Sie sind auf Seite 1von 134

Acoustic Emission for High-Energy Piping

A State-of-Knowledge Summary

2013 TECHNICAL REPORT


Acoustic Emission for
High-Energy Piping
A State-of-Knowledge Summary

EPRI Project Manager


K. Coleman

3420 Hillview Avenue


Palo Alto, CA 94304-1338
USA

PO Box 10412
Palo Alto, CA 94303-0813
USA

800.313.3774
650.855.2121

askepri@epri.com 3002002218
www.epri.com Final Report, November 2013
DISCLAIMER OF WARRANTIES AND LIMITATION OF LIABILITIES

THIS DOCUMENT WAS PREPARED BY THE ORGANIZATION(S) NAMED BELOW AS AN ACCOUNT OF


WORK SPONSORED OR COSPONSORED BY THE ELECTRIC POWER RESEARCH INSTITUTE, INC. (EPRI).
NEITHER EPRI, ANY MEMBER OF EPRI, ANY COSPONSOR, THE ORGANIZATION(S) BELOW, NOR ANY
PERSON ACTING ON BEHALF OF ANY OF THEM:

(A) MAKES ANY WARRANTY OR REPRESENTATION WHATSOEVER, EXPRESS OR IMPLIED, (I) WITH
RESPECT TO THE USE OF ANY INFORMATION, APPARATUS, METHOD, PROCESS, OR SIMILAR ITEM
DISCLOSED IN THIS DOCUMENT, INCLUDING MERCHANTABILITY AND FITNESS FOR A PARTICULAR
PURPOSE, OR (II) THAT SUCH USE DOES NOT INFRINGE ON OR INTERFERE WITH PRIVATELY OWNED
RIGHTS, INCLUDING ANY PARTY'S INTELLECTUAL PROPERTY, OR (III) THAT THIS DOCUMENT IS SUITABLE
TO ANY PARTICULAR USER'S CIRCUMSTANCE; OR

(B) ASSUMES RESPONSIBILITY FOR ANY DAMAGES OR OTHER LIABILITY WHATSOEVER (INCLUDING ANY
CONSEQUENTIAL DAMAGES, EVEN IF EPRI OR ANY EPRI REPRESENTATIVE HAS BEEN ADVISED OF THE
POSSIBILITY OF SUCH DAMAGES) RESULTING FROM YOUR SELECTION OR USE OF THIS DOCUMENT OR
ANY INFORMATION, APPARATUS, METHOD, PROCESS, OR SIMILAR ITEM DISCLOSED IN THIS
DOCUMENT.

REFERENCE HEREIN TO ANY SPECIFIC COMMERCIAL PRODUCT, PROCESS, OR SERVICE BY ITS TRADE
NAME, TRADEMARK, MANUFACTURER, OR OTHERWISE, DOES NOT NECESSARILY CONSTITUTE OR
IMPLY ITS ENDORSEMENT, RECOMMENDATION, OR FAVORING BY EPRI.

THE FOLLOWING ORGANIZATION, UNDER CONTRACT TO EPRI, PREPARED THIS REPORT:

Clarus Consulting, LLC

NOTE

For further information about EPRI, call the EPRI Customer Assistance Center at 800.313.3774 or
e-mail askepri@epri.com.

Electric Power Research Institute, EPRI, and TOGETHERSHAPING THE FUTURE OF ELECTRICITY are
registered service marks of the Electric Power Research Institute, Inc.

Copyright 2013 Electric Power Research Institute, Inc. All rights reserved.
Acknowledgments The following organization, under contract to the Electric Power
Research Institute (EPRI), prepared this report:

Clarus Consulting, LLC


P.O. Box 470318
Charlotte, NC 28247

Principal Investigator
J. Foulds

This report describes research sponsored by EPRI.

John Rodgers, Acoustic Emission Consulting, Inc., is thanked for


contributing the majority of data that went into this review.
Capturing the historic development of this AE monitoring
application might not have been possible without Mr. Rodgers
valuable input. Itzik Mizrahi, Margan Physical Diagnostics Ltd., is
thanked for contributing to the service experience data and providing
background on Margans procedures. Jonathan Parker, Kent
Coleman, and John Shingledecker, EPRI, helped obtain some of the
field experience information summarized in this report. The
following individuals are acknowledged for sharing their relevant
utility plant experience: Robert Griffin, NB Power Generation
Corp.; David Berger, PPL Generation; Michael Crichton, American
Electric Power; and Thomas Fox, NRG Energy. Finally, Acoustic
Emission Consulting, Inc. and Margan Physical Diagnostics Ltd. are
thanked for their review of the document.

This publication is a corporate


document that should be cited in the
literature in the following manner:

Acoustic Emission for High-Energy


Piping: A State-of-Knowledge
Summary.
EPRI, Palo Alto, CA: 2013.
3002002218.
iii
Product
Description This report pulls together information on the application of acoustic
emission (AE) testing for monitoring high-energy piping in fossil
power plants. The experience data reviewed include the Electric
Power Research Institute (EPRI) initiatives under research project
RP1893 (circa 19861995) that culminated in the 1995 monitoring
guidelines for hot reheat piping and available post-guidelines field
applications through to the present. EPRI has had more than two
decades of application experience with the use of AE in the field;
however, a clear understanding of the optimal method of testing, the
role AE can best play in the broader context of piping integrity
management, and what forms of damage may be detected in advance
of failure appear to remain a matter of subjective opinion. In an effort
to reduce the subjectivity associated with the application of AE to
high-energy piping, EPRI gathered and assimilated available AE
monitoring experience and provided its technical assessment of the
application.

Background
EPRI is currently engaged in evaluating the practical application of
AE testing as a nondestructive method to detect and assess creep
damage in operating high-energy piping in fossil power plants. As far
as application of AE to high-energy piping, the available information
appears to represent a range of application techniques and provides
varying indications of feasibility and performance, reflecting a need
for an independent review to suitably document the state of
knowledge.

Objectives
To summarize the state of knowledge regarding the application
of AE for detecting and assessing damage in fossil power plant
high-energy piping

Approach
The AE plant monitoring experience data obtained by EPRI from
commercial vendors and its utility members were reviewed. The data
included test methods that have been applied, interpretation of test
results, and correlations with other nondestructive and destructive
test findings. Particular attention was paid to the variability in test
methods and interpretation of test data; the practical aspects of
implementation such as manageability of indications, reduction of
background, and other irrelevant extraneous responses; and the
nature (form and extent) of damage detected.

v
Results
Nearly all of the data reviewed was provided by two service providers
with differing practices. Data from plant owner-operators were very
limited, so that with little significant plant-reported experience,
including nondestructive evaluations and damage and failure
experience, the emerging picture regarding the state of knowledge
may be unbalanced. In any case, the following observations have been
made:

There is substantial variability in the practice;


Aspects of the technology merit enhanced transparency.
The method as practiced appears to readily detect active thermal-
fatigue cracking under thermal stimulation.
A few cases have been reported where the method correlated
with creep damage, but they were insufficient for firm
conclusions on creep damage detectability.
The method provides potential value to managing
integrity/lifetime that may be realized with further streamlining
of the process, establishing a refined understanding of the
capability via controlled laboratory tests, expanding and
balancing the experience database, and updating the EPRI 1995
Guidelines.
In addition to providing a summary of the practiced methods,
including the 1995 Guidelines, and all of the reported experience
data, this document offers a synthesis of the experience that helps to
define user expectations.

Applications, Value, and Use


Given that there has not been a broad review of the AE method as
practiced for high-energy piping since development of the EPRI
Guidelines in 1995, and given that questions continue to arise
regarding what may and may not be expected with regard to damage
detection using the AE method, this state-of-knowledge summary
that pulls together the available post-1995 experience data gives the
plant engineer and maintenance planner an understanding of the
practical value of the method with its advantages and limitations.
This state-of-knowledge summary also provides the technical bases
for developing the steps to update the 1995 Guidelines and to
encourage consistency in field applications.

Keywords
Acoustic emission Creep
High-energy piping Nondestructive evaluation
Seam welds

vi
Abstract
In an effort to capture the state of knowledge regarding acoustic
emission (AE) monitoring of high-energy piping in fossil power
plants, both the available AE monitoring methods practiced and the
in-plant experience data have been reviewed. All of the information
collected by the Electric Power Research Institute (EPRI) was
reviewed. The reviewed data spanned the period from 1980 through
to the present, thus including some pre-Guidelines experience.
However, the main observations made for this summary are based
nearly exclusively on the post-1995 application of high-energy piping
AE monitoring.

The review focuses on two aspectsthe AE monitoring method as


practiced and the service or field experience. The data regarding the
method practiced primarily consists of AE monitoring conducted in
general conformance with the 1995 Guidelines with some
information obtained on a second method somewhat different from
the Guidelines practice. The known details of the practices have been
described, although many details of the applied data
analysis/evaluation procedures are unknown because they are
proprietary or have not been completely described in the information
made available to date. In addition, recommendations are made for
improved streamlining of the monitoring method.

Nearly all of the materials reviewed came from service providers and
may not adequately reflect the experience. Nevertheless, some
important observations have been made regarding the use of AE for
identifying non-seam-weld cracking, long seam-welded piping creep
damage, and its value with regard to helping prioritize nondestructive
surface and volumetric examinations. Recommendations are made to
address the obvious issues with the variability in application of the
AE method, including aspects of transparency, completeness of the
experience database, and clarity with regard to what can be expected
in any given AE monitoring program. Not surprisingly, the data
accumulated to date do not support the notion that the technology
by itself provides a foolproof method of preventing failures. There
are clearly many variables to implementation and interpretation;
however, the data also show that AE can, in addition to helping
prevent failures from at least some forms of cracking, help prioritize
nondestructive volumetric and surface examinations.
Recommendations are also made to expand the experience database
and update the 1995 Guidelines.

vii
Conversion
Factors for Units
Used in This Parameter English to Standard International
(SI) Units
Report Length 1 in. = 25.4 mm or 2.54 cm
1 ft = 0.305 m
Pressure 1 psig = 6.895 kPa
Temperature C = (F -32) x 5/9
Temperature Interval 1F = 0.5556C

ix
Table of Contents

Section 1: The AE Monitoring Method and Its


Application History ..................................1-1
AE Fundamentals ............................................................. 1-2
Terminology ............................................................... 1-2
The Basic AE Monitoring Process .................................. 1-9
Field Application History ................................................. 1-11
Pre-Mohave Experience (19751985) ........................ 1-11
Post-Mohave, 19851990 ......................................... 1-17
The 1995 EPRI Guidelines ......................................... 1-20
Current Practice ........................................................ 1-46
Standardization ............................................................. 1-54

Section 2: Experience Summary...............................2-1


Data Sources ................................................................... 2-1
Field Application Experience ............................................. 2-1
19851990............................................................... 2-1
19901995............................................................... 2-4
19961999............................................................. 2-11
2000Current .......................................................... 2-24
Key Observations ........................................................... 2-33

Section 3: Summary ................................................3-1


The AE Method as Practiced .............................................. 3-1
AE Performance and Capabilities ...................................... 3-2
Recommendations ............................................................ 3-4

Section 4: References ..............................................4-1

Appendix A: Field Experience Data....................... A-1


19851990 ................................................................... A-1
19901995 ................................................................... A-2
19961999 ................................................................... A-6
2000Current................................................................. A-9

xi
List of Figures

Figure 1-1 Example of the AE output from a burst-type event,


reproduced from EPRI report TR-111558, illustrating the
initial rapid rise to a peak followed by attenuation of the
energy wave. Identified features are defined in the text. ....... 1-3
Figure 1-2 Example of AE output with the application of a
floating threshold to eliminate noise. ............................... 1-4
Figure 1-3 Examples of waveguide configurations for high-
energy piping AE monitoring as presented in the 1995
EPRI Guidelines................................................................ 1-7
Figure 1-4 Schematic of a waveguide array and the overall
setup to detect and locate AE emission sources.................... 1-8
Figure 1-5 Definition of normalized event density ranking
(NEDR) of a location cluster, a source ranking method
developed in the post-1995 guidelines EPRI research. .......... 1-9
Figure 1-6 PG&Es Potrero #3 waveguide and
instrumentation setup. ..................................................... 1-23
Figure 1-7 Schematic illustrating event location accuracy. ........ 1-27
Figure 1-8 Schematic illustrating the laboratory test setup; LF
and SF represent known indication locations
corresponding to locations at 31 and 89.5 in. in
Table 1-1. ..................................................................... 1-29
Figure 1-9 Macroetched sections of the laboratory-tested
pipe showing (top) the slag inclusion corresponding to
the 89.5- in. location in Table 1-1 and the SF location
at 73 in. in Figure 1-8, and (bottom) the inactive crack
corresponding to the 31 in. location in Table 1-1 and
the LF location at 18.5 in. in Figure 1-8. ........................ 1-31
Figure 1-10 Attenuation measured for a lead break source
with a 300-kHz resonant sensor, 40-dB integral pre-
amp, and 80-dB gain. .................................................... 1-33
Figure 1-11 Key elements of the 1995 EPRI high-energy
piping AE monitoring guidelines. ..................................... 1-38

xiii
Figure 1-12 Key elements of the data analysis/evaluation. ....... 1-42
Figure 1-13 Definition of normalized event density ranking
(NEDR) of a location cluster, a source ranking method
developed in the post-1995 Guidelines EPRI research. ....... 1-49
Figure 2-1 Plug sample evidence from Valmy #2 of non-
creep thermal fatigue-appearing crack from a seam
weld lamination that was AE-located in cooldown, but
not in online monitoring. ................................................. 2-18
Figure 2-2 Graphics reproduced from EPRI report TR-
111558 in the case of CPL, ES Joslin HRH AE
monitoring AE response with the peak at a hanger
support (upper) and results of cryo-cracking reportedly
representing the early stage of creep cavitation (lower). ..... 2-20
Figure 2-3 AE activity (NEDR) as a function of pressure in
online monitoring, showing sensitivity to operating
pressure. ....................................................................... 2-22
Figure 2-4 Pie charts showing the breakdown in the
reviewed database of experience by components
(cases), seam-welded (S-W) piping versus non-seam-
welded piping, for (a) all data and (b) the post-1995
period. ......................................................................... 2-34
Figure 2-5 Breakdown of materials for the 46 post-1995 AE
monitoring cases reviewed. ............................................. 2-34
Figure 2-6 Breakdown of the stress method used in the 46
post-1995 documented AE monitoring cases. .................... 2-35
Figure 2-7 Breakdown of the AE monitoring cases with
specific located AE sources (AE Call Cases) and ones
with no identified AE sources (AE No Call Cases). ............. 2-36
Figure 2-8 Breakdown of the post-1995 AE monitoring
cases as correlated with nondestructive examination
results. The 15 No NDE Cases are ones where there
were no nondestructive examinations conducted................ 2-37
Figure 2-9 Breakdown of cases where damage or cracking
was identified. The AE-Cracking cases represent ones
where damage was not specifically identified as creep. ..... 2-37

xiv
List of Tables

Table 1-1 Identified AE location clusters compared with


results from other examinations and tests on Potrero #3. ..... 1-24
Table 1-2 Active AE clusters compared with NDE indications
on Potrero #3 East hot reheat lead. .................................. 1-36
Table 1-3 Active AE clusters compared with NDE indications
on Potrero #3 West hot reheat lead. ................................ 1-37
Table 1-4 Linear location accuracy for straight pipe. ............... 1-45
Table 1-5 Linear location accuracy for bend/elbow. ............... 1-45
Table 1-6 Current ASTM standards under the jurisdiction of
ASTM Subcommittee 07.04 on Acoustic Emission
Method. ........................................................................ 1-55
Table 2-1 In-plant AE monitoring cases in the period 1985
1990. Italicized text has been added to the original
table based on a review of the text from that reference......... 2-2
Table 2-2 Active AE clusters compared with NDE indications
on Potrero #3 East hot reheat lead. .................................... 2-7
Table 2-3 Active AE clusters compared with NDE indications
on Potrero #3 West hot reheat lead. .................................. 2-8
Table 2-4 Correlation between AE activity and results of
other examinations, as assembled from the EPRI report
TR-111558 table with the NEDR data added. ................... 2-14
Table 2-5 Tabulated listing of in-plant AE monitoring cases
prepared from a table provided by Rodgers...................... 2-26
Table 2-6 Listing of experience as tabulated from data
provided by Mizrahi. ..................................................... 2-30
Table A-1 In-plant AE monitoring cases in the period 1985
1990 from EPRI report RP 1893-4. Italicized text has
been added to the original table based on a review of
the text from that reference. .............................................. A-1

xv
Table A-2 Listing of In-plant AE monitoring cases. The listing
includes the 12 cases of 19851990 (underlined). ............. A-2
Table A-3 Listing of in-plant AE monitoring cases as reported
in EPRI report TR-111558. ................................................ A-7
Table A-4 Tabulated listing of in-plant AE monitoring cases
prepared from a table provided by Rodgers..................... A-10
Table A-5 Listing of experience as tabulated from data
provided by Mizrahi. .................................................... A-11

xvi
Section 1: The AE Monitoring Method and
Its Application History
Managing and assuring the integrity of high-energy piping in fossil-fueled power
plants continues to be a significant operations and maintenance related activity,
given that failures in such components can be catastrophic, as has been the case
with low-alloy steel long seam-welded hot reheat piping (for example, Mohave in
1985 [1] and Monroe in 1986 [2]). Key to piping integrity assurance is the
identification of damage sufficiently in advance of failure. Such damage
identification has been and continues to be achieved by nondestructive evaluation
(NDE) methods, primarily ultrasonic test (UT) inspections that are volumetric;
that is, they examine the through-thickness condition of the piping.

Following the heightened concern for the integrity of low-alloy steel long seam-
welded high-energy piping in the mid-1980s, EPRI began developing guidelines
for evaluating long seam-welded high-energy piping that have evolved over the
past 25 years (1987 [3], 1996 [4], 2001 [5], 2003 [6], 2012 [7]). The most recent
edition of EPRIs Guidelines for the Evaluation of Seam-Welded High-Energy
Piping [7] describes the current capability of EPRI-recommended UT methods
to detect cracks and creep microcracked zones of as little as 0.03 inches
(0.76 mm) in depth (through-wall dimension). Crack growth predictions
suggest that detectable cracking using proper UT procedures may be made three-
to-five years in advance of failure. The UT inspection approach offers a reliable
means of helping manage piping integrity. However, as the sole NDE tool for
the purpose, UT carries with it several practical disadvantages:
The conservative prediction of damage progression rates implies that piping
weldments merit inspection every three-to-five years, a frequency that can be
economically burdensome.
While locations and weldments may be prioritized for inspections on the
basis of design, operation, and maintenance experience, the default preferred
implementation calls for 100% inspection of the systems long seam
weldments, at least in cases where prior inspections have not been frequent.
Execution of the inspection requires significant time and effort relating to
insulation removal and scan time, the latter increasing with increased
refinement of the UT method.

1-1
A major benefit to the efficient implementation of UT inspections may be
realized if the required inspection volume can be significantly reduced. The
potential exists for acoustic emission (AE) monitoring of the high-energy piping
system to identify specific areas of the system that merit further evaluation by
UT. The ability to characterize the emission of transient elastic waves generated
from the release of energy (AE) associated with the progression of damage (for
example, creep cracking) is the basic premise for the application of AE to high-
energy piping. The advantage of an AE monitoring system is that it may be
installed and implemented relatively economically, in addition to its potential to
detect high rates of in-operation damage progression that reflects imminent
failure, a useful advance warning indicator.

EPRI is currently engaged in evaluating the practical application of AE as a


nondestructive method to detect and assess creep damage in operating high-
energy piping in fossil power plants. The potential of the method to provide
advance warning of failureand even an indication of remaining lifetime in these
componentshas been explored for roughly two decades, including by EPRI in
the form of laboratory testing and field monitoring. Specifically to address hot
reheat piping, guidelines for AE monitoring were published in 1995 [8],
following the research conducted in the immediate post-Mohave period [9, 10].

Since publication of the EPRI guidelines in 1995, the practice of AE monitoring


of high-energy piping has expanded (for example, [11, 12]), with monitoring
practices having changed from the 1995 EPRI Guidelines and also varying with
service provider and application conditions. Indeed, the information and findings
generated to date suggest a range of AE monitoring application techniques and
performance, reflecting the need for an independent review to suitably document
the state of knowledge with regard to AE application. This review is an effort to
extract and summarize the relevant factual details of the application experience.
As described here, the reviewed data include EPRI-obtained fossil plant high-
energy piping experience as offered by the AE monitoring service providers and
some fossil plant owner-operators.

AE Fundamentals

Terminology

In describing AE monitoring practice, including its many aspects related to


hardware, data, and interpretation of data, a number of technical terms are used.
To facilitate understanding of this review, fundamentals of the AE monitoring
process and definitions of commonly used terms are briefly covered. The
definitions, presented below in an order best describing the AE monitoring
method, are primarily from the ASTM standard E 1316: Standard Terminology
for Nondestructive Examinations [13].

acoustic emission (AE) [13]: The class of phenomena whereby transient elastic
waves are generated by the rapid release of energy from localized sources within a
material, or the transient waves so generated. In the case of AE for high-energy
piping monitoring, the fundamental premise is that the progression of relevant

1-2
forms of damage, for example creep, results in this type of emission. The
emission can be such that the emitted wave features allow it to be easily identified
as emanating from a source as a single burst, attenuating with time. This
phenomenon, characterized as a burst event or more generally, an event, is
illustrated in Figure 1-1 (reproduced from unpublished EPRI report TR-111558
[11] as an AE signal [electrical voltage output generated from the emission]).
Such relatively easily identified events are distinct from the multitude of lower
amplitude events that essentially blend together as what is referred to as
continuous emission, typical of the noise associated with flow in high-energy
piping. While there is no standard definition for what constitutes noise (ASTM
is working on a definition), the overall signal features of such continuous
emission make it generally unremarkable, of limited value, and confounding to
the discernment and analysis of data from more obvious, discrete events. Spectral
frequency of flow-related noise relevant to high-energy piping is generally lower
than 300 kHz, but may be higher near turbine leads and boiler outlets. A
common approach to eliminating such continuous emission from consideration
in analyzing AE data is to use an evaluation threshold (in data processing) or
system threshold (by hardware filtering) to effectively screen such data. Figure
1-2 is another graphic reproduced from EPRI report TR-111558 [11] that shows
the noise and burst event emissions and the (evaluation) threshold, floating in
this case, designed to eliminate from the analysis noise that varies in amplitude
over time. The deadband represents the signal data below the threshold.

Figure 1-1
Example of the AE output from a burst-type event, reproduced from EPRI report
TR-111558 [11], illustrating the initial rapid rise to a peak followed by attenuation
of the energy wave. Identified features are defined in the text.

1-3
Figure 1-2
Example of AE output [11] with the application of a floating threshold to eliminate
noise.

threshold floating vs. fixed: A floating threshold is one where the amplitude
below which data are ignored is set on the basis of a time-averaged measure of
the signal voltage; that is, the threshold can vary with the amplitude of the noise.
A fixed threshold, on the other hand, is preset and does not vary. See Figure 1-2
schematically illustrating the threshold.

AE signal [13]: An electrical signal obtained by detection of one or more acoustic


emission events. A detection device (sensor), typically piezoelectric, transforms
the particle motion of the transient elastic wave into an electrical signal. Such a
signal is often amplified to deliver the voltage-time output exemplified by Figure
1-1. An AE signal is typically associated with an AE event, so the terms signal
and event are often used interchangeably.

dB: The dB (origin from decibel as conventionally defined in the field of


acoustics) is the output voltage as logarithmically referenced to, and calculated
from, the sensor voltage before amplification. The final output dB voltage is
typically amplified from the AE dB. A specific level of dB, defined as the AE dB,
dBAE, to represent the peak amplitude, is defined as [13] dBAE=20[log10(A1/A0)]
where A0=1 V at the sensor before amplification and A1 is the sensor-measured
voltage before amplification. This standardized output representation is key to
permitting comparison of data from different sensors.

AE count [13]: The number of times the acoustic emission signal exceeds a preset
threshold during any selected portion of a test. Figure 1-1 identifies some of the
output counts making up the AE count, also sometimes referred to as the ring-
down count (RDC). The multiple peaks represent the number of cycles or AE
count. This event-specific count is distinguished from the count of events
occurring over a period of time.

1-4
event count [13]: The number obtained by counting each discerned acoustic
event once. This event count is a parameter used in assessing emission from a
given source/location over a period of time and is distinguished from the event-
specific count or AE count defined above.

event count rate [13]: The time rate of the event count. There is no standard
definition for the term with regard to the time period and source/location volume
for which the rate is computed. An example of units for event count rate is
# events/hour.

AE activity: This term is used very frequently and is qualitatively defined only as
[13] the presence of acoustic emission during a test. Quantitative measures of
AE activity include event rate, emission energy, energy rate, etc.

signal (event) energy [13]: The energy contained in an acoustic emission signal,
which is evaluated as the integral of the volt-squared function over time. As a
measure of AE activity, an aggregate energy may be so computed for one or more
events (one or more signals) over a selected time period.

signal (event) strength [13Error! Bookmark not defined.]: The measured area of the
rectified AE signal with units proportional to volt-sec.

AE RMS (signal RMS) [13]: The rectified time averaged AE signal, measured
on a linear scale and reported in volts.

signal amplitude [13] or peak amplitude: The peak voltage of the largest excursion
attained by the signal waveform from the emission event. The signal amplitude
is also commonly referred to as the peak amplitude. Figure 1-1 identifies the peak
amplitude or dB of the signal from the specific emission event.

amplitude distribution: This is a probability density or cumulative probability


function used to describe one aspect of the signature of one given event or to
describe the behavior of several events associated with a given emitting location.

In the case of a single event, the density and cumulative distribution functions,
ft(V) and Ft(V), respectively, are: (a) ft(V) is the fraction of the number of times
the given signal waveform has a peak (see AE count) between the voltage, V and
V+dV, also referred to as the differential threshold crossing distribution [13]; (b)
Ft(V) is the fraction of times the given signal waveform has a peak exceeding the
voltage, V, also referred to as the cumulative threshold crossing distribution [13].

In the case of multiple events, the density and cumulative distribution functions,
f(V) and F(V), respectively, are: (a) f(V) is the fraction of events with a peak
amplitude between the voltage, V and V+dV, also referred to as the differential
amplitude distribution [13]; (b) F(V) is the fraction of events with a peak
amplitude exceeding the voltage, V, also referred to as the cumulative threshold
crossing distribution [13].

1-5
signal (event) rise time [13]: The time between AE signal start and the peak
amplitude of that AE signal, typically measured in microseconds. The signal
start is usually defined as the first crossing of the threshold (see Figure 1-1).

Signal (Event) Duration [13]: The time between AE signal start and AE signal
end. This is also referred to as event duration (see Figure 1-1), typically in
microseconds or milliseconds. The signal end is usually defined by the last
crossing of the threshold.

pseudo frequency (average frequency): This term has been defined in the 1995
EPRI guidelines [8] as a signal feature of an AE event measured as the ratio of
the AE count of that event (also ring-down count) to the event or signal duration,
reported in kHz. Since this is unrelated to the sensor resonant frequency, the
term pseudo is used.

waveguide [13]: A device that couples elastic energy from a structure or other
test object to a remotely mounted sensor during AE monitoring. In the case of
high-energy piping monitoring, the waveguide is a standoff intended to provide
sufficient thermal gradient from the hot pipe surface to the sensor at the other
end of the waveguide to keep the sensor relatively cool. Figure 1-3 shows
examples of waveguide configurations presented in the 1995 EPRI high-energy
piping AE monitoring guidelines [8]. These configurations, minor variations in
waveguide rod dimensions notwithstanding, have remained essentially
unchanged. For the high-energy piping application, the waveguide is generally
permanently weld-attached to the pipe; the details of the attachment are
described later. Mechanically attached waveguides with a suitable high-
temperature couplant are not permanent, not reliable for the application, and not
generally used. The illustrated configurations indicate easily swappable sensors
coupled to the waveguide via a suitable couplant, although configurations with
permanently installed (bonded) sensors are also used.

array [13]: A group of two or more AE sensors positioned on a structure for the
purposes of detecting and locating sources. The sources would normally be
within the array. An array also refers to the array of waveguides. Figure 1-4
reproduced from the EPRI guidelines [8] is a schematic showing how the
waveguides are arrayed on a given pipe in order to locate an AE emission source.
The waveguides are almost always aligned (located at the same circumferential
pipe position) and spaced 1020 feet apart (span length in Figure 1-4),
depending on the noise level, type of sensor used and the attenuation for a
particular application.

1-6
Figure 1-3
Examples of waveguide configurations for high-energy piping AE monitoring as
presented in the 1995 EPRI Guidelines [8].

computed (source) location [13]: A source location method based on algorithmic


analysis of the difference in arrival times among sensors. For the high-energy
piping application, a simple linear algorithm is used, hence the preferred
waveguide alignment. For example, referring to Figure 1-4, the identified source
is located based on the difference in the arrival time between the first hit at #2
sensor and that at #3 sensor and with known wave velocity. Note that a source
outside the array (right of #3 sensor in the figure) may appear to be from within
the array or may be located at the array end, both having potential to be in error.
This is referred to as the end-of-array effect [8]. Also, note that clearly the linear
locating method provides no information on the circumferential position of the
source. The source location linear accuracy naturally varies, depending on the
waveguide spacing-to-pipe circumference ratio, assumption of the wave velocity,
and source location (higher error when source is close to sensor). Location error
can vary from as low as 5% to as much as nearly 20% of the waveguide
spacing [8].

location cluster: A group of AE events whose event locations are within a small
localized area and ideally with similar signal features. In high-energy piping
monitoring practice, evaluating the activity from clusters is key to application of
the AE technology.

source rank: A ranking of the location clusters based on their AE activity. There
are no standards for how a source rank should be determined. In fact, AE activity
can vary with the stimulation or loading stimulus (mode or cycle), sensitivity of
the monitoring system, parameters used to quantify activity, variation from one
situation to another, etc. Establishing a consistent source ranking method and
then correlating the source rank with the type of AE source is key to future
robustness of the high-energy piping monitoring application.

1-7
Figure 1-4
Schematic of a waveguide array and the overall setup to detect and locate AE
emission sources.

normalized event density ranking (NEDR) [11]: A ranking of the location clusters
based on their AE activity, specifically the number of events from a cluster per
unit time and normalized for the length of the cluster in terms of the standard
deviation length of the cluster. The method developed by Rodgers for EPRI as
applied to high-energy piping is summarized in Figure 1-5 [11]. The NEDR
value for an AE cluster consists of several basic analysis steps: (a) Defining the
location limits of the cluster; (b) Computing the number of located events in the
cluster over the time of analysis; (c) Computing the standard deviation of the
cluster in inches; (d) Determining the peak load time in the analysis period. The
NEDR value is then determined by the equation: Cluster Events/Cluster Std.
Dev/Peak Load Time (Units: Events/Inch-Hr).

attenuation [13]: The decrease in AE amplitude per unit distance, normally


expressed in dB per unit length. Attenuation effectively limits sensor/waveguide
spacing and also influences waveguide rod dimensions.

sensor narrowband vs. broadband: A sensor is a piezoelectric transducer that


converts the mechanical acoustic vibration energy into an electrical voltage
output. Sensors can be narrowband, resonant about a central frequency (for
example, 200 or 300 kHz) or broadband, having no central resonant
characteristic. In the case, of high-energy piping monitoring, the EPRI
guidelines recommended sensors [8] are high sensitivity, narrowband, with
frequency depending on the noise (lower noise lower resonant frequency).
Additionally, commercially available sensor systems include integrated
preamplification and a high-pass filtering capability to reduce noise output
(frequency-based filtering with lower noise lower high-pass filtering frequency
such as 200-kHz filtering with a 300-kHz narrowband sensor).

1-8
stimulation [13] (stress method): The application of a stimulus such as force,
pressure, heat, and so forth, to a test article to cause activation of acoustic
emission sources. In the case of high-energy piping monitoring, stimulation is
achieved in operation, and AE monitoring may be conducted under various unit
loading cycles such as startup, cooldown, full load, and load swings.

Kaiser effect [13]: The absence of detectable acoustic emission at a fixed


sensitivity level, until previous applied stress levels are exceeded.

AE system verification [13]: The process of testing an AE system to assure


conformance to a specified level of performance or measurement accuracy. (This
is usually carried out prior to, during and/or after an AE examination with the
AE system connected to the examination object, using a simulated or artificial
acoustic emission source). In high-energy piping monitoring practice, such
verification, also referred to as calibration, is typically conducted before
monitoring, as part of the installation of the system.

Figure 1-5
Definition of normalized event density ranking (NEDR) of a location cluster, a
source ranking method developed in the post-1995 guidelines EPRI research [11].

The Basic AE Monitoring Process

The primary assumption with the use of AE monitoring for high-energy piping
is that the in-operation damage processes of interest due to high-temperature
creep produce the acoustic elastic energy waves that can be detected. Further, the
monitoring process is implemented in a manner such that the area of piping
responsible for the detected energy emission can be identified. In field
application of the AE technology, however, there are a multitude of factors that
can confound the detection and identification process, particularly emissions due
to phenomena of little or no relevance to the main damage detection objective.

1-9
These include flow, surface oxide cracking, and system settling from thermal
effects including rubbing and frictional effects at contacts such as hangers. The
challenges associated with identifying relevant damage processes and their
location are discussed later. This subsection simply focuses on the basic process
and identifies the main issues with monitoring high-energy piping. Most of this
basic information has been reproduced from the EPRI AE monitoring guidelines
document [8].

The sound waves emanating from a source in the piping propagate radially
throughout the structure, attenuating with distance from the source. The wave
direction changes as it is reflected and refracted at the boundaries of the
structure. This radial propagation is one reason for the inaccuracy in locating
emission sources, increasing with pipe circumference to waveguide spacing ratio.
The waves are detected by a sensor that transforms their mechanical
displacement into an electrical signal. The electrical signal can be stored either
directly as a digital signal or indirectly as signal features. Figure 1-4 shows the
major components of an AE system for measuring and locating AE signals that
are typical of discrete events.

In order to distinguish between discrete events and the background noise, a


voltage threshold level is set above which most signal features are measured.
Obviously, the AE count (ring-down count), rise time, event duration, pseudo
frequency, and signal energy will all vary by raising or lowering the threshold.
One capability of AE technology is that of source location. The sensor closest to
an AE event detects the AE wave arrival and starts a clock. The arrival of the AE
wave at the second sensor and subsequent sensors stops the associated clock(s). A
computer algorithm using the distance between sensors, the velocity of sound in
the material, and the event arrival times calculates the source location. A pipe
lends itself to a linear location method of source location. The source is located in
a pipe an axial distance from a reference point. The circumferential distance
around the pipe is not used.

In order for an energy release to generate AE, stress needs to be induced into the
structure (stimulus). A number of methods exist for inducing this stress (or
straining the structure). These are typically employed during operation (for
example, load swings, cooldown, startup). The acoustic emission released from
the stressed structure is statistical in nature, and the interpretation of the signal(s)
represents the main challenge.

The AE monitoring of high-energy piping carries with it the following main


problems:
The operating pipe metal temperature is high. The waveguide standoffs help
overcome this problem.
Steam flowing through the piping causes high amplitude and fluctuating
background noise. The use of a threshold and high-pass filtering with
narrowband sensors help alleviate this problem.
Control over how piping is stressed is limited. Generally, plant operation
dictates how the pipe is stressed, along with its stress level.
1-10
The pipe is not always accessible for monitoring. If normal plant operation
stresses the pipe (that is, plant startup, on line, or plant shutdown) the
monitoring technique will need to be tailored to match that situation. Test
preparations such as waveguide installation and sensor installation may
require scaffolding or the use of ropes to access the pipe.
AE signals need to be quantified and differentiated from a multitude of noise
sources. For each test, a large amount of data is generated from these
combined AE background noises and AE sources of interest. Thus, there is a
compelling need to separate AE sources of interest from other sources.
The AE data may need to be analyzed quickly. Confirming conventional
inspections (for size, depth, and orientation), and defect repairs await the AE
results. All of these activities are performed during short outages. AE
monitoring during plant cooldown, the AE data processing, interpretation,
and reporting must be done quickly (generally within a few days).

Field Application History

The use of AE for monitoring fossil power plant equipment has existed for over
35 years with applications ranging from the earliest boiler tube leak detection and
turbine bearing monitoring (for unbalance, blade deterioration, and bearing
degradation) systems to the post-Mohave (1985) focus on high-energy piping for
damage detection. This section briefly describes the historic evolution of the use
of AE for fossil plant field application, the 1995 EPRI guidelines for high-energy
piping monitoring, and the current technology as practiced for high-energy
piping.

Pre-Mohave Experience (19751985)

Well before the 1985 Mohave hot reheat piping incident, the potential
application of AE for monitoring fossil plant equipment was being investigated,
as evidenced by (a) the documented proceedings of the 1982 EPRI Workshop:
Incipient Failure Detection for Fossil Power Plant Components [14] where more
than 25% of the workshop papers were related to the use of AE; and (b) the 1981
EPRI report On-Line Acoustic Emission Monitoring of Fossil Power Plants: A
Critical Assessment [15]. The primary AE monitoring focus was then, and
continues to be, the use of AE in detecting damage in advance of failure,
although its application to the early detection of boiler tube leak failures was also
being researched at that time. Interestingly, in this pre-Mohave era, high-energy
piping was not the focus, rather, the potential for monitoring steam turbine
rotating components (rotors and disks, following the 1974 Gallatin rotor failure)
and detecting tube leaks (boiler tube failures affecting outage costs were in the
forefront) were of greatest interest. However, that early experience with the use
of AE significantly influenced the high-energy piping application, and this
summary of the early work includes relevant lessons learned with regard to
hardware, data interpretation, and limitations of the AE monitoring application.

1-11
The EPRI 1982 workshop proceedings [14] included experience on incipient
failure, leak detection, or crack growth monitoring applications for various fossil
plant applications as follows:

Steam Turbines
Field and laboratory experiments on EPRI projects RP 1266 and RP 734
(Graham et al. on Monitoring of TVA Turbines by Acoustic Emission;
Harris et al. on Fracture Mechanics-Acoustic Emission Correlations for
Turbine Rotor Cracking):
Monitoring for rotor bore crack propagation using bearing mounted sensors
was inconclusive mainly because there no physical cracking occurred during
the monitoring process. Laboratory correlations developed between AE and
crack advance under fatigue conditions for a 1-in. thick standard compact
tension CrMoV steel specimen indicated that the energy release rate with
fatigue crack advance correlated with the rate of emission in terms of the
event rate. Interestingly, the laboratory fatigue crack growth related that AE
event rates (with a broadband sensor with a 100-kHz high-pass noise filter
and a threshold of about 40 dB) were very low, estimated at about four events
per 0.010 in. of fatigue crack extension. It was thus inferred that for the
turbine rotor bore cracking application, the event rate would be also very low
(perhaps just one AE event for one or two major load cycles), requiring
careful monitoring and sophisticated instrumentation. Separately, it was
found that AE via direct-mounted sensors on bearing caps would be useful
for detecting unbalance, blade rubbing, and bearing deterioration.
Philadelphia Electric field experience (McElroy on Application of Acoustic
Emission On-Line Monitoring of Fossil-Fueled Power Plants):
One of the earliest (1979) field applications of AE to monitoring cracking
was on the Eddystone #2 HP-turbine outer casing. Three sensors, each
located surrounding two through-wall cracks (sensors were within 6 in. of
surface crack tips) that had been metallock sutured, were used to detect AE
under various startup conditions. Based on the AE event rate at the transient
load (MW) condition when the unit load transferred to the bus, a safe
transient was established for the load level at which the load transfer would
be made for continued operation. The linear-algorithm-located source
(crack-tip) emission dropped from over 800 events per hour to 80 events per
hour when reducing the transfer load to an acceptable level. No correlations
were made regarding AE as a function of crack growth, but in this case, AE
was used to establish an acceptable load cycle in the face of known crack
damage.
Brayton Point field experience (Thiele on Diagnostic Monitoring System
Development at Brayton Point):
This was another early study by Rockwell International (19771980)
supported by EPRI and New England Power Services Company, where a
total of 77 AE channels were employed to monitor a range of components on
four units, including rotating equipment (turbines, pumps, fans for bearing
and rotor problems), tubes (boiler, condenser, feedwater heater for leaks),

1-12
and high-energy piping (for cracking and exfoliation). All the AE
monitoring was conducted with 1/8-in. diameter metal rod waveguides and
sensors with an integrated pre-amplifier that was designed to significantly
reduce the effects of electromagnetic noise disturbance. Sensors were
changed to a 500-kHz type following the preliminary tests showing that
broadband or low-frequency transducers would get saturated with 50-kHz
noise. In case of the turbines, it was found that the bearing-located sensors
received highly attenuated signals such that the desired sensitivity for crack
monitoring in the rotor could not be achieved.

Boiler Tubes

The focus of all boiler tube monitoring has been the early detection of tube leaks
in water wall, superheater, and reheater tubes.
Central Electricity Generating Board (CEGB) (Perratt)
The Central Electricity Generating Board (CEGB), located in the UK,
developed and installed acoustic-monitoring boiler tube leak detection
systems in most, if not all, of the generating units in the UK Midlands
region, beginning in the early 1970s. The system consisted of a tube probe,
about 3/5 to 1 in. diameter. Attenuation was reportedly acceptable for probe
length to about 12.5 ft. The system used a high-frequency band pass filter
with a logarithmic amplifier. The technology for leak detection contrasts
with that used for crack monitoring in that acoustic emission from leaks have
a broadband spectral characteristic in a relatively low-frequency range (an
example of a 2-kHz optimum resonant sensor has been cited for one case), so
that the background noise is at a higher frequency than that of interest. In
the case of high-energy piping, the background noise frequency is at the low
end or lower than the AE frequency of interest. Additionally, since
attenuation decreases with decreasing frequency, the acoustic leak detection
system is easier to configure than for detecting high-energy piping damage.
Brayton Point field experience (Thiele on Diagnostic Monitoring System
Development at Brayton Point):
As noted above for Steam Turbines, this was an early study by Rockwell
International (19771980) supported by EPRI and New England Power
Services Company where a total of 77 AE channels were employed to
monitor a range of components on four units, including tubes (boiler,
condenser, feedwater heater for leaks). All the AE monitoring was conducted
with 1/8-in. diameter metal rod waveguides and sensors with an integrated
pre-amplifier that was designed to significantly reduce the effects of
electromagnetic noise disturbance. Sensors were changed to a 500-kHz type
following the preliminary tests showing that broadband or low-frequency
transducers would get saturated with 50-kHz noise. For the boiler water wall,
waveguides attached to the wall were used. Details of the setup are reported
in the Rockwell study to be available in EPRI report CS-1938 (July 1981).

1-13
ENEL-CISE:
A leak detection system developed by CISE, was adopted by ENEL (Italy)
for feedwater heater and boiler tubes in 32 units in the period 19791981. In
the case of the high-pressure feedwater heater, the AE from the flow into the
water volume was found to be more easily detected (the water behaves as a
waveguide), and a single sensor (with a 5-kHz high-pass filter) mounted on a
short duct connected to the feedwater inlet or outlet was adequate for the
simple detection. Leak location was not a priority in adopting this method
for the feedwater heaters. The AE activity metric uses was the signal root
mean square (RMS) in the frequency range 515 kHz. In the case of boiler
tubes, the signal RMS was also used as the emission metric, but in a slightly
broader frequency band of 1.510 kHz. The sensors consisted of an array of
transducers mounted on boiler observation windows, with a total of 10
sensors being adequate for each boiler. While the monitoring was concluded
to be effective, an interesting conclusion made was that the leak rates could
not be ascertained, even in a very approximate manner. This was reportedly
due to the many variables affecting the emission signal output, including the
geometry of the tube rupture and the varying attenuation.

Headers
Philadelphia Electric field experience (McElroy on Application of Acoustic
Emission On-Line Monitoring of Fossil-Fueled Power Plants):
Another early (19801981) field application of AE to monitor cracking was
on the Southwark economizer inlet header. This case was an interesting
learning experience. Direct-mounted sensors rated to 850F were used
(header temperatures were about 800F in startup transients); however, the
preamplifier had to be mounted about 12 ft distant, outside the economizer,
resulting in signal loss. Sensors were spaced 1 ft apart, although the event
rates measured were considered too low to warrant further monitoring. A key
finding of this experiment was that high-temperature sensors were not
reliable and that waveguides are preferred.
Brayton Point field experience (Pekrul et al. on Vibration Signature Analysis
and Acoustic Emission Monitoring at Brayton Point, EPRI CS-1938 [July
1981], excerpted from the final (unpublished) report on EPRI project RP
1893-4) [9]:
As part of the two-year, EPRI-sponsored program, a boiler penthouse
header was monitored. The 3-in. diameter steel rods that hang the steam
header from the I-beam superstructure and that extend out through the roof
were used as waveguides. The sensors were placed at the ends of the rods,
located outside the penthouse. The monitoring did not provide useful
information regarding header damage. However, a key finding was that the
rods appeared to deliver poor acoustic transmission (due to excessive
attenuation) and that the 3-in. diameter was an undesirable size for a
waveguide.

1-14
Babcock & Wilcox laboratory ambient temperature hydrostatic tests of a 3-ft
section of an ex-service, cracked secondary superheater outlet header (Coulter et
al. on EPRI RP 734-6, Acoustic Emission Monitoring of Cracks in Fossil
Boilers, EPRI CS-5264 [July 1987], summarized in EPRI report RP 1893-4)
[9]:

Crack growth was detectable and locatable, with the AE events of short
duration and rise time, low ring-down count, and medium peak
amplitude. The Kaiser effect was observed on repeated pressurization.
Long waveguides (25 ft) of 1/8-in. diameter were able to detect and
locate sources, and in one pressurization case where a leak occurred, the
leak was readily detected.

Babcock & Wilcox-Wisconsin Public Service secondary superheater outlet


header at Pulliam #7, July 1985 (Coulter et al. on EPRI RP 734-6, Acoustic
Emission Monitoring of Cracks in Fossil Boilers, EPRI CS-5264 [July 1987],
summarized in EPRI report RP 1893-4) [9]:

Waveguides (long-length waveguides as used for the laboratory header


tests described earlier) were found to be useful and permitted
continuous monitoring of high-temperature components for extended
periods. The application of a floating threshold allowed for detection
even in the presence of considerable background noise. An active AE
source was located at a tube hole and was later inferred to be from a
crack.

Piping
Philadelphia Electric field experience (McElroy on Application of Acoustic
Emission On-Line Monitoring of Fossil-Fueled Power Plants):
This field application of AE to monitoring the growth of an existing crack
was on the supercritical Eddystone #1 main steam piping (5000 psig,
1000F). Waveguides were tack-welded to the pipe. The AE activity/location
in startup appeared consistent with flaw growth, although that was not
verified. Event rates were of the order of 400 events per hour.
Brayton Point field experience (Thiele on Diagnostic Monitoring System
Development at Brayton Point):
As noted above for Steam Turbines, this was an early study by Rockwell
International (19771980) supported by EPRI and New England Power
Services Company where a total of 77 AE channels were employed to
monitor a range of components on four units, including tubes (boiler,
condenser, feedwater heater for leaks). All the AE monitoring was conducted
with 1/8-in. diameter metal rod waveguides and sensors with an integrated
pre-amplifier that was designed to significantly reduce the effects of
electromagnetic noise disturbance. Sensors were changed to a 500-kHz type
following the preliminary tests showing that broadband or low frequency
transducers would get saturated with 50-kHz noise. The main and reheat

1-15
steam lines of Unit 3 were monitored for cracking and exfoliation. There
were no reported correlations between AE and physical or other
nondestructive test observations. Details are reported in the Rockwell study
to be available in EPRI CS-1938 (July 1981).

Summary of Pre-Mohave Observations Relating to the Use of AE [14, 15]


Leak detection with AE appears relatively straightforward.
The cracking process is detectable and locatable with AE, although the
configuration of sensors, type of sensor, waveguide design, filtering, etc., are
variables that need to be refined, given the noise- and attenuation-related
issues associated with the emission output.
Noise and attenuation are the most significant factors affecting data
acquisition, analysis, and interpretation.
Noise data reduction may be achieved with the use of band pass filtering and
sensors with resonant frequency removed from the characteristic broadband
noise frequency range (typically <300 kHz in most flow-related applications).
The AE frequency range of relevance differs for boiler tube leaks (low
frequency is typically <20 kHz) versus for cracking (higher frequency is
typically >100 kHz).
Event amplitude distribution, F(V) is suggested as a source-type identifier:
the number of events exceeding a peak amplitude (PA) versus the PA
exceeded.
Event energy distribution is another potential means of characterizing the
source.
Spectral analysis is difficult and complex for many reasons related to
camouflaging the effects of frequency-dependent modification processes in
the material and limitations of bandwidth in the sensor and associated
electronics. A good substitute may be the rise time.
Signal duration tends to be proportional to counts, so pulse width may not be
needed when the ring-down count (RDC) is used.
The background noise in a 500-MW fossil plant varies from unit to unit.
Most noise consists of low frequencies (<50 kHz at least should be filtered).
The noise level on shutdown is low, lower than instruments, whereas noise
levels on runup are much higher.
Waveguides are needed for high-temperature component monitoring.
Attenuation (damping) through waveguides increases significantly with
waveguide diameters <0.11 in. and for frequencies >400 kHz. Waveguide
length has no effect on damping at <400 kHz, although waveguide diameter
needs to be limited. Recommended waveguide diameter is 0.20.4 in.

1-16
Post-Mohave, 19851990

Shortly following the Mohave and Monroe hot reheat long seam-welded low-
alloy steel piping failures [1, 2], EPRI initiated project RP 1893-4 in November
1986 to help advance methods for condition monitoring of steam lines. The
project, executed by Babcock & Wilcox (B&W), recognized the potential and
focused on the development of AE for such monitoring. Using prior knowledge
and experience (for example, [14, 15] as summarized earlier) and additional field
and laboratory experience, B&W produced a set of guidelines for applying AE to
monitoring high-energy piping [9]. These guidelines were intended to
supplement EPRIs 1987 edition of its long seam-welded piping evaluation
guidelines [3].

In-Plant Testing

Several in-plant AE tests provided useful experience that went into the
development of a suitable AE test approach. Tests were conducted at:
Ohio Edison (New Castle)
Pennsylvania Power & Light (Brunner Island #1, 2)
Orange & Rockland (Lovett)
New York State Electric & Gas (Somerset)
United Illuminating (Bridgeport Harbor)
TVA (Kingston #7)
Pennsylvania Electric (Keystone, Seward)
Philadelphia Electric (Cromby)
Houston Lighting & Power (PH Robinson #1)

The high-energy piping plant AE test experience lessons learned, as reported in


EPRI report RP 1893-4 [9], are briefly summarized:
Testing showed that at least some minor defects or UT indications could be
detected with AE.
- For high-frequency sensors up to 500 kHz, waveguides cannot be spaced
more than about 17 ft apart for a source location accuracy on the order of
1 ft or better.
- AE testing is highly dependent on instrumentation, methodology, and
the users engineering expertise.
- AE from startup can be quite noisy and difficult to analyze and interpret,
while AE data from over-pressure and over-temperature stressing were
relatively easy to analyze.
- AE could be used to show the absence of defects, based on one case with
New York State Electric & Gas Somerset station monitoring where
startup and temperature swing (to 75F below normal then back to

1-17
normal) using 500-kHz sensors and a fixed threshold gave no significant
AE, correlating with no recordable UT indications.
- Using a floating threshold is an effective way to eliminate noise in data
analysis.
- A 300-kHz resonant sensor with a 200-kHz high-pass filter may be the
optimal choice for locations with substantial noise, such as near the
turbine (concluded from the TVA Kingston #7 unit monitoring).
- Correlating AE with plant loading conditions is important to help
identify the nature of the source.
- Sensor location and placement may be important, particularly at
connections (tees, etc.) where girth welds are also of interest.
- Thermal transients are the primary AE-causing stress (as opposed to
pressure changes) during startup. In fact, limited offline testing using
steam injection to induce thermal transients in one case (Philadelphia
Electric-Cromby) gave an AE source location identical to that seen with
online startup monitoring.
- Load swings and cooldown stressing cycles are attractive since they are
not associated with excessive background noise and since they do not
represent unit overstress situations.
- None of the in-plant testing involved any cases where the AE response
was correlated with in-service cracking. This uncertainty drove the
laboratory testing for cracking detection verification.

Laboratory Simulation

The B&W laboratory testing involved simulated rupture of an approximate 5-ft


section of retired Grade 11 (1-1/4Cr1/2Mo) steam line under flowing steam
conditions. The simulation was achieved by introducing two deep (95% wall
penetration) electrodischarge-machined (EDM) notches in the heat-affected
zone (HAZ) of the longitudinal weld. Six waveguides were welded to the surface,
four offset from the weld and two on either end of the weld seam. Multiple
sensors were mounted on each waveguide (using oversized conical receptors).
These included wideband and 150-, 500-, and 800-kHz resonant frequency
sensors. Multiple data collection instruments were also used. Pipe rupture was
achieved at one of the EDM notch locations at about 1300F and 520 psi. Key
findings were:
Use of a floating threshold was necessary in the presence of high background
noise.
Burst emission data can produce source location that correlates with notch
growth and breakthrough.
Filtering was important to eliminate noise.
Leak detection following breakthrough was clearly identified by all sensors,
with the leak typically broadband with amplitude more than the background
noise typically seen with in-plant lines.

1-18
Noise was primarily below 200 kHz.
Optimized sensor for the test would be a 300-kHz resonant sensor with a
high-pass filter at about 200 kHz, consistent with the TVA Kingston #7 in-
plant test findings.
The AE signals (events) from notch propagation had broadband frequency
content, indicating that sensor selection for crack monitoring may be made
on the basis of background noise only.

Summary of AE Monitoring Guidelines Recommendations

The first set of high-energy piping guidelines was developed [9] with the
following main features:

Waveguides

Diameter: Preferred waveguides are 1/81/4 in. stainless steel rods.

Attachment: Permanently welded to the pipe surface using appropriate ASME


Codecompliant weld attachments.

Length: Waveguide length is preferably short (12 in.); although, for the
recommended diameter, the attenuation effects are minimal for longer lengths.

Spacing: Depends on several factors relating to sensor response frequency and


acoustic attenuation (spacing can be higher with reducing frequency and
consequent lower attenuation); potential interference from extraneous sources
such as hangers, radiographic testing (RT) plugs, etc.; geometric and path
constraints and location accuracy needs in cases of connections such as tees. A
300-kHz resonant sensor would require waveguides spaced no more than 20 ft
apart.

Sensors

To avoid noise, sensor frequency is preferably above the noise band (typically,
<300-kHz noise for the high-energy piping application). However, too high a
resonant frequency means greater attenuation and energy loss, so the required
spacing becomes smaller. 300 kHz resonant sensors with high-pass filtering at
200 kHz were found to be generally appropriate. In case of high noise conditions,
such as near the turbine, 500-kHz sensors may be used, and these would be
spaced at 15 ft or less. Uncommon, but in low noise situations, 150-kHz sensors
with a 100-kHz high-pass filter may work, and these may be spaced up to 25 ft
apart.

Sensors with integrated preamplification are convenient for mounting and are
preferred.

1-19
Threshold

Due to typically varying noise levels, it was concluded that fixed thresholding
does not generally work well. Instead, use of a floating threshold that varies
automatically with background noise levels is recommended.

Stress Methods

No conclusions were made regarding the most suitable stress method, and
startup, cooldown, temperature and pressure overload, and temperature and load
swings were all shown to induce AE activity. Startups, although producing the
most stressful conditions, were found to be very noisy. Overload conditions on
the order of 5% in temperature or pressure were found enough to activate
sources. Similarly, temperature and load swings can be useful and relatively quiet.
The temperature swing tests are recommended to be performed as rapidly as
possible.

At the time, startups and overload conditions were the recommended stressing
methods, although it was noted that more experience was needed with other
methods.

An important recommendation was that the AE data needed to be correlated


with operating conditions.

Source Location

Identifying the source location is key to the acceptance and rejection of data. The
use of automated time arrival difference algorithms to identify regions where
events are clustered (clustering algorithms) is recommended. No specific
algorithmic methods (single versus multi-dimensional) were recommended.

Offline Monitoring

At the time of the development of these guidelines, offline testing using thermal
stressing by steam injection was being investigated and reportedly showed
promise.

The 1995 EPRI Guidelines

Following the development of the high-energy piping AE monitoring guidelines


under EPRI RP1893-4 as described above, EPRI project RP1893-20 was
launched to expand and refine the guidelines and to demonstrate the application.
The joint EPRI-Pacific Gas & Electric (PG&E) effort executed by PG&E
resulted in publication of the high-energy piping AE monitoring guidelines in
1995 [8]. The unpublished details of the testing and bases for the guidelines have
been documented [10] and made available to this project. The focus of the
project was long seam-welded hot reheat piping, and the primary damage
mechanism of concern was creep, although the guidelines were developed to
include monitoring of all high-energy piping.

1-20
The project was executed in three phases:

Phase 1: Field testing on a section of long seam-welded hot reheat Grade 22


steam line at PG&Es Potrero #3 unit in the period 19901992. This formed the
bulk of the field testing to examine monitoring strategies, instrumentation, and
stimulation (stress) methods and to correlate results with other nondestructive
examination (NDE) findings.

Phase 2: A laboratory test program conducted in 19921993 to generate and


study typical AE sources using a section of hot reheat pipe taken from the
structure tested in Phase 1.

Phase 3: Development of standardized guidelines and their proof-of-method


application to a full-scale test conducted in 1994 at PG&Es Potrero #3 hot
reheat lines.

This section summarizes the key aspects of the field and laboratory testing,
followed by a brief description of the guidelines.

Field Testing

Based on the findings from the previous EPRI guidelines development study [9],
a portion of the PG&E Potrero #3 hot reheat line that was constructed of Grade
22, 18-in. diameter x 0.875-in. nominal wall thickness and that had 160,000
hours of service was instrumented and tested. The operating maximum
conditions were 1010F and 600 psig. The section that was the focus of the study
was a 100-ft segment of the East lead. A much smaller West lead segment and
the cross-tie were also instrumented. The waveguides and instrumentation setup
is shown in Figure 1-6. The waveguides were linearly aligned, spaced 13 to 15 ft
apart, and consisted of 1/4-in. diameter stainless steel rods. Full penetration weld
attached configurations were used, although mechanically coupled configurations
were also examined for practical feasibility. Several resonant sensors were used, as
illustrated in Figure 1-6, but the vast majority of data analyzed were with the
300-kHz sensors. A linear locating algorithm was used for clusters, and activity
data (# located events per hour) were analyzed using a floating threshold.

Initial testing in AprilJune 1990 was geared toward confirming that AE signals
were distinguishable from background noise and exploring the best stimulation
(stress) method. The loading cycles investigated were offline hydro-
pressurization, offline steam injection, plant startup, and plant cooldown.

1-21
More detailed testing was conducted in 19911992 using essentially the same
setup as the 1990 tests, but restricted to a 73-ft segment on the East lead.
Additional stressing methods, specifically online load swings, and the
repeatability of monitoring results were investigated. A summary of the testing
with the identified source locations and their correlation with other examinations
is given in Table 1-1, reproduced from the final (unpublished) report on EPRI
project RP1893-20 [10]. The referenced data files and figures are in that
reference. The correlation of the identified clusters with the results of other
examinations and tests is discussed later.

1-22
Figure 1-6
PG&Es Potrero #3 waveguide and instrumentation setup [10].

1-23
Table 1-1
Identified AE location clusters compared with results from other examinations and
tests on Potrero #3 [10].

1-24
The following were the significant findings from the field testing:

Stimulation (Stress) Method


Offline hydro-pressurization was difficult and costly to implement, and the
tests gave no evidence of emissions that passed the filtering criteria used. The
findings reflected a concern with a possible Kaiser effect and as to whether
this stress method would provide indications of damage in operation.
Offline steam injection appeared difficult to control, producing uneven and
unknown stimulation of various areas. As in the case of the offline hydro-
pressurization, the tests gave no evidence of emissions, and the findings
reflected a concern as to whether this stress method would provide
indications of damage in operation.
Startups were accompanied by relatively high background noise (RMS
averaged 56 dB with a maximum of 62 dB in the 192 tests), and a floating
threshold of 56 dB was used. The higher threshold setting runs the risk of
missing relevant AE sources, although in this case, location clusters were
identified nonetheless.
Online load swings were found to be the most consistent in AE with noise
levels at manageably low levels. Temperature changes of 30F50F were
sufficient to stimulate some sources.
Cooldown tests appeared to be useful only with large thermal changes
(200F300F drop).
As discussed below, since there were no identified instances of creep
cracking, the optimal stress conditions for detecting creep damage were not
established. However, as a matter of data acquisition and analysis feasibility,
the online conditions represent the ones with the most potential for the
high-energy piping application.

Waveguides
A limited study in conjunction with the field testing indicated that
mechanically coupled waveguides (as opposed to the weld-attached ones) had
significant transmission losses, on the order of 1520 dB, much greater than
an approximate 6-dB loss seen with the welded configurations. Additionally,
the couplant (required for the mechanical coupling) typically degrades at the
high pipe surface temperatures, resulting in changing losses and frequent
couplant replacement. The performance of all available high-temperature
couplants was not examined, although any couplant is expected to degrade
with time at temperature.
The full penetration gas tungsten arc welding (GTAW) or shielded metal arc
welding (SMAW) weld attachment ensured good acoustic connection.
The 1/4-in. diameter stainless steel rods proved adequate, and the
temperature gradient offered with these rods was a drop from about 1000F
to less than 120F over the 12-in. rod length used.

1-25
Calibration
Pencil lead breaks for generating a wave were found capable of providing
adequate signal amplitude 15 ft away with application of an 80-dB gain.
The mechanical pencil, however, does not survive when used on a hot pipe
surface, so the method is limited to application in the cold pipe condition.
A hand-held center punch with the center punch pin removed (to preclude
damaging the pipe) generated reasonably consistent signal amplitudes in the
near and far field and is a calibration option.
The lead break, having lower signal dispersion than the punch, is better
suited to velocity calibration, while the punch with its relative ruggedness and
larger signal amplitude works well for near-field sensor sensitivity checks.
Several broadband pulsers were experimented with as a calibration source,
with the highest energy pulser at 220 V peak-to-peak with 94 J of energy.
The pulsers did not generate sufficient energy to be detected. It was
concluded that this method of calibration has potential with pulsing integral
preamp sensors, but that spike pulsers are not capable of producing sufficient
energy over a narrow bandwidth. Later laboratory testing developed an in situ
pulsing system (waveguide mounted for signal to adjacent waveguide/sensor)
consisting of a pulsable integral pre-amp sensor with high voltage pulsing
capability.

Pipe Signal Attenuation (see Figure 1-7)


For the 80-dB gain used, a near-field 8085-dB signal attenuated to
2025 dB 15 ft away. The attenuation is rapid (exponential with distance)
over the first 2 ft or so (geometric spreading of the wave), but then it
becomes dominated by the structure and material characteristics, resulting
in a more gradual, linear loss of amplitude with distance.
The linear algorithm location method requires that the AE event be detected
at two adjacent sensors. If the source is very near a waveguide/sensor, the
attenuation to the adjacent waveguide/sensor is a maximum and has the
lowest probability of detection by that sensor. In contrast, an event source
roughly midway between waveguides/sensors experiences the lowest
attenuation before arrival at the adjacent waveguide/sensor. Thus, the least
sensitive locatable area is near a waveguide, and the most locatable area is
midway between waveguides.

1-26
Location Accuracy (see Figure 1-7)
Accuracy varies with distance to the nearest sensor and the circumferential
position around the pipe from the waveguide line or axis. The best accuracy
is for a location on the waveguide axis, midway between waveguides, and the
least accuracy is at the axial waveguide location on the diametrically opposed
side of the pipe from the waveguide.
For the 18-in. diameter pipe and the 15-ft waveguide spacing, accuracy varies
from a best case of 5% to as high as 17% of the span length.
Events outside the waveguide array can cause location errors (end-of-array
effect) due to a pile-up at the arrays end sensor, creating the appearance of a
location cluster.

Figure 1-7
Schematic illustrating event location accuracy [10].

AE Test Performance (see Table 1-1)


The online load swing (160F temperature drop) and cooldown from high-
temperature tests produced the best match of located clusters with the results
of UT and destructive examination.
Both the cooldown and online temperature load swing tests identified a
location in the approximate vicinity (with 23 ft error) of an observed
thermal ratchet crack at a seam weld (89.5-in. location). The thermal
stimulation appears to provide a means of detecting thermal- or fatigue-type
cracks.
An apparently inactive slag inclusion area at a seam weld (31-in. location) did
not produce detectable AE, supporting the potential of AE to screen inactive
defects.
Likely thermal fatigue cracks near a girth weld (location at 429 in.) were
detected by the cooldown in online temperature load swing tests.

1-27
A seam-weld UT indication location (525 in.), nature unknown, was
identified with the cooldown and online load swing tests.
Locations with UT indications at an elbow circumferential weld (667 in.)
and RT and UT indications on a cast elbow (691710 in.) were source
identified approximate locations by the cooldown and load swing tests. RT
indicated micro-shrinkage porosity in the case elbow.
A support saddle-to-pipe fillet weld crack was conclusively located with all of
the stress methods, except by the cooldown from a medium-temperature
stress, which appears to be a stress method that did not produce useful
indications.
The AE system did not produce an unwieldy number of calls for disposition
and also did not produce known false calls.
The AE system appears to readily identify active fatigue-type cracks and, in
at least one instance of a slag inclusion area, was shown to be an effective
screener for inactive defects.
There were no cases of known creep damage/cracking for the pipe section
monitored. As a result, the system was not proven to detect creep damage in
the field.
The AE system results correlated reasonably well with observations of cycle-
or time-independent cracking, and its performance for time-dependent creep
damage and cracking remained unknown.

Laboratory Testing

A 15-foot section of the previously field-monitored Potrero #3 hot reheat piping


was removed for laboratory testing conducted in MayOctober 1993. The testing
objectives were to identify signal features associated with source types, determine
effects of wavepath through thin-wall pipe, determine the best classification
strategies for different signal types, and relate thermally stimulated sources to the
field test data. Most testing was conducted on the segment of piping from
location 0 to 110 in. (see Table 1-1) that had two long seam-weld indications.

Axially aligned waveguides were spaced at 51.5 in., that is, 1/4 of the field test
span length. Instead of a single waveguide at an axial position, three waveguides
were positioned at a single axial position, spaced at <1.5 in. apart, as seen in
Figure 1-8.

1-28
Figure 1-8
Schematic illustrating the laboratory test setup [10]; LF and SF represent known
indication locations corresponding to locations at 31 and 89.5 in. in Table 1-1.

A number of different sensors (300, 400, and 500 kHz with 40-dB integral
preamp and single-ended broadband sensors) were experimented with and
responses characterized for a pencil lead break and for a pulser. Additionally, AE
was documented for a range of induced conditions including lead break, attached
steel bolt cracking, center punch, oxide cracking (oxide drop, fracture and
crushing), impact, air blast, rubbing, and the thermal simulations involving
heatup and cooldown with induction heating. For these tests, to perform source
location and collect data, a standard AET 5500 computer-based AE system with
80-dB gain was employed. Thresholds were usually fixed, except where varying
noise was expected. This was typically less than 46 dB (1.5 V). Filtering was
high-pass, post-amp, at 200 kHz.

The major findings from the laboratory experiments were as follows:

Source Type
Oxide sources (cracking, spalling) and metal cracking (from the attached
cracked bolt experiment) are distinguishable from other mechanical sources
such as rubbing, impact, sudden flow perturbations (for example, air blast).
The signal pseudo-frequency (ring-down count/event duration) is typically
>50 kHz for the cracking sources, while noise signals from flow turbulence
and mechanical rubbing statistically cluster below 50 kHz. Signal rise time
for crack-like sources is relatively short, between 50 and 400 s, whereas
noise signals have a wide spread of values, without any indication of
clustering at low rise time values.
Review of the test findings indicate that distinguishing oxide cracking and
spalling from crack growth may be difficult. Indeed, the heatup-cooldown
experiments produced general area signals from pipe surface oxide spalling,
comparable to those from the bolt cracking experiments, except that there

1-29
was no apparent clustering. Further, as seen in Table 1-1, the field AE
produced consistent signals in cooldown and online monitoring from
locations roughly corresponding to the 89.5 in. UT indication location that
was subsequently sectioned and found to contain a small inclusion (no crack).
Figure 1-9 [10] contains macrographs showing the cross sections at the
locations corresponding to 31 and 89.5 in. in Table 1-1 (re-located in the
laboratory tests to locations at 18.5 and 73 in.).
Since the tested pipe segment apparently did not have any propagating flaws
and did not have any existing creep damage, the experiments were unable to
provide any information on the AE relating to damage that may be
encountered in high-energy piping service.

Source Location
Location accuracy using a linear locating algorithm is a function of the
sensor-spacing-to-pipe-diameter aspect ratio. As this ratio decreases, the
circumferential wavepath length increases, and the locating axial error for
events off-waveguide alignment axis increases. Increasing sensor spacing to
1718 ft may be called for with pipe diameters in excess of 25 in.
Events outside the waveguide array cause location errors (end-of-array effect)
due to a pile-up at the arrays end sensor, creating the appearance of a
location cluster. Emissions as much as 5 ft outside an array can cause event
piling.

1-30
Figure 1-9
Macroetched sections of the laboratory-tested pipe [10] showing (top) the slag
inclusion corresponding to the 89.5- in. location in Table 1-1 and the SF location
at 73 in. in Figure 1-8, and (bottom) the inactive crack corresponding to the 31 in.
location in Table 1-1 and the LF location at 18.5 in. in Figure 1-8.

1-31
Threshold
The experiments showed that floating thresholds under reasonable
background noise conditions (<1V RMS noise at 80 dB gain, equivalent to
40 dB) would enable sources to be accurately located.
The AE equipment should employ a moderate time constant (rapid, but not
too rapid as to truncate high-amplitude events), floating threshold deadband
that is a linear, not logarithmic, function of the background RMS noise.
Pseudo frequency becomes erratic when there is less than 13 dB difference
between the signal amplitude and the threshold.
Linear location is subject to increased error when the difference between the
signal amplitude (at the second-hit sensor) and threshold is less than 10 dB.

Calibration
A pulser system was developed that has application to the field. A high-
voltage tone burst (100200 V) was used to pulse through a waveguide-
mounted transducer to the adjacent waveguide/sensor (in situ pulsing).

Attenuation
Figure 1-10 shows the measured attenuation from a lead break with 80-dB
gain and the IR 300 SWT sensor (300 kHz resonant with 40 dB integral
pre-amp). Note the drop of about 15 dB over the first 28 in. (1/2 pipe
circumference). The curve-fit indicates an overall attenuation of about 25 dB
for a 15-ft waveguide span, comparable to the field testing measured
attenuation of 2025 dB. Since the far-field attenuation (after the initial
geometric wave spreading) is ~ 1 dB/ft, increased waveguide spacing has a
minimal effect on signal loss.
For an AE event next to a sensor, the first-hit sensor will detect it without
significant attenuation, whereas the second-hit sensor will detect the signal
with 2025 dB attenuation. For a mid-span event, the prediction is about 20
dB attenuation to the sensor.

Waveguides
Waveguides do not significantly alter frequency content, but they do affect
signal loss. The measured loss was 35 dB for wide and narrowband sensors.
This includes the attenuation through the 12-in. rod. Signal loss along the
rod length was estimated to be about 0.2 dB per additional foot beyond the
12-in. length.
The waveguide orientation to the pipe surface creates a wavemode conversion
from plate wave to compressional rod wave, and the efficiency of this
conversion relates to the rod diameter. If the rod diameter is less than the
wavelength transmitted, phase cancellation is not a factor. Diameters larger
than the 1/4 in. used may affect the mode conversion of higher-frequency
signals in addition to increased attenuation.

1-32
Figure 1-10
Attenuation measured for a lead break source with a 300-kHz resonant sensor,
40-dB integral pre-amp, and 80-dB gain [10].

Proof-of-Method Test and the Guidelines

The third phase of the Guidelines development project involved a proof-of-


method test conducted at Potrero #3 hot reheat East and West leads in a plant
cooldown during February 1994 and in online operation in AprilMay 1994.
Considerations in the test included the ability to install waveguides during
operation, the ability to perform and automate calibration of sensors/channels,
the capability to scale up AE monitoring efforts to include all hot reheat piping,
and the need and method for monitoring operating conditions. In addition, the
inspections following the AE monitoring provided more correlations with
identified sources/location clusters.

1-33
Findings were as follows.
In the case of cooldown, most of the AE information of value was obtained
in the first 300F temperature decrease.
Online monitoring identified clusters in addition to those seen in the
cooldown; about 40% of the identified locations were also identified in the
cooldown. The online condition cycled between high and low plant
load/pressure, typically creating a temperature transient of about 20F30F.
A source ranking system was used for the located clusters, details of which
are not reported. Qualitatively, in the cooldown, the source rank was based
on the total events, total counts, and normalized amplitude ranking (factor
based on signals >70 dB). For online monitoring data, the source ranking was
based on event density (time rate), count density (time rate), and normalized
amplitude ranking as it was for the cooldown.
Tables 1-2 and 1-3 summarize the AE sources and results of other NDEs on
the East and West leads, respectively. Note:
- Ten identified clusters with source ranks from 0.10 to 0.67 that include:
o Seven inaccessible locations without UT correlations
o Three true positives (UT indications)
- No significant false calls, except that in at least one instance there was a
substantially high event rate on startup that was rejected. The example
highlights the need to compare results from various test modes.
- Reasonable correlation with UT
- Good capability to detect thermal cracking
- No correlation with creep damage made in the study

It was clear from the field and laboratory testing that an optimal AE monitoring
method was developed and demonstrated for detecting high-energy piping crack
damage, specifically time-independent and fatigue cracks. The performance of
the system for detecting time-dependent creep damage as cavitation or cracking
remained to be seen. Nevertheless, a set of guidelines for AE monitoring of high-
energy piping was drawn up [8] based on this work [10]. Figure 1-11,
reproduced from EPRI report TR-105265-V1 [8], is a schematic showing the
various key elements of the AE test method. The important features of the
elements are summarized.

Stress Method

Online monitoring (steady-state), online load swings, and cooldown are the
recommended stimulation modes. These are to be executed within the envelope
of plant operation; that is, as passive stress methods. Active stress methods
outside the operational envelopes, such as online over-pressurization or over-
temperature and offline hydro-pressurization or steam injection, are not
recommended. Stress methods are suited to types of damage; for example,

1-34
thermal fatigue cracks are more likely activated by thermal stimulation, whereas
creep cracks are more likely activated under online conditions steady state or
with pressure increases. The guidelines recommend a combination of monitoring
modes:
Online monitoring under steady-state load conditions: At least one week of
continuous monitoring for each section of line, with at least 25% of the
monitoring period at peak load and for at least a 2-hour period.
Load swing monitoring: Normally combined with online monitoring. A
minimum 25% reduction in peak load pressure during low load cycles is
recommended, with at least five load cycles being monitored.
Cooldown monitoring: Monitoring should begin at operational conditions,
preferably at peak load. Target cooling rate for the initial decrease is 50F
100F/hour with monitoring over at least the range of 950F650F (as
developed for the low-alloy steels).

1-35
Table 1-2
Active AE clusters compared with NDE indications on Potrero #3 East hot reheat
lead [10].

1-36
Table 1-3
Active AE clusters compared with NDE indications on Potrero #3 West hot reheat
lead [10].

1-37
Figure 1-11
Key elements of the 1995 EPRI high-energy piping AE monitoring guidelines [8].

Stress Methods

The Guidelines recommend three usable methods:


Cooldown
Online load swings
Online steady state

1-38
Cooldown

The cooling rate should use the typical plant cooldown practice as much as
possible. A minimum average cooldown rate of 50F/hour, between 1000F and
700F (for a full load temperature of 1000F) is needed. A target rate of
100F/hour is preferred.

The piping system depressurization and the start of the temperature cooldown
should coincide. The piping system should not be completely depressurized until
the temperature has fallen at least 50F from the initial operating temperature.

The AE monitoring should start at the plant operating temperature and pressure.
It is recommended that AE monitoring start at the system pressure associated
with full plant MW load. Monitoring should continue until: (1) the pressure is
zero, and (2) the temperature is less than 600F (or until AE data activity reduces
to <1 locatable event/minute).

Online Load Swing

The loading rate needs to match or exceed normal plant load changes. This
includes the rate of change of temperature, pressure, and gross megawatts. Gross
megawatts provide an indication of changes in the steam flow rate.

The load swing should include a pressure change of at least 25% of the piping
operating pressure range (for example, operating pressure of 500 psig down to
375 psig).

Monitor for several load cycles to provide a statistical representative sampling of


normal plant operation. Typically, this should be at least five cycles for each pipe
section.

Online (Short-Term) Steady State

Monitoring needs to be performed at full power steady-state plant operating


conditions. Steady state is typically where temperature change is less than 20F
and pressure change is less than 20 psi.

Monitoring should continue until a statistically representative sampling of


normal plant operation is obtained. Typically, this should be at least five days of
steady-state operation for each pipe section monitored. Steady-state operation is
defined as periods of constant load, with a minimum of 2 hours between load
swings.

1-39
Waveguides
1/4-in. diameter stainless steel rod with either a conical or flat receptor for
the sensor (see Figure 1-3)
Waveguide length is not critical, although signal loss is on the order of 0.2
dB/foot beyond the 12-in. length used in development of the guidelines.
Waveguides are to be permanently weld-attached to the pipe with full
penetration GTAW or SMAW or by stud welding such as capacitive
discharge, all compliant with the ASME B31.1 Power Piping Code. Details
of the recommended rod tip geometry in EPRI Report TR-105265-V1 [8]
describe a recessed area 0.078 in. diameter x 0.050 in. depth in the rod tip to
house an aluminum flux ball.
Mechanical waveguides may be used for some short-term monitoring, but
inconsistency in performance relating to the couplant and varying contact
losses make this configuration undesirable.
Waveguide spacing of 1518 ft is recommended. Larger diameter piping
(>25 in.) may benefit from larger spacing (1718 ft) to minimize the
geometry effect (spacing to diameter aspect ratio) on location inaccuracy.

Sensors and Instrumentation


Multi-channel computer-based systems with standard AE signal
characterization, a floating threshold capability, and active source location are
required for all monitoring modes.
Sensors with resonant frequencies in the range 300400 kHz are
recommended. To reduce the influence of flow-related noise, high-pass
filtering at 200300 kHz is recommended in the secondary amplification
stage.
Instrumentation capability should include active source linear location and
storage of all AE event data that pass the initial filtering criteria.
Floating threshold is recommended for all monitoring modes. Criteria and
tests specified in the guidelines should be followed in order to ensure the
floating threshold performance, including deadband settings and the
response time constant of the threshold circuit.
The instrumentation preferably has the ability to record time-sampled
operating data such as temperature, pressure, unit load, and flow rate.
High-temperature coaxial cable is recommended.

1-40
Data Analysis/Evaluation

Figure 1-12 is a schematic from EPRI Report TR-105265-V1 [8] that shows the
steps recommended for data analysis/evaluation. Stored AE data are to be
evaluated for signal characteristics, specifically peak amplitude, ring-down count,
event duration, and pseudo frequency. However, specific criteria for event
selection were not developed, and guidelines for identifying significant events
(signals) are given in a qualitative fashion. Important aspects listed here,
nevertheless, are the following:
The basis for AE data interpretation is identifying and locating source
clusters. Source location cluster analysis and cluster ranking are to be
performed based on cluster features, relative activity levels (event rate, count
rate), and correlation with plant parameters. Statistically significant clusters
are based on activity exceeding a minimum level under a given stimulation
(stress) mode.

1-41
Figure 1-12
Key elements of the data analysis/evaluation [8].

1-42
Guidelines for the analysis and interpretation of the data were not fully specified
in quantitative terms (for example, for step 4 ranking, no details of the criteria,
such as for event duration, are given), and the potential subjectivity of data
analysis and interpretation remains an issue (see the Current Practice discussion
below).

Comments on the steps of Figure 1-12:


1. Identify event locations: Identify AE event locations using the linear location
method. Typical filtering limits for acceptance of individual sensor data are
>10 ring-down count, and >10 microseconds event duration. Use of linear
location to identify locatable AE events is key to data analysis.
2. Identify loss of detection sensitivity: Identify data where a raised floating
threshold may decrease detection sensitivity. Sensitivity loss can be detected
by evaluating recorded threshold levels for located AE events. It can also be
detected by recording RMS voltage per channel.
3. Identify source location clusters: Identify location clusters that are statistically
significant. Filter on located AE events to eliminate AE events with signal
features similar to flow or rubbing noise. Typical filtering limits for
acceptance are: >25 ring-down count, >50 KHz pseudo frequency (ratio of
ring-down count/event duration), and <400 microsecond rise time. This
filtering is to be performed only on the first hit sensor data of located AE
events.
4. Rank location clusters: The location clusters should be ranked based on their
level of activity, intensity, and stimulus type. The intent is to identify
dominant location clusters. The ranking needs to consider the following:
- Normalization of data to address signal attenuation effects
- Effects of a floating threshold
- Differences in emission, from different source types (for example, signal
amplitude from crack growth may be less than from oxide fracture)
- Differences in stimulus from different stress methods

Ranking can be performed using event activity (number of events) and peak
amplitude. Peak amplitude is not sensitive to variations in AE monitoring
equipment design and floating threshold effects. Filtering for ring-down
count, pseudo frequency and rise time will eliminate many low energy events.
The remaining events can be adjusted for expected signal attenuation,
separated into bands of peak amplitude ranges, and then ranked based on the
activity level from those bands. The ranking should be performed for each
stress method test separately. Then the rankings compared and combined as
part of identifying active AE sources.
5. Identify active AE sources: Those location clusters with a significant source
ranking should be evaluated to determine if they are active AE sources.
This provides an interpretation of the monitoring results. For each location
cluster, perform this evaluation by considering (a) correlations between event
activity and expected stimulus; (b) loss of detection sensitivity on data;

1-43
(c) cluster statistics and sensor location (signal feature statistics and cluster
statistics, such as a clusters standard deviation and location relative to sensor
position); and (d) trending (use AE monitoring data from previous tests to
establish trending of the AE source's activity).
6. Identify follow-on inspection locations: Identify those active AE sources to
be of interest for follow-on inspection. In selecting locations, a priority for
inspection needs to be given. This should be based on:
- Source ranking
- Criticality for crack growth in different types of pipe components (for
example, potential for through-wall crack growth in seam welds versus
drain or vent connections)
- Safety concerns for personnel protection in different plant areas

Use location accuracy to determine the most likely location of a cluster and
range of pipe length for follow-on inspection. Typically, this is 2 feet from
the average location.

Location Accuracy

The accuracy of a sources linear location is influenced by several factors:


Aspect ratio of pipe diameter to sensor spacing distance
Distance between the AE source and waveguide (signal attenuation)
Pipe geometry (straight pipe, bend, elbow, tee, and reducer)
Signal amplitude relative to equipment threshold and background noise level
Source type (signal shape characteristics)

Straight pipe: Table 1-4 provides the average linear location accuracy for straight
pipe. The table shows that accuracy improves for smaller aspect ratios and
increasing distance from a waveguide. The aspect ratio distorts the AE source
location by making the location appear to be closer to the center of the span
length than its true location. But when a source location is very close to a
waveguide, within 11%, the true location may be in the next waveguide span.
The maximum error occurs for events originating on the opposite side of the pipe
from the waveguide.

Elbows, bends, and tees: When an elbow, pipe bend, or tee exists between two
waveguides, the accuracy error is increased by their geometry. Table 1-5 shows
the average accuracy for a pipe elbow, for medium to high peak amplitude events.
For low-amplitude events, the error can be much larger. As with the straight
pipe, the aspect ratio distorts the AE source location. True linear location is more
likely to be closer to the nearest waveguide than measured location. The
likelihood increases as the elbow/bend radius decreases and the aspect ratio
increases.

1-44
Table 1-4
Linear location accuracy for straight pipe [8].

Table 1-5
Linear location accuracy for bend/elbow [8].

Monitoring Frequency

The recommended maximum AE monitoring interval is 36 months. The


maximum monitoring interval was based on potential for crack growth and the
need to establish trending of data, and developed on the basis of the low-alloy
steel long seam-welded piping creep damage experience. It was also
recommended that the evaluation of the cooldown and online monitoring data be

1-45
performed together. The online monitoring can be performed prior to a plant
outage and the cooldown monitoring performed at the start of the outage. The
evaluation and follow-on NDE inspections would be performed during that
outage.

Current Practice

An effort was made by EPRI to obtain information on current high-energy


piping AE monitoring practice from commercial service providers and EPRI
utility members that have been using AE to monitor their fossil power plant
piping. A substantial amount of information on practice and service experience
was provided by J. Rodgers, Acoustic Emission Consulting, Inc. (AEC), Fair
Oaks, California, a service provider, in the form of published and unpublished
documents, and verbal and written communications. Additional limited
information on practice and service experience was provided by I. Mizrahi,
Margan Physical Diagnostics Ltd. (Margan), Netanya, Israel, a service provider,
in the form of verbal and written communications.

Evolution of the 1995 Guidelines to Current Practice

19951999

In the 19951999 period following publication of the EPRI Guidelines, a


substantial number of high-energy piping AE monitoring projects were
conducted, mainly in accordance with the Guidelines recommendations [11, 16]
in order to build a database, verify the methodology, and refine evaluation
criteria, and evidently also as a result of the utility industry interest in using the
technology to prevent failures. The specific service experience relating to roughly
30 monitored low-alloy steel hot reheat units in this period is summarized in the
following chapter, but the main elements of the practice and its evolution are
described here. The information in EPRI reports TR-111558 and TR-113534
[11, 16] was supplemented by Rodgers in a written communication [17].

Stress Method

Since cooldown monitoring requires the simultaneous monitoring of all channels


and can be logistically difficult to plan and implement, this stress method began
to lose favor and was used in only about one-third of the 30 cases reported on for
the period. Where applied, the cooldown stress AE monitoring method
continued to exhibit the ability to identify sources that represented time-
independent or cyclic forms of cracking such as thermal fatigue.

Startup monitoring, as previously noted, is relatively noisy. As with cooldown


monitoring, it requires simultaneous monitoring of all channels. Although not
recommended in the Guidelines, startups were occasionally monitored and used
in roughly a third of the reported monitoring cases for that period. The data from
startups were not used as a primary indicator of a relevant source, and in most

1-46
instances, the source clusters identified in a startup persisted in online
monitoring. In any case, startups were seen to have potential value in identifying
the forms of crack damage (time-independent and fatigue) also identifiable in
cooldown monitoring.

Online monitoring under steady-state operation and with load swings became
the primary stimulation modes in the immediate post-1995 Guidelines period.
Given that a key objective was to identify creep damage locations and given that
online monitoring is relatively easy to execute (segments may be separately
monitored at different times, requiring fewer channels for any one monitoring
session), it was only natural that this stress method evolved as the main
stimulation method.

Waveguides

Waveguide configurations used were in accordance with the Guidelines, 1/4-in.


diameter stainless steel, typically 12 in. long, weld-attached to the pipe, axially
aligned, and spaced 1518 ft apart.

Sensors and Instrumentation

Sensors and instrumentation used were as recommended by the guidelines.


Sensors had resonant frequencies in the range 300400 kHz, with high-pass
filtering at 200300 kHz in the secondary amplification stage. Multi-channel
computer-based systems with standard AE signal characterization, a floating
threshold capability, and active source location were used for all monitoring
modes. Additionally, the instrumentation capability included active source linear
location, storage of all AE event data that passed the initial filtering criteria, and
ability to record time-sampled operating data such as temperature, pressure, unit
load, and flow rate.

Data Analysis/Evaluation

The data analysis/evaluation process underwent some change, as described by


Rodgers [11, 17]. The Guidelines-recommended Source Ranking system
reportedly based on an event and count rate with other stated, but non-
quantified, criteriawas refined mainly to include consideration of the pipe
length over which event clusters were identified, as noted below.

The initial filtering of signals for relevance was very similar to that recommended
in the Guidelines. According to Rodgers [17], located events are first filtered
based on ring-down count ( 25) and then by pseudo frequency (>50 kHz) levels,
as recommended in the Guidelines. The resulting source location clusters are
then evaluated for response to thermal or pressure gradients, as well as the
structure present at that location (seam welds, girth welds, hangers, welded
fittings, etc).

1-47
A more detailed cross-plot analysis is conducted on a selected area representing a
cluster location (extracted set of data) to examine signal features within the
cluster location and compute average values for the cluster, such as amplitude,
rise time, ring-down count, and pseudo frequency. The data are also used to
compute the standard deviation of the source location cluster, , as illustrated in
Figure 1-5 and reproduced here as Figure 1-13 for convenience. A normalized
event density ranking (NEDR)the event rate normalized for the length of the
location cluster given as # events/time/cluster length standard deviationis a
change from the straight event rate recommended in the Guidelines. Units of
NEDR are events/in.-hr., with the time period equal to the hours at peak load or
period of stressing.

Since the NEDR, developed in the 19951999 period [11], has become the
metric for interpretation of AE monitoring data generated in accordance with the
1995 Guidelines and has continued to be used extensively in high-energy piping
monitoring, comments regarding this parameter are in order. Most of the
description has been reproduced from EPRI report TR-111558 [11].

Normalized Event Density Ranking (NEDR) [11]

The conventional, published data interpretation parameter developed for


characterizing AE activity on high-energy piping monitored using the 1995
EPRI Guidelines is the NEDR, schematically shown in Figure 1-13. The
NEDR value for an AE cluster consists of several basic analysis steps:
Defining the location limits of the cluster
Computing the number of located events in the cluster over the time of
analysis
Computing the standard deviation of the cluster in inches
Determining the peak load time in the analysis period

The NEDR value is then determined by the equation: # cluster events/cluster


standard deviation/peak load time. The NEDR value is useful in comparing
activity on different line segments that may have been monitored at different
times and peak load conditions.

1-48
Figure 1-13
Definition of normalized event density ranking (NEDR) of a location cluster, a
source ranking method developed in the post-1995 Guidelines EPRI research [11].

Computing and interpreting NEDR is complicated by several factors:


The event rate is related to pressure achieved on each peak load cycle and its
consistency.
The numbers of cycles of loading in the analysis period will have some
effectcyclic loading produces higher event rates per hour of peak loading
than steady-state high load operation.
Loading may depend on other load factors, such as thermal changes and
accompanying load cycling. Elbows/bends will often show some thermal
sensitivity in addition to pressure sensitivity.
Unusual upset events, such as temporary overload or over-temperature
conditions, fast cooldowns, etc., can have profound effects on the event rate.
NEDR values are much higher under startup or cold restart conditions and
may take several days to a week to settle to steady-state values.
Piping that is operated in the load following mode may see peak loading only
periodically. Accumulated creep damage can then be released at higher rates
under peak loading than would be normal if the piping were operated under
base load conditions.
Higher noise levels on some areas of line, particularly near turbine inlets,
diminish the prospect of an accurate calculation because lower amplitude
signals (5060 dB) are excluded.

The use of the NEDR calculation for trending on successive tests must therefore
take into account the conditions under which each test was conducted.

1-49
In the 19951999 period, the roughly 30 reported cases of AE monitoring [11]
had NEDR values in the range, none or 0 to as high as 125 events/in.-hr seen
under startup conditions and relatively lower upper-range values of about
25 events/in.-hr under online conditions. Correlation of NEDR with other
nondestructive and destructive examinations is discussed in the next chapter.

The NEDR values are used to present a relative ranking of sites for follow-up
inspection. The following recommendations were developed based on the
locations and range of NEDR values calculated (since no = signs were reported
originally, the > sign has been conservatively changed to in this summary):
No pressure-related seam-weld sources No NEDR. Re-inspect with AE in
three to four years (the age of the line is a factor). The Guidelines
recommend a maximum 36-month interval, although that recommendation
was premised on the conservative damage prediction rates used in
development of EPRIs long seam-welded high-energy piping evaluation
guidelines of 1996 [4].
NEDR <1.0 No immediate inspection requirement. Re-inspect with AE in
two to three years (age of line a factor).
NEDR 1 and <5.0 Low activity sources. If confined to relatively short line
segments (<50 inches), inspect with advanced UT methods in one to two
years. If affecting long line segments (typically, 100300 inches) or if the line
is subcritically annealed, perform baseline UT in 612 months. Re-inspect
with AE within two years.
NEDR 5.0 and <20 Moderate activity sources, usually accompanied by
higher amplitude range of signals (5090 dB). Perform baseline UT in 612
months. Re-inspect with AE within two years.
NEDR 20 High activity sources. Perform baseline UT within six months.
Reinspect with AE in one to two years.

It is not uncommon to see a range of NEDR values on successive monitoring


periods covering the same structure. Ranges of 13 NEDR points would not be
uncommon for comparatively similar periods. Much larger differences can be
seen between periods near startups and several weeks later, on the order of 510
NEDR points. The use of NEDR for trending from one test to another several
years later must, therefore, take into account these factors.

2000Current

AEC

Since completion of the first phase of the post-1995 Guidelines field experience
(19951999), the methods and procedures described by Rodgers [11, 17] and
summarized above have remained unchanged. The method and procedure thus
described is representative of current practice in case of AECs high-energy
piping monitoring services.

1-50
Margan

Margan has separately provided the project with information regarding its
current methods and procedures for high-energy piping AE monitoring via
verbal and written communications [18, 19]. The main features, different in
many respects from the EPRI 1995 Guidelines recommendations, are
summarized.

Waveguides

Waveguides are typically (for low-alloy steel) constructed according to the EPRI
Guidelines, that is, 1/4-in. diameter stainless steel rod weld-attached- to the
pipe. Reportedly for sensitivity and due to attenuation, monitoring of Grade 91
piping uses a 0.1-in. diameter rod attached via capacitive discharge welding to
limit heat input as required by the ASME Boiler & Pressure Vessel Code,
Section I Code Case 2649.

Waveguide length is typically about 13.5 in. For Grade 91, a shorter waveguide
rod of about 9.8 in. is used, reportedly to partially compensate for the loss of
attenuation with the smaller diameter.

Waveguide spacing is typically 12 ft. For Grade 91, the spacing is reduced to 9 ft.
The spacing is lower than that recommended by the EPRI Guidelines, reportedly
due to the locating accuracy needed. The waveguide spacing in proximity of
wyes, tees, and the like are reduced to 3 ft or less for improved locating accuracy
and due to increased attenuation.

Waveguide locations are typically pipe axially aligned, and the linear locating
algorithm is used, consistent with the EPRI Guidelines.

Typically, an entire piping system is instrumented, involving a large number of


installed waveguides and sensors.

Sensors and Instrumentation

The sensor is grease-coupled to the waveguide, as has been conventionally done


in the EPRI work.

Sensors are 200 kHz resonant, different from the EPRI Guidelines 300400
kHz. Margan indicates that its R&D results with such sensors have produced
satisfying correlations between AE indications and propagating flaws in steels.
The sensors in this case, however, also require more complex noise filtering that
Margan reportedly practices. Background noise generally represents more than
95% of the acquired data, and there is expectedly a significantly greater data
processing burden associated with this approach. The philosophy appears to be
aimed at minimizing the risk of losing valuable data by using a sensor resonant
frequency band that is as low as possible, rather than choosing a sensor band
simply to avoid collecting noise data.

1-51
Calibration

Calibration is generally done under full load, steady-state unit operating


conditions using a pulse source at a waveguide. Each sensor is checked
individually, and the tests used are:
Sensor-detected energy response to a preset level of noise
The measured energy spatial distribution response
Attenuation measured at a sensor as a function of distance from the source

Stress Method

The stress methods do not include any startup or cooldown cycles. Three online
stress methods are employed:
Full load, steady-state: Most of the data are collected under these conditions.
Load swings (referred to as dynamic conditions, done repeatedly): Can be as
low as 10%; differs from the EPRI Guidelines recommendation for at least a
25% pressure drop.
Low load: 70% of full load conditions, particularly used under noisy
conditions to help set up filtering, not in the EPRI Guidelines
recommendations.

Margan does not consider major transients such as startup and cooldown to be of
value reportedly due to levels of noise in these transients and their irrelevance to
creep damage related activity that is of primary interest.

Data Acquisition

There are two primary modes of acquisition:


Continuous mode, typically taken during the day under full load operating
conditions. In this case, the threshold is initially set and remains fixed
thereafter; that is, not a floating threshold.
Burst mode, typically used during the night under low and varying load
conditions. In this case, a floating threshold is employed, so only significant
burst event data are acquired.

Data are typically acquired over a three-day period.

Margan reportedly emphasizes periodic monitoring to establish changes and


trends; they call this TMI or trend monitoring inspection.

1-52
Data Analysis/Evaluation

There are four basic AE event-/signal-related sets of data that are assessed:
AE energy: This is typically done on a total or aggregate basis for a time step
of 0.1 and 0.01 sec. This event feature as a primary AE parameter differs
from the EPRI Guidelines.
The pseudo frequency of an event (counts/duration).
AE (event) counts: These are separated into energy band bins and computed
for a given exposure period.
Amplitude.

The above allow identification of located AE sources and identification of source


type. According to Margan, these include:
Flaw accumulation and development: Cracking, fracture, and de-bonding of
hard inclusions and indications of possible creep development
Mechanical impact and friction: Impacts and knocks at hanger locations,
friction at supports
Leaks, malfunctioning valves, and steam fluctuations due to process control
problems

The specifics of the data screening process and the evaluation are proprietary and
present a limitation to any future efforts to standardize the process to include
proper consideration of the Margan practice.

Limitations

As indicated by Margan, limitations are:


Sensitivity decreases with increases in background noise.
Sensitivity decreases with increased fluctuations in noise.
Error in source locating is about 10% of the waveguide interspacing distance.
Source locating surface versus subsurface is not identified.
Non-active flaws are not detected.
Sometimes AE is too sensitive (small flaws with high activity).

The only indicated limitation not clearly identified in the development of the
Guidelines and any other post-Guidelines work is the last one indicated. Indeed,
the varying AE response of flaws not necessarily dependent on flaw size or extent
of damage makes any attempt to use AE monitoring as a method of identifying
extent of damage potentially futile and, at best, uncertain.

1-53
Standardization

There is evidently substantial variability in the use of AE for monitoring of fossil


plant high-energy piping. Further, there remain many details of the applied data
analysis/evaluation procedures that are unknown. While consistency in the
application is naturally challenging due to the many variables that affect the
measured output (for example, attenuation, varying emission with stress method,
location, etc., signal characteristics changing similarly, etc.), the high-energy
piping case begs for some level of standardization and greater transparency to
further its development and value.

There are many ASTM standards relating to acoustic emission; ASTM


Subcommittee E07.04 on Acoustic Emission Method oversees as many as 26
standards and practices shown in Table 1-6 produced here as a ready reference
for the interested user. However, there is no standard or practice that supports
the high-energy piping monitoring application. Following the development of
the 1995 EPRI Guidelines, Subcommittee E07.04 has been working on a New
Standard Test Method for Acoustic Emission Examination of Seam-Welded
High-Energy Steam Piping for a decade (ASTM Work Item WK658 initiated
in April 2003). Indications, however, are that the development of this standard
has stalled due to the persistent, significant differences of opinion among the
practitioners. Perhaps reflective of this condition is a relatively recent (initiated
April 2011) standard development work item, ASTM WK33018 New Practice
for Acoustic Emission Examination of Steam Piping, that appears to, at a
minimum, overlap the WK658 initiative and, at worst, be in technical conflict
with it. The apparent lack of consensus presents a potential obstacle to
streamlining use of the method for the high-energy piping application.

Again, to be clear, the variability in practice and the accompanying lack of


transparency beg for a defined set of clear guidelines or standards for the high-
energy piping monitoring application.

1-54
Table 1-6
Current ASTM standards under the jurisdiction of ASTM Subcommittee 07.04 on
Acoustic Emission Method.

ASTM Standard Title


E569/E569M-13 Standard Practice for Acoustic Emission Monitoring of
Structures During Controlled Stimulation
E650/E650M-12 Standard Guide for Mounting Piezoelectric Acoustic
Emission Sensors
E749/E749M-12 Standard Practice for Acoustic Emission Monitoring
During Continuous Welding
E750-10 Standard Practice for Characterizing Acoustic Emission
Instrumentation
E751/E751M-12 Standard Practice for Acoustic Emission Monitoring
During Resistance Spot-Welding
E976-10 Standard Guide for Determining the Reproducibility of
Acoustic Emission Sensor Response
E1067/E1067M-11 Standard Practice for Acoustic Emission Examination of
Fiberglass Reinforced Plastic Resin (FRP) Tanks/Vessels
E1106-12 Standard Test Method for Primary Calibration of
Acoustic Emission Sensors
E1118/E1118M-11 Standard Practice for Acoustic Emission Examination of
Reinforced Thermosetting Resin Pipe (RTRP)
E1139/E1139M-12 Standard Practice for Continuous Monitoring of Acoustic
Emission from Metal Pressure Boundaries
E1211/E1211M-12 Standard Practice for Leak Detection and Location Using
Surface-Mounted Acoustic Emission Sensors
E1419-09 Standard Practice for Examination of Seamless, Gas-
Filled, Pressure Vessels Using Acoustic Emission
E1495/E1495M-12 Standard Guide for Acousto-Ultrasonic Assessment of
Composites, Laminates, and Bonded Joints
E1736-10 Standard Practice for Acousto-Ultrasonic Assessment of
Filament-Wound Pressure Vessels
E1781-08 Standard Practice for Secondary Calibration of Acoustic
Emission Sensors
E1888/E1888M-12 Standard Practice for Acoustic Emission Examination of
Pressurized Containers Made of Fiberglass Reinforced
Plastic with Balsa Wood Cores
E1930/E1930M-12 Standard Practice for Examination of Liquid-Filled
Atmospheric and Low-Pressure Metal Storage Tanks
Using Acoustic Emission
E1932-12 Standard Guide for Acoustic Emission Examination of
Small Parts

1-55
Table 1-6 (continued)
Current ASTM standards under the jurisdiction of ASTM Subcommittee 07.04 on
Acoustic Emission Method.

ASTM Standard Title


E2075/E2075M-10 Standard Practice for Verifying the Consistency of AE
E2076/E2076M-10 Standard Practice for Examination of Fiberglass
Reinforced Plastic Fan Blades Using Acoustic Emission
E2191/E2191M-10 Standard Practice for Examination of Gas-Filled Filament-
Wound Composite Pressure Vessels Using Acoustic
Emission
E2374-10 Standard Guide for Acoustic Emission System
Performance Verification
E2478-11 Standard Practice for Determining Damage-Based
Design Stress for Glass Fiber Reinforced Plastic (GFRP)
Materials Using Acoustic Emission
E2598/E2598M-13 Standard Practice for Acoustic Emission Examination of
Cast Iron Yankee and Steam Heated Paper Dryers
E2661/E2661M-10 Standard Practice for Acoustic Emission Examination of
Plate-like and Flat Panel Composite Structures Used in
Aerospace Applications
E2863-12 Standard Practice for Acoustic Emission Examination of
Welded Steel Sphere Pressure Vessels Using Thermal
Pressurization

1-56
Section 2: Experience Summary
The in-plant AE monitoring experience data that have been made available to
the project were reviewed and evaluated with respect to:
Methods used, including stimulation modes, AE activity metric used, etc.
Reported AE findings such as identified cluster locations and their AE
activity measures
Correlations between the reported AE activity at located clusters and results
from other examinations, nondestructive and destructive

The experience data are presented and discussed in chronological order,


beginning with the experience leading up to the 1995 EPRI Guidelines [8], the
period 19961999 representing the early implementation of the Guidelines, and
the experience acquired from 2000 to the present. In a few cases, the plant or
owner-operator are not identified at the request of the data supplier.

Data Sources

The majority of experience data have been provided by J. Rodgers, Acoustic


Emission Consulting, Inc., (AEC) via published and unpublished documents.
These data were mostly gathered using methods generally consistent with the
1995 EPRI Guidelines. Information on three plants was provided by the
respective owner-operator utility in the form of email communications and
reports. These utility-provided cases involved monitoring in accordance with the
EPRI Guidelines. Summaries of eight case studies and tabulation for two more
for a total of 10 cases) were provided by I. Mizrahi, Margan Physical Diagnostics
Ltd. (Margan) that has reportedly been operating since 2000 (Margan-provided
data thus pertain only to the post-2000 period). These case studies represent the
current monitoring method practiced by Margan. Where listed and discussed,
the data source has been referenced.

Field Application Experience

19851990

The AE field effort of this period in the immediate aftermath of the Mohave and
Monroe reheat piping failures [1, 2] represents the early high-energy piping
monitoring work. The plant experience was documented by Babcock and Wilcox
(B&W) in its experimental work for EPRI [9]. While the monitoring methods

2-1
used then were varied and differed from the 1995 EPRI Guidelines method used
in the bulk of the experience reviewed, the experience has relevance to persistent
issues of AE monitoring data interpretation and is, therefore, summarized.

Table 2-1 lists the in-plant AE monitoring cases from EPRI report RP 1893-4
[9]. The italicized text in the table has been added based on a review of the text
of that reference. The added plant unit numbers were obtained from a list
published in EPRI report RP 1893-20 [10].

Table 2-1
In-plant AE monitoring cases in the period 19851990 [9]. Italicized text has been
added to the original table based on a review of the text from that reference.

#7
July 1985 Large active ligament crack

#5 Plug sample: Oxide Inclusions; inactive stringer


Sep 1986 (40 minutes)
(LOF at Circ welds)

Mid-1987 (rubbing)
Startup &

#4 March 1987 as minor indications. Noisy startup not useful.

#1 May 1987
March-April 1988 & econ. inlet header
Startup, shutdown & swing

Nov 1987

#7
Jan 1988, Sep 1989

Magnetically coupled waveguides and 112 sensors


Late 1988

May, Oct 1988 Misses by AE attributed to noise; indications


not characterized

, 8 minor
April 1989

3
#1

AE misses under thermal stimulation
AE in single case seen with thermal transient

Used RP1893-4 Guidelines [9] & large scale


March 1990 monitoring

2-2
The 12 cases of Table 2-1 indicate the following:
Six of these cases were focused on long seam-welded hot reheat pipe, one on
a clamshell elbow and economizer inlet header, two on main steam lines
without welds specified, one on circumferential welds of a main steam line,
one on the circumferential welds of both main and reheat steam lines, and
one on a secondary superheater outlet header.
Except for the superheater outlet header case (Wisconsin Public Service
Pulliam #7), none of the field tests provided any verification that AE
monitoring correlates with cracking. However, separate laboratory testing
and later experience support the fact that AE can detect active cracking.
The thermal transient stimulation that generated AE in case of the Pulliam
#7 header and the nature of the header ligament crack suggested that the
crack AE activity was time-independent or cyclic and not due to creep.
The Ohio Edison New Castle #5 case illustrated AE under thermal stimulus
from oxide inclusions and a stringer, with no evidence of these being active
damage sources in operation. AE, as applied there, appears to have tagged
otherwise inactive material discontinuities, and there appears to be a
possibility that inclusion-matrix interface thermal strains produced the AE.
The AE misses in the Pennsylvania Electric Keystone case were attributed to
noise, highlighting the problem with noise in this early phase of field testing.
The nature of the sources was not reported.
The Tennessee Valley Authority Kingston #7 case illustrated an interesting
phenomenon where a new replacement weld exhibited highly active AE in
online load swings and in an offline cooldown. Although the activity in the
1989 tests was lower than the tests in 1988 following installation of the new
weld, the evidence suggests that sources of significant AE activity that are
cluster localized need not be from active in-operation cracking or even a form
of damage of concern.
The AE misses in the Philadelphia Electric Cromby #1 case highlight the
potential inadequacy of a startup and offline steam injection stimulation.
Although the setup was not optimal (sensors were generally high resonant at
500 kHz and a fixed threshold was almost always employed), the monitoring
process could detect locations that correlated with other NDE indications
and, in one header case, a large active crack.
An early promising finding of the application is reflected in the fact that the
AE monitoring process did not produce an unwieldy number of calls.
Accompanying the manageable AE call rate, the system also appears to
produce few, if any, false calls as evidenced by the New York State Electric &
Gas Somerset #1 case, where there were no identified AE sources and no UT
indications.
The database did not have a case where AE was correlated with creep
damage.

2-3
19901995

This was the period during which the EPRI AE monitoring guidelines were
developed under joint EPRI-Pacific Gas & Electric (PG&E) sponsorship.
Volume 2 of the EPRI Guidelines [11] contains a summary table of the in-plant
AE monitoring cases through 1995. This table has been reproduced here as
Table A-2 in the Appendix. The table lists a total of 41 cases that include the 12
cases described for the 19851990 period (Table 2-1 and Table A-1). In effect,
this documentation indicates an additional 29 cases of AE monitoring beyond
the cases itemized (Table 2-1) for the 19851990 period.

The tabulation provides good detail on the


Plant/unit
Dates of AE monitoring
Component monitored
AE monitoring organization
Stimulation (stress) modes for monitoring
Source reference of the information

Unfortunately, the tabulation does not give any information on the AE results
and provides no data regarding other inspections, tests or examinations that
would help assess how the AE monitoring systems performed in that period.
Most of the cited references could not be retrieved and, except for the data on
Potrero #3 that comprised an integral part of the EPRI-PG&E project and has
been covered in detail in an unpublished report that has been available [11], this
particular experience dataset could not be evaluated. In any case, the following
observations are made pertinent to the 29 post-19851990 units listed in
Table A-2:
Twenty-six of the 29 cases included cooldown as a stress mode, by far the
leading stimulation method used.
Ten of the 29 cases used a startup stress mode in monitoring.
Eight of the 29 cases had online and/or load swing mode monitoring, a
significantly small fraction of the total, and trailing the number of cases using
cooldown and startup cycles. Evidently, testing during that period did not
appear to consider online/load swing monitoring a preferred mode.
Three of the 29 cases involved over-temperature, two of which also had
overpressure cycles of monitoring.
The only case of offline testing using hydro-pressurization and steam
injection was on Potrero #3 in the early exploratory phase of testing in
development of the Guidelines (see the preceding chapter of this report).

2-4
The emphasis on long seam-welded piping is apparent from the 25 cases
where seam-welded hot reheat piping was monitored.
The cases include one on a superheater outlet header, two on main steam
wye blocks and one on a main steam line.
The table lists 10 vendors conducting the AE testing, in sharp contrast to
what appears to be less than a handful of currently active service providers.

Because the data on Potrero #3 generated by PG&E has been available for review
and because the experiments were key in the development of the Guidelines, the
results are briefly discussed below.

Potrero #3

These tests were conducted in 1994 using cooldown (February 1994) and
online/load swing (AprilMay 1994) monitoring on both the East and the West
hot reheat leads of the unit. At this stage of development, the AE monitoring
system had reasonably matured, and the hardware and data processing methods
used closely matched the forthcoming 1995 Guidelines recommendations.
Location clusters were ranked using a source ranking system. A part of this has
been covered in the previous chapter, but it is repeated here.

In the case of cooldown, most of the AE information of value was obtained in


the first 300F temperature decrease. Online monitoring identified clusters in
addition to those seen in the cooldown; about 40% of the identified locations
were also identified in the cooldown. The online condition cycled between high
and low plant load/pressure, typically creating a temperature transient of about
20F30F. The results support use of online load swings as a stress method.

A source ranking system was used for the located clusters. Qualitatively, in the
cooldown, the source rank was based on the total events, total counts, and
normalized amplitude ranking (factor based on signals >70 dB). For online
monitoring data, the source ranking was based on event density (time rate), count
density (time rate), and normalized amplitude ranking as for the cooldown.

The nondestructive and destructive examinations following the AE monitoring


provided correlations with identified sources/location clusters.

2-5
Tables 2-2 and 2-3 summarize the AE sources and results of other NDEs on the
East and West leads, respectively. Note the 10 identified clusters with source
ranks from 0.10 to 0.67 that include seven inaccessible locations without UT
correlations and three true positives (UT indications). The main findings from
the reported results are:
There were no significant false calls.
In one instance of startup (temperature increase), there was a substantially
high event rate near the turbine on the East lead (30 in. location, Fig. D.1-3
of EPRI report RP 1893-20 [10]) that was rejected; the location also showed
no indications in other inspections. The example highlights the potential
issues with using the startup transient and the subjectivity and experience
that may be needed in interpretation of the AE data.
Reasonable correlations of the AE identified significant clusters were made
with UT.
AE showed good capability to detect thermal cracking.
Again, there was no case that permitted correlation of AE with creep
damage.

2-6
Table 2-2
Active AE clusters compared with NDE indications on Potrero #3 East hot reheat
lead [10].

2-7
Table 2-3
Active AE clusters compared with NDE indications on Potrero #3 West hot reheat
lead [10].

PPL Generation

Plant high-energy piping AE monitoring experience of PPL Generation (then


Pennsylvania Power & Light or PP&L) in the 19931997 period was provided
in summary written form by D. Berger, PPL Generation [20]. Although this
subsection is focused on the 19901995 period, the PPL Generation experience
through 1997 is included here for convenience and completeness.

2-8
A Grade 22 hot reheat line of Plant A operating at nominally 1000F steam
was monitored in the 19931997 period. The monitoring procedures, at least in
1995 and thereafter, were likely to have been generally consistent with those of
the 1995 Guidelines [10], since at least part of the work (in 1995) was reportedly
done in an effort to demonstrate application of the Guidelines.

Waveguides were welded onto the piping in 1993 while the unit was operating at
normal pressure and temperature. Over the period 19931997, three sets of tests
were conducted on that line, one each in 1993, 1995, and 1997. Most of the
waveguides stayed on the pipe long enough to do AE tests through 1997, with
roughly about 20% of the waveguides needing to be reattached in 1995 and 1997.
Considering each set of tests as one case, this experience brings the total case
count to 44.

1993

A cooldown AE test performed on shutdown identified two potential source


locations. Post-AE monitoring UT and magnetic particle testing (MT)
examinations were also conducted. UT identified two small, acceptable
subsurface slag inclusions in a turbine lead. MT examination produced three
types of indications: hanger attachment cracks (two hangers), cracks on the drain
pots and associated control piping attachments, and hydrogen cracks related to
alignment lugs used during original field erection (several locations that were
near two field welds and on the inlet of a boiler outlet reducer). The hydrogen
cracks were concluded to be non-propagating. The other cracks were believed to
be service-induced, with fatigue as the expected mechanism for the shallow drain
pot cracks and thermal stresses driving the hanger cracks. All cracks were
removed by grinding, without need for weld repairs. Although no details have
been provided on the correlation of the two AE source locations with the other
NDE results, the AE monitoring evidently missed identifying more than one
crack location. The specifics of the monitoring method were not reported.

1995

A cooldown AE test was performed during a shutdown. This was the first phase
of acoustic emission monitoring under an EPRI co-funded project to
demonstrate application of the Guidelines [10]. AE identified several areas for
examination. The subsequent UT examination results, including focused beam
UT, found four indications that were concluded to be from slag inclusions. These
were likely not damaging in service and would not necessarily emit detectable
AE, although service-inactive inclusions had been earlier found to be emitters in
cooldown and thermal transients (see Ohio Edison New Castle #5 case listed in
the 19851990 data description). However, the AE-identified locations
reportedly had no UT indications; that is, apparent AE false calls. Separately, a
later report by Rodgers [11] identified a PP&L unit where a South turbine lead
clamshell elbow had a located source with low NEDR, <1.5 but no indications
with a time-of-flight diffraction (TOFD) UT examination.

2-9
1997

In a period of the three weeks preceding an overhaul, AE monitoring of the


entire line was conducted. The testing program included nine days of steady-state
on-line monitoring and swings involving a thermal cycle from 975F gradually up
to 1025F, then from 1025F down to 950F in 1 hour. Four areas were
identified as AE sources requiring follow-up examination. Most all of the seam-
welded segments of the line were inspected by wet fluorescent MT and TOFD
UT. MT examination identified three areas of indications. Two were determined
by subsequent replication to be superficial liquation cracking. The third area was
at a hot reheat fabricated lateral (branch connection). The indications in this area
were located in an area not inspectable by TOFD due to its geometry. There
were a number of indications on both fusion lines of the fabricated lateral branch
weld. The longest was 1.9 in. long and wholly confined to the weld metal of the
longitudinal seam in the 28-in. OD branch (that is, across the width of the
longitudinal weld in the 28-in. branch pipe). Replication indicated heavy creep.
During excavation, the indication remained fairly constant in length until it
disappeared at a depth of approximately 1/2 in. At that point, a second indication
appeared at the upper fusion line of the longitudinal seam in the 28in. OD
branch (that is, at roughly 90 degrees to the original indication, which was
parallel to the fusion line of the branch weld). Replication of this second
indication led to two different opinions of its morphology. One was that it was
fabrication-related and accompanied by nonmetallic inclusions. The second was
that it was creep cracking, accompanied by a heavy population of creep voids.
The AE-identified locations were not specifically correlated with the liquation
cracking and branch lateral creep damaged locations discovered with MT.
However, separately, Rodgers has reported [17] on a PP&L hot reheat unit
branch lateral having been AE-source-located, but with a low normalized event
density ranking (a NEDR <1.0) that is very likely the one cited here. What is
nevertheless noteworthy is that an area of significant creep damage (the branch
weld location had to be repaired) was located, but did not apparently produce
much AE activity in that case. This is considered one of the cases supporting
creep damage detection with AE.

The following observations are made:


The 1993 AE monitoring methods may not have been adequately consistent
with the 1995 Guidelines, but it is noted that the monitoring during a
cooldown missed more than one location with fatigue and thermal cracking.
The 1995 AE monitoring during a cooldown, likely to have been closely
consistent in procedure with the 1995 Guidelines, had source locations that
did not evidence any damage in UT examination; apparent AE false calls
were in effect.
The PPL Generation-reported information from online load (temperature)
swing AE monitoring in 1997 suggests that AE missed a significant creep-
damaged area. It is inferred from other data [17], however, that this case is
likely one that was, in fact, source-located, but with the reported NEDR for
that location being <1.0, a low-risk output that carries with it no

2-10
recommended immediate action (see the discussion on Current Practice in
the previous chapter). In contrast, as it relates to low NEDR-located sources,
there appears to be an instance in 1995, based on Rodgers report [11], where
a clamshell elbow located with a low NEDR, <1.5, did not show any
indications with a TOFD UT. In effect, the results suggest that a simple AE
activity measure need not reflect the extent of in-service damage and the risk
of future damage progression. This is an important issue with regard to the
best use of AE monitoring.
Although there were apparent false calls and perhaps one significant weak or
near-miss call, the monitoring did not produce an unwieldy number of AE
calls for disposition.
Based on other data [17], this experience likely includes one example of an
AE-detected creep cracked area, albeit with weak AE activity.

19961999

This period represents a short, but apparently productive time during which the
1995 EPRI Guidelines were exercised. The reviewed information for this period
has been largely produced and provided by Rodgers [11, 16, 17]. As such, the
findings and inferences made here need not reflect wider application of high-
energy piping AE monitoring technology, although no substantive information
or data have been found to support a wider application.

Table A-3 is a tabulated listing of the plant AE monitoring cases, reproduced


from EPRI report TR-111558 [11]. The list of 30 cases includes two cases
previously covered (the first two listed), that is, 28 additional cases, bringing the
running total through to 1999 equal to 72. At least 17 of the 28 listed cases were
reportedly from tailored collaboration EPRI projects. Findings from the table
and from the report [11] are as follows.

Stress Method
Twelve of the 28 cases included cooldown and startup as a stress mode. In
contrast to the pre-1996 work, the cooldown stress method was replaced by
the online mode conducted in every one of the 28 cases. Evidently, the
monitoring process evolved into one with a preferred online stimulus mode
(less than a third of the cases prior to 1996 used the online mode), while the
cooldown became a secondary stress method.
The online mode was typically in accordance with the Guidelinespeak
loads with load swings of 25% for a minimum of three cycles.
The cooldown cycle was typically a full cooldown of the piping from
operational temperatures to 600F.

2-11
The online mode (steady-state and load swings) was considered more
significant for evaluation of seam welds. The cooldown was thought to be
better for evaluating welded hangers, circumferential welds, drains, and vents.
Startups were concluded to be impractical because of unpredictable
schedules, high noise, and the absence of near-term follow-up inspection
opportunities.
There were no documented offline tests.

Components
The emphasis on long seam-welded piping is apparent. All of the cases
involved long seam-welded Grade 11 or 22 piping26 hot reheat including
clamshell elbows and two main steam link.
The piping length monitored per case varied from as low as 23 ft to >1000 ft.

AE Data Analysis Refinement


This phase of the work resulted in development of a region- or length-
normalized measure of located cluster AE activity termed normalized event
density ranking or NEDR (see Figure 1-13 for a graphic definition and the
text of the preceding chapter for a description).
The NEDR was reportedly developed to account for significant variation in
located emitting cluster sizes and in amplitude. This approach appears to be
logically intended to categorize as a low-activity (low NEDR) cluster, a
relatively large area of low-level emission density that produces a total high
emission rate. At the other extreme, the approach is meant to categorize as a
cluster of high activity (high NEDR), a discrete small area of intense
emission (high emission density as may occur from crack advance) with a
total emission that is relatively low. In theory, NEDR evaluations could well
correlate with the extent of damage.
Startups produced extremely high NEDR values, in some cases 50 to 100
times higher than in online cycles. Startup NEDR values were not used by
themselves for cluster identification.

Results and Correlation with Other Examinations

Table 2-4 lists the cases as correlated with the results of other examinations. The
details of the AE monitoring stimulation method and dates of monitoring are
found in Table A-3. The table has essentially been reproduced from the tables of
EPRI report TR-111558 [11], with the added seam weld AE activity (NEDR)
data column imported from Table A-3 for a convenient read.

2-12
The summary observations from Table 2-4, excluding the first two listings
previously discussed, are:
Of the 28 cases, five had no significant AE findings. These five cases also
had no other inspection data associated with them and do not provide useful
information regarding possible missed AE calls.
One of the five no significant findings or NSF casesCPL Nueces Bay
#6lists a seam-weld NEDR at an exit elbow of <1.2, and it is not clear why
this located cluster, albeit with a low NEDR, was considered an NSF.
Nine cases with reported significant AE findings had no other examination
data (eight with no inspections and one with pending inspection results) and
are, therefore, not useful for any correlations.
One case (UI New Haven Harbor) with significant AE findings had
incomplete inspections, although the reported rigid support AE cluster
location appears to have been UT inspected (UT reported at rigid hanger,
assumed to be the same location), and no UT indications were reported. This
may be considered a false call case. NEDR values were not reported for the
significant AE findings that were all located at hangers (non-seam-weld
areas).
This leaves a total of 13 cases with other examination data that are discussed
here in more detail.

2-13
Table 2-4
Correlation between AE activity and results of other examinations, as assembled from the EPRI report TR-111558 [11] table
with the NEDR data added.

2-14
Table 2-4 (continued)
Correlation between AE activity and results of other examinations, as assembled from the EPRI report TR-111558 [11] table
with the NEDR data added.

SH
KU
HRH

2-15
Table 2-4 (continued)
Correlation between AE activity and results of other examinations, as assembled from the EPRI report TR-111558 [11] table
with the NEDR data added.

2-16
Several important observations were made with respect to the 13 cases for which
other examination data are available:
Twelve of the 13 cases had NDEs, typically UT and MT. The report
indicates good correlation of the AE-located clusters with UT and MT
results. In the one case where there are no reported NDEs that were
conducted (KU Brown #3 superheater [SH] links), link section seam-
weldment samples were removed, and the results from a destructive
examination on one of the samples is discussed below. In any event, the
sample removal suggests that there was a conclusion of significant indications
of damage in service that could well have been a result of UT examinations
not reported.
There was one case (KU Brown #3 hot reheat [HRH] links) where there
were no significant AE findings (note that the seam weld NEDR Values
entries in Table A-3 for the Brown #3 SH links and the HRH links have
evidently been reversed and reported in error based on the details of EPRI
report TR-111558 [11], and this has been corrected in Table 2-4). In this
case, there were also no significant indications by other NDEs, that is, good
correlation and no false calls.
In two instances (Sierra Pacific Valmy #2 and KU Brown #3 SH Links),
repeat AE monitoring was conducted. For Valmy #2, this was in 1998 and
1999, and for Brown #3 SH links, the repeat AE monitoring was conducted
in 1998 before an outage. Results were repeatable, and the cases support
repeatability.
There were four cases (GRDA Coal #2 HRH, Sierra Pacific Valmy #2, CPL
Lon C. Hill #4 HRH and GRDA Coal #1 HRH) where the clusters located
with transient stimulation modesstartup and/or cooldowndid not
respond to the online pressure changes. These all appeared to be cases of
thermal cracking (hanger related and one case of near-seam weld
laminations), supporting the likelihood that thermal or time-independent
forms of cracking need not be activated and detected by online pressure
change (load swing) monitoring, rather, thermal cracking can be best located
by AE under thermal stimulus modes (startup, cooldown, and temperature
swing transients). The Valmy #2 seam weld lamination indicated area was
plug-sampled, and the cracking, seen in Figure 2-1 and reproduced from
EPRI report TR-111558 [11], was found to be not related to creep, having a
thermal fatigue appearance.
AE was able to locate an advanced visually identified drain line fillet weld
crack in one case (WTU Oklaunion HRH) using only online stimulation.
The experience suggests that online testing could be effective in identifying
time-independent or cyclic/thermal forms of cracking in cases where the
cracking is so advanced as to be active under online stress conditions.

2-17
Figure 2-1
Plug sample evidence from Valmy #2 of non-creep thermal fatigue-appearing
crack from a seam weld lamination that was AE-located in cooldown, but not in
online monitoring.

2-18
Specifically, with regard to seam weldment creep damage detectability with
AE, the following observations were made for each one of the eight cases for
which there were seam weld NEDR values reported; that is, for which there
were AE-located seam weldment clusters (see Table 2-4):
- AEP Gavin #1 HRH: While there was good correlation between the
AE-located cluster in online tests over about a 30-in. length on one
turbine lead (low-activity NEDR ~1.5) and manual UT showing
intermittent nearinside diameter (ID) indications, the nature of the
indications was not reported, so the case provides no information
regarding creep. Interestingly, a second lead long seam weldment that
was not AE active was found to have stringers by TOFD, supporting the
potential value of AE in focusing on active defects and enhancing the
efficiency of follow-on NDEs.
- CPL, ES Joslin HRH: A highly active AE cluster (NEDR ~40) was
located in the vicinity of a hanger support on the lower horizontal section
of line during online testing, related to peak pressure loading. The two
most active sites were in the seam weld under the hanger strap. Several
other seam weld areas with activity in the range of NEDR 712 were
also located on the line. A subsequent monitoring of the line during
cooldown conditions produced very little emission in these areas,
supporting the possibility that the seam weld clusters were from active
creep-related sources and not thermal or time-independent damage.
These suspect locations were later inspected by UT and confirmed, both
with manual as well automated methods and TOFD. A 10-in. square
plate section was removed from the hanger strap location to further
evaluate the metallurgical condition of the seam weld. A vendor using
standard optical metallographic practice evaluated one portion containing
the strongest UT indication. Thin slicing every 0.020 in. over a 1-in.
section was used to successively examine the weld cross-section. This
evaluation resulted in no detection of cracking or visible cavitation up to
500X magnification. The other section containing the lesser of the two
indications was retained by EPRI and later evaluated by M&M
Engineering Associates, Inc. (M&M) by cryo-cracking. The fracture
found evidence of early stage creep cavitation in the lower weld bead, and
this example has been cited as one where the AE monitoring process is
detecting a very early stage of creep damage [11, 16]. The AE evidence
and fractograph of the cryo-fracture is shown in Figure 2-2. Issues
remaining to be resolved are the following:
o The absence of creep cavitation or cracking in the high NEDR (~40)
section examined metallographically suggests a weak correlation (if
any) between NEDR and the extent of damage. Additionally, the
high AE NEDR call represents one of minimal concern for failure
and may be considered a false call. Finally, the background on the
UT data with regard to assessment of the indications merits review.

2-19
o The relevance of calling out significant AE in a case where the
damage, seen only via an unconventional cryo-cracking examination
technique, is at such a low level as to provide no useful indicator to
assist in integrity/life management.
o Potential sources of emission other than those reported need to be
examined.

Figure 2-2
Graphics reproduced from EPRI report TR-111558 [11] in the case of CPL, ES
Joslin HRH AE monitoring AE response with the peak at a hanger support (upper)
and results of cryo-cracking reportedly representing the early stage of creep
cavitation (lower).

2-20
- KU Ghent #1 HRH link: The AE-located clusters at four bends
(initially assumed elbows were later found to be bends) and in a
horizontal straight segment (near a bend) all showed UT indications that
matched well in severity also. The highest NEDR location at an East
lead bend had very high startup NEDR values (8085) and online values
of 68 at normal operation, reportedly sensitive to pressure changes. As
reported, the AE activity in terms of NEDR decreased rapidly with a
pressure decrease, going from about 20 NEDR at 510 psig to about 1
NEDR at 415 psig. Figure 2-3 from EPRI report TR-111558 [11]
shows the apparent power-law pressure dependence of AE activity, as
may be expected in case of creep damage progression. The pressure
dependence highlights the potential sensitivity of AE activity to load (in
this case, about a 95% decrease in NEDR for about a 20% decrease in
pressure). A 10-in. sample from the most active AE bend was removed
for laboratory metallographic examination, and all bends were repaired.
It is noteworthy that the post-repair AE activity (see Figure 2-3) was
comparatively low (~2 at 510 psig), but identifiable. The 10-in. sample
did not show any evidence of creep cavitation or cracking via optical
metallography examinations to 500X. As with CPL ES Joslin, the AE
findings, although consistent with the UT indications, raise these issues:
o The absence of creep cavitation or cracking in a case where there was
significant NEDR (68 normal and about 20 at peak pressure)
represents one of minimal concern for failure and may be considered
a false call. The background on the UT data with regard to
assessment of the indications merits review.
o The practical relevance of calling out significant AE is questioned in
a case where the damage is at such a low level as to provide no useful
indicator to assist in integrity/life management.
o Potential sources of emission other than creep damage in the form of
conventionally identifiable cavitation or cracking need to be
examined. The post-repair AE activity, not insignificant, suggests
sources of emission possibly unrelated to easily identifiable forms of
creep damage.
KU Brown #3 SH link: The AE monitoring was on a 10-ft segment, each on
an East and West outlet at the penthouse roof. Both segments showed
comparable AE activity with a relatively high NEDR (~35) in online
monitoring. No in-service UT data were reported. However, both link
segments were removed, and a 6-ft section of the West segment was sent to
EPRI for evaluation. EPRI conducted two UT inspections of the seam weld
on the section. The first was a standard multi-angle pulse-echo examination
according to 1996 EPRI Guidelines criterion 3 [4]. This inspection did not
find any significant indications. The second was an automated UT inspection
using a phased array capability that had been added to EPRIs Tomoscan
system. The phased array inspection found evidence of indications in the
weld centerline over a 24-in. area matching the AE source location cluster.
A 3-in. core sample was taken from the suspect weld area and examined via
optical metallography at 500X. No evidence of cracking or cavitation was

2-21
visible. The sample was then sent to M&M Engineering Associates, Inc., for
further evaluation by cryo-cracking. Early stage creep cavitation of the type
seen in CPL ES Joslin was seen at the weld centerline, which correlated with
the results of the phased array UT examination. Again, the finding raises
these issues:
- The absence of creep cavitation or cracking in the high NEDR (~40)
section that was examined metallographically suggests a weak correlation
(if any) between NEDR and the extent of damage. Additionally, the
high AE NEDR call represents one of minimal concern for failure and
may be considered a false call. Finally, the background on the UT data
with regard to assessment of the indications merits review.
- The practical relevance of calling out significant AE in a case where the
damage, seen only via an unconventional cryo-cracking examination
technique, is at such a low level as to provide no useful indicator of
lifetime.
- Potential sources of emission other than reported need to be examined.

Figure 2-3
AE activity (NEDR) as a function of pressure in online monitoring, showing
sensitivity to operating pressure [11].

2-22
- CPL Lon C Hill #4 HRH: The AE monitoring in this case (startup,
cooldown, and online) showed detectable clusters, but these were
insensitive to pressure. The two identified locations (NEDR 620,
unknown stress method) were at a seam weld near the turbine deck and
at a batwing hanger correlated with UT and MT resultsthe UT
showing laminar indications in the seam weld and the MT showing OD
cracking at the batwing hanger. The AE sources were apparently non-
creep-related and their insensitivity to pressure supports the value of
online load swing testing to potentially screen creep from non-creep
forms of damage.
- OG&E Horseshoe Lake #7 HRH: The AE monitoring in this case
(startup, cooldown, and online) showed detectable clusters at several
elbows (NEDR 41 at one and 1222 at others) and at a batwing hanger
in online and startup (NEDR 18). The hanger location correlated with
MT findings of cracking. The elbow seam weld locations correlated with
midwall UT indications up to 1 in. long. The case illustrates good
correlation with other examinations and detectability of non-creep-
related cracking (at batwing hanger). No information is available on the
seam weld indications, so no inferences may be made regarding the
detectability of creep damage.
- Salt River Project Navajo #2 HRH: Two clamshell elbows of a hot
reheat line produce AE with the maximum NEDR reported for a startup
on one elbow at 56 and an online NEDR for the other at 8,1 at 1000F.
Interestingly, the online NEDR for both elbows decreased to <0.5 after
two weeks at a lower temperature of 950F. As in the case of pressure,
the online AE activity appeared sensitive to temperature, suggestive of
the source being related to creep. The locations reportedly correlated
with UT (TOFD and focused array) indications of creep damage,
although no details are available.
- Illinois Power Baldwin #1 HRH: Six locations at bends were AE
identified, with the maximum AE activity recorded at a bend near a
hanger at 5 NEDR. A clamshell elbow was also AE-located in online
monitoring with a shakedown to 2 NEDR. Other NDE results are not
reported for the bends; however, the clamshell elbow was subject to
detailed UT (TOFD and focused array), with the focused array
indicating early stage creep damage in the near-midwall of the extrados
weld and the TOFD unable to identify any indications of creep damage.

The eight seam-weld NEDR cases for which other examination data were
available indicate the following:
There was generally good correlation of AE-located clusters with UT and
MT indication locations.
The AE activity sensitivity to pressure and temperature in online monitoring
suggests a means of identifying creep damage-related sources.
Non-creep damage sources (for example, hanger-related, weld laminations)
were found to be pressure-insensitive, providing a means (online load swing)
of identifying such damage sources.

2-23
The three cases with significant AE activity in online monitoring that
showed no evidence of creep damage (cracking or optically observable
cavitation) present some pending issues requiring further research into the
sources of the AE activity.
Even considering the above three cases as false calls, the AE monitoring
results as reported have very few false calls and also did not evidently produce
an unwieldy number of calls to manage. As a tool to help focus on areas to
inspect with well established nondestructive volumetric and surface methods
(UT and MT), the findings are positive.

2000Current

The service experience data for this period have been provided mainly by Rodgers
[17] (six cases) and Mizrahi [18, 19, 21] (nine cases). Supplemental details on
one of the cases of Rodgers were provided by Griffin, NB Power Generation
Corp. [22]. Information on two additional cases was provided, one each by
Crichton, American Electric Power [23, 24] and Fox, GenOn Energy, Inc. (now
NRG Energy, Inc.) [25]. These 17 cases bring the running total of cases to 89.

Acoustic Emission Consulting, Inc. (AEC) Data [17]

Table 2-5 summarizes the data prepared from tabulated data provided by
Rodgers [17]. References to vendors of the examinations have been removed.
The AE monitoring dates for these cases were not provided, but most of the
cases other than the first three listed are likely from the post-1999 period. The
first three cases in the table above the separator horizontal line are updates of
ones previously discussed for the 19961999 period. The significant updates are:
The Horseshoe Lake at batwing hanger cracking phenomenon was further
characterized as crack progression from the support into the pipe wall. It was
also reported [17] that the AE from this location was pressure-sensitive and
that the cracking into the pipe wall exhibited creep damage.
The Navajo clamshell elbow case reportedly had a material sample removed
and examined conventionally, exhibiting creep cavitation. This confirmed the
UT-inferred condition of creep damage discussed earlier. Recall that the AE
activity was highly sensitive to temperature in this case.

2-24
Of the remaining six cases, the last one listed NB Power Corp.s Belledune #2
did not involve weldment creep damage or cracking and is discussed later. Of the
five cases left, the experience showed varying levels of damage correlating with
AE.
The online load swing stress method was employed in all cases.
Located clusters correlated with nondestructive or destructive examination
results as follows:
- The AEP Conesville #5 main steam seamless segment exhibited an
active cluster (NEDR ~18) at a vent connection that failed nine months
later. As has been seen with much of the experience, located clusters
generally correlate well with time-independent or cyclic/thermally driven
forms of cracking.

2-25
Table 2-5
Tabulated listing of in-plant AE monitoring cases prepared from a table provided by Rodgers [17].

2-26
- One case of Grade 22 seam-welded hot reheat piping bends had AE-
located clusters (Great River Energy Stanton #1) that correlated with
cavitation damage observed in samples removed from the bends. The
NEDR was <5 and there were no NDE data. However, the highest
damage predicted by an OmegaPipe analysis coincided with the most
active AE-located clusters. Although the samples were taken before the
AE, there is every reason to surmise that the AE-identified locations
would also have been creep cavitated.
- In two cases (SWPSCo/Xcel Harrington #2 and PGE Boardman) of
Grade 22 hot reheat pipe bends and clamshell elbows, respectively, the
NEDR values were significant (<114 at 80% peak load and >5,
respectively) . In the case of Harrington #2, multiple UT techniques were
unable to identify indications, although two sections were removed for
further examination. In the Boardman case, midwall indications were
UT- identified and repaired, followed by removal of a plug sample away
from the repair. In both cases, sections removed were later found to have
isolated cavitation identified by cryo-cracking. The AE activity may have
correlated with more advanced creep damage in the Boardman case since
the UT-indicated region was not destructively examined. However, the
Harrington #2 and Boardman cases are the fourth and fifth examples
(three seen in the 19961999 experience) of where the AE activity was
significant, but creep damage was minimal and not identifiable via
conventional metallographic means.
- In the Exelon Power Schuykill #2 case, the moderate AE activity
correlated with indications of creep cavitation damage made by focused
array UT. Additionally, the AE activity responded to pressure changes,
supporting the correlation.

NB Power Belledune #2: This is an interesting case where failure occurred in a


seamless Grade 22 hot reheat pipe bend after about 113 khr of service. The
failure, manifested as an axial creep crack at the extrados, was attributed [26, 27]
to an excessively large grain size (~1000 m) likely from original forming and
heat treatment, resulting in poor creep ductility. Additionally, increased extrados
pressure hoop stress from the observed ovality was also concluded to be a likely
contributor. NB Power opted to use AE to monitor other bends. The AE
monitoring occurred in four phases (two in 2008 and one each in 2010 and
2012). Details of the AE monitoring have been provided by Griffin [22]. The
relevant results summarized for one of the hot reheat bends (#19) listed in Table
2-5 that has been repeatedly tested are:
The located AE clusters were in the vicinity of the bend extrados where in-
service strains from ovality and internal pressure are expected to be at a
maximum.
The UT examinations roughly about the time of the first phase of testing in
2008 showed no evidence of creep damage. However, the AE monitoring
located clusters, albeit with low NEDR (<1). The AE was reported to be
indicative of early stage creep damage. However, the reports provide no
physical or other evidence to support the conclusion.

2-27
Repeat monitoring through 2012 did not show a significant increase in AE
for the bend.
The AE rate was pressure sensitive, supporting the case for the source being
related to the mechanical stress effect.
The AE source remains to be identified. It is possible that early creep
damage is responsible for the emission activity as reported. However, the
consistency in AE for several bends and for the extrados location and the
apparent response to pressure suggest that localized strain, rather than
damage as traditionally defined (creep cavitation and cracking), may be an
AE source.

Margan Physical Diagnostics Ltd. (Margan) Data [18, 19, 21]

Table 2-6 is a tabulation of the experience data provided by Mizrahi [18, 19, 21].
The plant identifications have not been provided. The set of nine cases includes
five hot reheat (HRH), two main steam (MS) and two cold reheat (CRH) piping
systems. The HRH cases include one of Grade 91 and the remainder of Grade
22, with all units having long seam-welded piping. For reference, the previous
chapter gives a summary description of the Margan AE monitoring method and
how it differs from the 1995 EPRI Guidelines recommendations. While Margan
has provided a qualitative description of its data interpretation (using event
energies aggreggated over small increments of time) with regard to AE activity,
specific cluster locating criteria are proprietary and all of the data have been
reported in relative units (r.u.). Thus, no cross-comparison can be made with
other data such as that of AEC (using NEDR), nor can an assessment be made
with regard to potentially correlating acttivity levels with the nature of the source.
Finally, the information on other examinations appears limited to areas where
significant activity has been reported, so no assessment can be made regarding
missed and false AE calls.

Following are the relevant observations made from the experience data provided:
Three cases of apparent fatigue-driven cracking from thermal effects were
detected with AE monitoring under steady-state and low-load conditions
the two MS cases and one CRH case (X-YY#6) of girth weldment cracking.
The MS cases may also represent damage by creep-fatigue. Interestingly,
steady-state and low-load conditions appeared to have provided sufficient
stimulation to activate sources that are apparently characteristic of fatigue.
The second CRH case (X-YZ#4) of a high-activity location (just upstream
of an attemperator) that subsequently failed likely due to thermal fatigue
involved monitoring under steady-state and load changing (swing) conditions
(<20% load change).

2-28
The five seam-welded HRH cases appear to show the following:
- The single Grade 91 case had 10 AE active (low and medium
activity) locations at girth welds. MT and UT inspections found
indications corresponding to these locations with the larger MT
indications correlating with the higher medium activity AE locations.
To date, none of the indications have been concluded to be in-service
creep related. The monitoring modes included steady-state, low-load,
and load changes.
- In one Grade 22 HRH case (XX-YY#7), the reported locations of high
AE activity in the most recent trend monitoring inspection (TMI) under
steady-state and low-load conditions showed creep microcracking and
microvoids in replications, the one example of creep damage
identification with AE.

2-29
Table 2-6
Listing of experience as tabulated from data provided by Mizrahi [18, 19, 21].

2-30
Table 2-6 (continued)
Listing of experience as tabulated from data provided by Mizrahi [18, 19, 21].

2-31
- Pipe movementrelated impacts at hanger locations were shown to be
picked up with AE monitoring in one HRH case (XXX-YYY#2) under
steady-state and low-load conditions.
- AE activity was reported in two HRH cases (XX-YY#6, X-YY#6) to
correlate with non-propagating defects. In one case, the activity was
classified as low. The observations raise a question regarding the extent
of AE calls for any given monitoring cycle and unit. Other experience
reviewed indicated that AE monitoring as practiced according to the
1995 EPRI Guidelines does not produce an unwieldy number of calls to
manage.
The Margan-provided data show that AE monitoring can correlate with
creep damage, thermal fatigue damage, impacts (for example, pipe-to-
hanger), and even non-propagating defects. The data, however, are
insufficient to assess the extent of AE calls needing to be managed in any
given monitoring cycle, the rate of false calls, and the rate of missed calls.

Utility Data

Plant experience data were offered by Crichton, American Electric Power [23,
24] and Fox, GenOn Energy, Inc. (now NRG Energy, Inc.) [25].

AEP

This is the case of an HRH Grade 22 clamshell elbow that suffered a leak. The
results of AE monitoring conducted on the piping system (including the elbow of
interest) about 40 months before the leak were also available for review. An
obvious issue then relates to the effectiveness of AE monitoring in providing
advance warning of such failures. Also reviewed for this summary is a laboratory
metallurgical examination report of a portion of the fracture.

The laboratory examination report indicated the fracture to have been initiated
and propagated by fatigue. Plant observations relating to a malfunctioning
support appeared consistent with the location of the damage. The metallurgical
examination was of a portion of the fracture in the elbow seam. The fracture
surface appeared and was reported to be consistent with fatigue and propagating
from the OD inward. Select metallographic sections through the fracture surface
showed the crack path to be in weld metal with creep cavitation and microcrack
damage adjacent to the path. The creep damage appeared to be localized to the
near-crack path region, and the report concluded that creep assisted the fatigue
mechanism of cracking.

Separately for this review and report, the fracture surface oxide thickness was
used to approximately date the crack. A preliminary estimate is that the near-
mid-crack length area had been steam exposed (connected to the inside surface)
for about 13,000 hours or 18 months. The remainder of the crack, however, was
not dated, and the approximate time of initiation and period of growth remains
unknown.

2-32
The AE monitoring, about 40 months prior, was conducted under online
conditions with load changes, and the specific failure location was not then
identified as an active significant AE source. Since the age of the crack is
unknown, no firm conclusions can be made regarding the condition of the piping
at that location at the time of monitoring and whether the experience represents
a missed call. What is apparent from this and other experience is that AE
monitoring, while a demonstrated useful tool in helping manage integrity by
optimizing follow-on nondestructive evaluations, is not a foolproof approach to
preventing failures. There are a number of possible reasons for this relating to the
variability in detected AE with stimulus, location of the source, frequency of the
monitoring program, approach to screening noise, data processing method, etc.

GenOn

This is the case of a C-1/2Mo fabricated main steam tee that was observed to
have suffered cracking visible at the OD. The crack was a result of graphitization
in the weldment HAZ and base metal and had a depth in excess of two-thirds of
the wall thickness. The pipe had been AE monitored multiple times about two
years prior to the discovery. An obvious issue then is the effectiveness of AE in
detecting this form of cracking.

A laboratory examination report concluded that the crack had existed well before
the last AE monitoring period (based on oxide dating). However, it was also
found that the crack tip was blunted and oxide filled, leading to the reported
conclusion that the crack had been dormant for some time. Crack-tip and near-
crack-tip oxide thickness appeared comparable to oxide thickness distantly
removed from the tip, supporting the possibility that the crack had been dormant
for some time. The report on the AE monitoring conducted about 24 months
prior to the cracking discovery has not been reviewed. The crack, almost certainly
present at the time of the last AE monitoring, was nevertheless likely to have
been dormant. This case highlights the fact that AE monitoring differs from
traditional NDE methods such as UT, MT, etc., in that it has the potential to
detect only damage that produces acoustic emission (that is, an active source)
under the imposed stress condition. Naturally, a source that is inactive during AE
monitoring may not remain inactive, which is another reason why AE
monitoring cannot be a foolproof method to avoid failure.

Key Observations

To summarize the key observations that can be made from the body of
experience described above, it is perhaps best to begin with a global view on the
breakdown of the various cases by the numbers. It should be recognized that
since the majority of data have been provided by service providers, at least in the
post-1995 period (post-EPRI Guidelines) of primary interest, this summary
presentation may not adequately reflect the experience. The numbers represent
the cases for which data were provided. Each case typically comprises one or
more AE monitoring cycles on a given plant and unit.

2-33
Figure 2-4 is a pie chart showing the breakdown of the number of monitored
components in the reviewed database as identified by the individual cases for
which data were providedseam-welded (S-W) piping versus non-seam-welded
piping for (a) the entire database (89 cases) and (b) for the post-1995 period (46
cases). The charts reflect the emphasis on seam-welded piping.

(b)
(a)
Figure 2-4
Pie charts showing the breakdown in the reviewed database of experience by
components (cases), seam-welded (S-W) piping versus non-seam-welded piping,
for (a) all data and (b) the post-1995 period.

Figure 2-5 shows the breakdown in materials. The vast majority of experience
has been on the low-alloy steelsGrades 11 (15%) and 22 (76%)with one case
of Grade 91 and three cases tagged as Other representing one C-1/2 Mo steel,
one SA-106 carbon steel, and the last unknown.

Figure 2-5
Breakdown of materials for the 46 post-1995 AE monitoring cases reviewed.

2-34
Figure 2-6 shows the breakdown of the stress method used in the 46 post-1995
documented AE monitoring cases. Note that the online mode (with load swings)
was used in every case, whereas offline testing was not used at all. Multiple stress
methods were used in some cases, so the stress methods count exceeds the
number of cases.

Figure 2-6
Breakdown of the stress method used in the 46 post-1995 documented AE
monitoring cases.

Note that the experience database includes nine (out of 46) cases from Margan,
using procedures that differ in certain respects from those recommended by the
1995 EPRI Guidelines. This is roughly 20% of the post-1995 experience.

The next few graphics present a global, numeric view of what the AE monitoring
cases tell us with respect to the detection of damage. No attempt has been made
to break down the individual AE calls for a given case. Rather, the presentation
simply breaks down the information on a case-by-case basis.

Figure 2-7 is a simple breakdown of the cases where there were AE activity-
located sources (39) and where there were not any such sources (7). The cases
where there were no reported sources (all EPRI Guidelines-typical procedures)
are telling because they indicate:
AE, as practiced, is unlikely to produce an excessive number of calls.
The occurrence of false calls with this practice is not expected to be large.

The next question is: Do the non-AE source cases represent missed calls? This is
a natural question, also because the reported AE monitoring cases do not involve
many calls even in cases where there is significant recorded activity. As far as
correlations with other examinations, the seven non-AE source cases correspond
to the following:
Two cases where the NDEs also found no indications. This further supports
the low false call probability.
Three cases where there were no NDEs performed.

2-35
Two cases (AEP and GenOn) that have been previously discussed; these
cases did not appear to have NDE data for a suitable correlation. The
reviewed evidence in these cases is not conclusive with regard to whether the
AE monitoring did, indeed, miss damage that was present and active.
As noted, the data have been predominantly from the service providers and
need not adequately reflect the experience.

Post-1995 AE Calls

AE No Call
Cases, 7

AE Call Cases,
39

Figure 2-7
Breakdown of the AE monitoring cases with specific located AE sources (AE Call
Cases) and ones with no identified AE sources (AE No Call Cases).

With regard to the AE monitoring results correlating with other examinations, as


a first and relevant step, the cases where AE monitoring results correlate with
results of nondestructive examinations is presented. Figure 2-8 is a breakdown of
all of the post-1995 monitoring cases as correlated with instances where
nondestructive examinations were conducted. The 28 cases with AE calls
correlated with the results of nondestructive examinations. As noted earlier, there
were two cases where the AE was silent but where the nondestructive
examinations also had no indications, a good correlation. Of the remaining 16
cases, 15 had no nondestructive examinations associated with them. One AE-
active caseSWPSCo Harrington #2 HRH in Table 2-5did not show any
indications in multiple UT examinations. A sample from the location is reported
to have later been found to have isolated cavitation seen by cryo-cracking. This
may be considered a false call and is discussed later in the context of the value of
AE monitoring.

2-36
Figure 2-8
Breakdown of the post-1995 AE monitoring cases as correlated with
nondestructive examination results. The 15 No NDE Cases are ones where there
were no nondestructive examinations conducted.

Finally, the general perspective on detectability of damagecreep versus other


forms of cracking (thermal fatigue, for example)is presented. The cases where
damage in the form of creep voids, creep cracking, and other forms of cracking
not specifically known to be due to creep have been separated into cases of seam-
weld creep damage and cracking versus other forms of cracking. This breakdown
is shown in Figure 2-9.

Figure 2-9
Breakdown of cases where damage or cracking was identified. The AE-Cracking
cases represent ones where damage was not specifically identified as creep.

The five cases of reported seam-weld creep damage include PPL (1997), OG&E
Horseshoe Lake #7, Salt River Project Navajo #2, Great River Energy Stanton
#1, and XX-YY#7 of Table 2-6. The PPL and Horseshoe Lake cases were AE-
creep cracking that correlated well after the initial monitoring. The Navajo and

2-37
Stanton instances were of cavitation (not cracking). The XX-YY#7 case is a very
recent reported instance of creep cavitation and microcracking. The details of the
AE monitoring and the damage discovery in all of these cases have not been
reviewed, so no qualifying comments can be made regarding the robustness of
these correlations and their implications for the broader issue of the efficacy of
AE monitoring in preventing long seam-welded high-energy piping creep failure.

There were four other cases where the AE-located source was reportedly
correlated with creep damage, but in the form of a very early stage of creep
cavitation. The form of isolated cavitation, concluded to be the source for the
recorded emission, was not observable in conventional optical metallography to a
magnification of 500X (see Table 2-4, CPL ES Joslin, KU Brown #3 SH link;
and Table 2-5, SWPSCo Harrington #2 and PGE Boardman HRH). A fifth
case, KU Ghent #1 HRH link (Table 2-4), also exhibited no creep damage in
conventional metallographic examination, and results of a cryo-cracking type
investigations were reportedly pending. These five cases present an interesting
issue with regard to the application of AE monitoring practice. Indeed, in some
of these cases, the AE activity, reported as NEDR, was significantly high (for
example, CPL ES Joslin at ~40 NEDR), and it is not clear how this form of
isolated cavitation can be so AE-active. The specifics meriting further
investigation are the following:
The relevance of calling out significant AE in a case where the damage, seen
only via an unconventional cryo-cracking examination technique, is at such a
low level as to provide no useful indicator to assist in integrity/lifetime
management.
Confirmation of the detectability of this stage of cavitation through
controlled laboratory experiments. The ability of AE to identify cracking has
been well accepted, and indeed, the EPRI experiments and Guidelines
development have always sought to use AE to detect cracking.
Potential sources of emission other than those reported need to be examined.

On the positive side, such calls appear to be infrequent, and clearly, the reported
experience data have not shown there to be an excessive number of AE calls to
deal with, nor have there been many false calls. Recognizing the broader
potential value of AE monitoring as a tool to help focus on areas to inspect with
well-established nondestructive volumetric and surface methods (UT and MT), a
possible correlation between AE and this form of isolated cavitation is secondary.
Of course, a better understanding of the sources of emission in cases where there
is no apparent damage level of concern will go a long way toward enhancing the
robustness of the monitoring method.

As seen in Figure 2-9, there were many cases of non-seam-weld cracking that
appear to have been called out with AE. These include cases of thermal/thermal
fatiguetype cracking related to hangers, supports, and attemperators, and
cracking at thick section (main steam) pipe girth welds that likely relate to
thermal gradients and transients with a creep component. The number of cases
documented indicates the potential capability of AE monitoring to prevent
failures that result from the cracking of these forms.
2-38
Summary observations are as follows:
The AE method appears to readily detect active thermal/thermal fatigue
cracking with sometimes greater response sensitivity to thermal transient
stimulation than to online load pressure swings.
The AE activity from a given source is highly dependent on the load or stress
cycle; for example, a 20% decrease in operating pressure can produce as much
as a 95% decrease in activity.
Perhaps due to the sensitivity to load and load changes and the inherent
variability in emission, there does not appear to be a direct correlation
between the activity measured in any one instance and the nature of the
emitting source with regard to extent of damage.
Indications are that an approach to periodic AE monitoring with consistent,
reproduced stress modes for a measure of AE response changes has potential.
The experience data show that AE monitoring as practiced does not produce
an unwieldy number of calls, nor does it produce many false calls. This makes
using the technology manageable.
AE appears to have detected the non-seam-weld forms of cracking in many
cases, supporting its potential value in preventing failures that result from
such cracking.
The details of the AE monitoring and the damage discovery in the five
reported detected cases of seam-weld creep damage are not known, so no
qualifying comments can be made regarding the robustness of these
correlations and their implication to the broader issue of the efficacy of AE
monitoring in preventing long seam-welded high-energy piping creep failure.
The experience data show that AE has the ability to prioritize nondestructive
surface and volumetric examinations.
The reported correlation of AE with a very early stage of creep cavitation
merits further investigation. As noted earlier, detection of this form of
damage provides no useful indicator to assist in integrity/life management.
Additionally, other possible sources of emission (for example, strain) that
may confound the interpretation need to be explored.
The issue of AE monitoring missing calls is of particular interest since the
technology may be viewed as a means of preventing failures. By itself, the
data accumulated to date do not support the notion that the technology
provides a foolproof method of preventing failures, as is the case with other
NDE methods. There are clearly many variables to implementation and
interpretation, not to mention the inherent variability in the AE response
associated with any given source (see the first chapter on the method).
However, the data indicate that AE monitoring has value, and the experience
reviewed puts the technology and its high-energy piping application potential
in an overall positive light.

2-39
Section 3: Summary
This chapter briefly summarizes the important observations and findings made in
the earlier chapters on the AE method and the service experience. In addition,
recommendations are made to enhance the robustness of the technology for the
high-energy piping application.

The AE Method as Practiced

There is substantial variability in the use of AE for monitoring of fossil plant


high-energy piping. Further, there remain many details of the applied data
analysis/evaluation procedures that are unknown. The following observations are
made:
The monitoring hardware (waveguides, sensors) have some variability from
the EPRI Guidelines recommendations:
- Waveguides are 1/4-in. diameter stainless steel rods weld-attached to the
pipe. Waveguide lengths used are about 12 in. and spaced 1218 ft apart.
However, in monitoring Grade 91 piping, smaller-diameter and shorter-
length rods have been used and attached to the pipe with lower heat
input capacitive discharge welding to conform to the ASME Boiler &
Pressure Vessel Code (Code Case 2649) requirements.
- Sensors with resonant frequencies in the range 300400 kHz are EPRI
recommended. To reduce the influence of flow-related noise, high-pass
filtering at 200300 kHz is recommended in the secondary amplification
stage. Lower resonant frequency (200 kHz) sensors with more aggressive
noise filtering are also used.
Instrumentation capability to include active source linear location, storage of
all AE event data passing the initial filtering criteria, and ability to record
time-sampled operating data such as temperature, pressure, unit load, and
flow rate are commonly used.
The floating threshold is generally used for all monitoring modes, although a
fixed threshold has also been used in a continuous monitoring mode.
The online stress methodsteady-state and with load swingshas emerged
as the primary stimulation method. Other modes include startup, cooldown,
and low-load operation.

3-1
The data analysis source ranking system referenced in the 1995 EPRI
Guidelines has evolved into the NEDR (normalized event density ranking)
to characterize located clusters, as reported for the vast majority of the post-
1995 work. However, the event energy measure of activity has also been used
since 2000, although the specific details of how the measurement is made are
proprietary.
There remains uncertainty regarding interpretation of the AE data. To date,
there is no evidence to support a reliable means of correlating activity levels
or intensity with the nature of the source. Indeed, the AE response of flaws
does not necessarily depend on flaw size or extent of damage. AE activity has
been shown to vary greatly with the magnitude or type of stimulation.
Obvious limitations of the practice are:
- Sensitivity decreases with increases in background noise.
- Sensitivity decreases with increased fluctuations in noise.
- An error in source locating can approach 20% of the waveguide
interspacing distance.
- Surface versus subsurface source location is not identified.
- Non-active flaws are not detected.
- Sometimes AE is too sensitive (small flaws with high activity), although
the EPRI Guidelines recommended practice has not produced an
unmanageable number of calls.
While consistency in the application is naturally challenging due to the many
variables that affect the measured output (for example, attenuation, varying
emission with stress method, location, signal characteristics changing
similarly, etc.), the high-energy piping case begs for some level of
standardization and greater transparency to further its development and
value.

AE Performance and Capabilities

The body of fossil plant high-energy piping AE monitoring experience data


provided was reviewed. Most of the data were provided by AE service providers,
at least for the post-1995 period (post-EPRI Guidelines) of primary interest. As
such, the inferences made here may not adequately reflect the experience.

Summary observations are as follows:


The AE method appears to readily detect active thermal/thermal fatigue
cracking with greater response sensitivity to thermal transient stimulation.
The number of instances where AE identified locations of such forms of
cracking (for example, hangers, supports, near-attemperator girth welds,
thick-section girth welds near wye blocks, etc.) supports its ability to detect
such forms of damage.

3-2
The AE activity from a given source is highly dependent on the load or stress
cycle; for example, a 20% decrease in operating pressure can produce as much
as a 95% decrease in activity. This sensitivity aspect of a source makes data
interpretation potentially problematic with regard to attempting to
characterize a source based on the measured activity.
Perhaps due to the sensitivity to load and load changes and the inherent
variability in emission, there does not appear to be a direct correlation
between the activity measured in any one instance and the nature of the
emitting source with regard to extent of damage.
Given, however, that indications are that a potentially useful measure of
damage may be made via repeat monitoring of activity, an approach to
periodic AE monitoring with consistent, reproduced stress modes for a
measure of AE response changes has potential value.
The experience data show that AE monitoring as practiced does not produce
an unwieldy number of calls, nor does it produce many false calls. This makes
using the technology manageable.
AE appears to have detected the non-seam-weld forms of cracking in many
cases, supporting its potential value in preventing failures resulting from such
cracking.
The details of the AE monitoring and the damage discovery in the five
reported detected cases of seam-weld creep damage are not known, so no
qualifying comments can be made regarding the robustness of these
correlations and their implication to the broader issue of the efficacy of AE
monitoring in preventing long seam-welded high-energy piping creep failure.
The experience data show that AE has the ability to prioritize nondestructive
surface and volumetric examinations.
The reported correlation of AE with a very early stage of creep cavitation
merits further investigation with regard to:
- The relevance of calling out significant AE in a case where the damage,
seen only via an unconventional cryo-cracking examination technique, is
at such a low level as to provide no useful indicator to assist in
integrity/life management.
- Confirmation of the detectability of this stage of cavitation is through
controlled laboratory experiments. The ability of AE to identify cracking
has been well accepted, and indeed, the EPRI experiments and
Guidelines development have always sought to use AE to detect
cracking.
- Potential sources of emission other than those reported need to be
examined.

3-3
The issue of AE monitoring missing calls is of particular interest since the
technology may be viewed as a means of preventing failures. The data
accumulated to date do not support the notion that the technology by itself
provides a foolproof method of preventing failures, as is the case with other
NDE methods. There are clearly many variables to implementation and
interpretation, not to mention the inherent variability in the AE response
associated with any given source (see the first chapter on the method).
However, the data indicate that AE monitoring has value, and the experience
reviewed puts the technology and its high-energy piping application potential
in an overall positive light.

Recommendations

With regard to the application of AE monitoring to fossil plant high-energy


piping, there are obvious issues with the variability in application of the AE
method, including aspects of transparency, completeness of the experience
database, and clarity with regard to what can be expected in any given AE
monitoring program. The following preliminary recommendations are offered to
help enhance the benefits of using AE for high-energy piping:
Develop a consensus on the variables (including test conditions)that are
optimally suited to the high-energy piping application:
- Obtain a level of transparency from the service providers and other
technicians that is sufficient to establish variables relating to hardware,
software, and data analysis/interpretation.
- Provide an information exchange forum for all practitioners, technicians,
and users (plant owner-operators).
Update the existing experience database with regard to:
- Details on select cases relating to creep
- Additional experience data from users for a possibly more balanced set
Empirically establish the AE response of various sources, ranging from creep
strain and isolated creep cavities to macro-creep cracks, under controlled
laboratory test conditions and using monitoring methods preferably with the
variables of consensus.
Update the 1995 EPRI Guidelines to incorporate all of the findings from the
variables consensus, the expanded experience database, and the results of the
laboratory experiments.

3-4
Section 4: References
1. Boiler Reheat Line Explosion, National Board Bulletin. Vol. 43, No. 2,
The National Board of Boiler and Pressure Vessel Inspectors, Columbus,
OH, October 1985.
2. Detroit Edison Today, Vol. 14, No. 4, Detroit, MI, February 1986.
3. Guidelines for Evaluation of Seam-Welded Piping. EPRI, Palo Alto, CA: 1987.
CS-4774.
4. Guidelines for the Evaluation of Seam-Welded High-Energy Piping. EPRI, Palo
Alto, CA: 1996. TR-104631.
5. Guidelines for the Evaluation of Seam-Welded High-Energy Piping. EPRI, Palo
Alto, CA: 2001. 1003978.
6. Guidelines for the Evaluation of Seam-Welded High-Energy Piping. EPRI, Palo
Alto, CA: 2003. 1004329.
7. Guidelines for the Evaluation of Seam-Welded High-Energy Piping. EPRI, Palo
Alto, CA: 2012. 1025326.
8. Acoustic Emission Monitoring of High-Energy Steam Piping, Volume 1: Acoustic
Emission Monitoring Guidelines for Hot Reheat Piping. EPRI, Palo Alto, CA:
1995. TR-105265-V1.
9. Unpublished Report: Steam Line Inspection and Condition Monitoring. EPRI,
Palo Alto, CA: June 1991. Final report on EPRI project RP 1893-4.
10. Unpublished Report: Acoustic Emission Monitoring of High-Energy Steam
Piping, Volume 2: Basis for Acoustic Emission Monitoring. EPRI, Palo Alto,
CA: 1995. Final report on EPRI project RP 1893-20.
11. Unpublished Report: Field Experience Summary for Acoustic Emission
Monitoring of High Energy Steam Piping. EPRI, Palo Alto, CA: 1999. TR-
111558.
12. J. M. Rodgers, Acoustic Emission Testing of Seam-Welded High Energy
Piping Systems in Fossil Power Plants, In Proceedings 6th International
Conference on Acoustic Emission, ICAE-6 Advances in Acoustic Emission - 2007,
Acoustic Emission Working Group (AEWG), CA.
13. Standard Terminology for Nondestructive Examinations, ASTM Standard E
1316-08a, ASTM International, West Conshohocken, PA, 2008.
14. Incipient Failure Detection for Fossil Power Plant Components, 1982 Conference
and Workshop. EPRI, Palo Alto, CA: 1982. CS-2920.
4-1
15. Online Acoustic Emission Monitoring of Fossil Power Plants A Critical
Assessment. EPRI, Palo Alto, CA: 1981. CS-1896.
16. J. M. Rodgers and R. M. Tilley Summary of Field Experience for Acoustic
Emission Monitoring of Seam-Welded High Energy Piping, In Proceedings
1999 EPRI Fossil Plant Maintenance Conference, EPRI, Palo Alto, CA: 1999.
TR-113534.
17. J. M. Rodgers, Acoustic Emission Consulting, Inc., to J. R. Foulds, Clarus
Consulting, LLC, via email, August 19, 2012.
18. I. Mizrahi, Margan Physical Diagnostics Ltd., Presentation at EPRI,
Charlotte, January 24, 2013.
19. I. Mizrahi, Margan Physical Diagnostics Ltd. to J. Foulds, Clarus
Consulting, LLC, via email, January 28, 2013.
20. D. Berger, PPL Generation to J. Parker, K. Coleman, and S. Walker, EPRI,
via email, October 17, 2012.
21. I. Mizrahi, Margan Physical Diagnostics Ltd., to J. Foulds, Clarus
Consulting, LLC, via email, February 21, 2013.
22. R. Griffin, NB Power Generation Corp., to J. Parker, EPRI, August 30,
2012.
23. M. Crichton, American Electric Power Co., to K. Coleman, EPRI, via
email, July 31, 2012.
24. M. Crichton, American Electric Power Co., to J. Foulds, Clarus Consulting,
LLC, via email, February 21, 2013.
25. T. Fox, GenOn Energy, Inc. (now NRG Energy, Inc.), to J. Shingledecker,
EPRI, via email, May 29, 2012.
26. J. C. Thornley, J. L. Speelman, and R. Griffin, Failure of a Seamless 2-
1/4Cr-1Mo Hot-Reheat Pipe Bend - Metallurgical Considerations, In
Proceedings of the 2010 Pressure Vessels and Piping Conference, ASME
PVP2010, Paper PVP2010-25713, ASME, New York, NY, 2010.
27. G. A. White et al., Failure of a Seamless 2-1/4Cr-1Mo Hot-Reheat Pipe
Bend - Acoustic Emission Testing and Fitness for Service of Other Steam
Pipe Bends, In Proceedings of the 2010 Pressure Vessels and Piping Conference,
ASME PVP2010, Paper PVP2010-25779, ASME, New York, NY, 2010.

4-2
Appendix A: Field Experience Data
19851990

Table A-1
In-plant AE monitoring cases in the period 19851990 from EPRI report RP 1893-
4 [9]. Italicized text has been added to the original table based on a review of the
text from that reference.

#7
July 1985 Large active ligament crack

#5 Plug sample: Oxide Inclusions; inactive stringer


Sep 1986 (40 minutes)
(LOF at Circ welds)

Mid-1987 (rubbing)
Startup &

#4 March 1987 as minor indications. Noisy startup not useful.

#1 May 1987
March-April 1988 & econ. inlet header
Startup, shutdown & swing

Nov 1987

#7
Jan 1988, Sep 1989

Magnetically coupled waveguides and 112 sensors


Late 1988

May, Oct 1988 Misses by AE attributed to noise; indications


not characterized

, 8 minor
April 1989

3
#1

AE misses under thermal stimulation
AE in single case seen with thermal transient

Used RP1893-4 Guidelines [9] & large scale


March 1990 monitoring

A-1
19901995

Table A-2
Listing of In-plant AE monitoring cases [10]. The listing includes the 12 cases of
19851990 (underlined).

A-2
Table A-2 (continued)
Listing of In-plant AE monitoring cases [10]. The listing includes the 12 cases of
19851990 (underlined).

A-3
Table A-2 (continued)
Listing of In-plant AE monitoring cases [10]. The listing includes the 12 cases of
19851990 (underlined).

A-4
Table A-2 (continued)
Listing of In-plant AE monitoring cases [10]. The listing includes the 12 cases of
19851990 (underlined).

A-5
19961999

A-6
Table A-3
Listing of in-plant AE monitoring cases as reported in EPRI report TR-111558 [11].

A-7
Table A-3 (continued)
Listing of in-plant AE monitoring cases as reported in EPRI report TR-111558 [11].

A-8
2000Current

A-9
Table A-4
Tabulated listing of in-plant AE monitoring cases prepared from a table provided by Rodgers [17].

A-10
Table A-5
Listing of experience as tabulated from data provided by Mizrahi [18, 19, 21].

A-11
Table A-5 (continued)
Listing of experience as tabulated from data provided by Mizrahi [18, 19, 21].

A-12
Export Control Restrictions The Electric Power Research Institute Inc., (EPRI, www.epri.com)
Access to and use of EPRI Intellectual Property is granted with the spe- conducts research and development relating to the generation, delivery
cific understanding and requirement that responsibility for ensuring full and use of electricity for the benefit of the public. An independent,
compliance with all applicable U.S. and foreign export laws and regu- nonprofit organization, EPRI brings together its scientists and engineers
lations is being undertaken by you and your company. This includes as well as experts from academia and industry to help address challenges
an obligation to ensure that any individual receiving access hereunder in electricity, including reliability, efficiency, affordability, health, safety
who is not a U.S. citizen or permanent U.S. resident is permitted access and the environment. EPRI also provides technology, policy and economic
under applicable U.S. and foreign export laws and regulations. In the analyses to drive long-range research and development planning, and
event you are uncertain whether you or your company may lawfully supports research in emerging technologies. EPRIs members represent
obtain access to this EPRI Intellectual Property, you acknowledge that it approximately 90 percent of the electricity generated and delivered in
is your obligation to consult with your companys legal counsel to deter- the United States, and international participation extends to more than
mine whether this access is lawful. Although EPRI may make available 30 countries. EPRIs principal offices and laboratories are located in
on a case-by-case basis an informal assessment of the applicable U.S. Palo Alto, Calif.; Charlotte, N.C.; Knoxville, Tenn.; and Lenox, Mass.
export classification for specific EPRI Intellectual Property, you and your
Together...Shaping the Future of Electricity
company acknowledge that this assessment is solely for informational
purposes and not for reliance purposes. You and your company ac-
knowledge that it is still the obligation of you and your company to make
your own assessment of the applicable U.S. export classification and
ensure compliance accordingly. You and your company understand and
acknowledge your obligations to make a prompt report to EPRI and the
appropriate authorities regarding any access to or use of EPRI Intellec-
tual Property hereunder that may be in violation of applicable U.S. or
foreign export laws or regulations.

Program:
Boiler Life and Availability Improvement Program

2013 Electric Power Research Institute (EPRI), Inc. All rights reserved. Electric Power
Research Institute, EPRI, and TOGETHER...SHAPING THE FUTURE OF ELECTRICITY are
registered service marks of the Electric Power Research Institute, Inc.

3002002218

Electric Power Research Institute


3420 Hillview Avenue, Palo Alto, California 94304-1338PO Box 10412, Palo Alto, California 94303-0813 USA
800.313.3774 650.855.2121askepri@epri.comwww.epri.com

Das könnte Ihnen auch gefallen