5 16 16 Signed Final Report To PST On ECA
5 16 16 Signed Final Report To PST On ECA
5 16 16 Signed Final Report To PST On ECA
By
Richard Kuprewicz
President, Accufacts Inc.
8040 161st Ave. NE, #435
Redmond, WA 98052
Ph (425) 802-1200
Fax (805) 343-2373
[email protected]
II
Description
of
what
an
Engineering
Critical
Assessment
is
as
related
to
this
proposed
rule
...............................................................................................................................
2
III
Description
of
what
an
ECA
should
include
to
adequately
assess
the
safety
of
a
pipeline
...........................................................................................................................................
3
IV
Discussion
of
the
strengths
and
weaknesses
of
ECAs
for
different
types
of
pipe
and
failure
modes
........................................................................................................................
6
V
Discussion
of
possible
concerns
with
the
use
of
ECAs
for
integrity
management
and
integrity
verification
..........................................................................................................
9
VI Suggested changes for the ECA proposal in this proposed rule ................................. 11
Accufacts Inc.
I Introduction
This paper focuses on the important subject of Engineering Critical Assessments, or ECAs,
as presented in the recent Notice of Proposed Rulemaking (“NPRM”) for natural gas
transmission and gathering pipelines.1 The NPRM, proposes to allow ECA under a new
regulation addition, §192.624(c)(3), as one of six methods to ascertain gas pipeline maximum
allowable operating pressure, or MAOP, on certain often called “grandfathered” pipelines
containing specific mentioned threat anomalies.2, 3 In addition, the NPRM also incorporates
additional detailed fracture mechanics modeling regulatory requirements to analyze pipeline
that may contain specific identified anomalies that could result in pipeline rupture as a new
section §192.624(d).
The September 9, 2010 San Bruno pipeline rupture and tragedy has been attributed to a
crack-like “seam” anomaly that became unstable and that grew with time on a pipeline
segment that had never undergone hydrotesting, under grandfathered regulations defining
MAOP. PHMSA reports additionally that, between 2010 and 2014, fifteen other reportable
incidents were attributed to seam failures.4 Seam failures fall into the category of crack
threats, and tend to fail as ruptures. While one intent of the integrity management (“IM”)
rule was to avoid such failures, these recent failures come as no surprise to this author given
the number of pipeline investigations on pipelines using ECA that have rupture failed
decades before ECAs predicted, and at pressures much lower than MAOP, in what I call the
“negative safety factor” region of pipeline normal operating pressures. Further clarifications
in high pressure pipeline integrity management regulatory approaches regarding ECA fitness
for service of gas pipelines are definitely warranted, and PHMSA has expended considerable
effort in identifying key parameters that influence prudent ECAs for pipelines.
A careful study of the proposed NPRM 549 page document explaining many of PHMSA’s
approaches will demonstrate, in my opinion, that PHMSA has done their research and
homework and is taking appropriate steps to codify needed clarification actions into
additional federal pipeline safety regulation utilizing ECAs as they apply to pipeline MAOP
1
Department of Transportation, Pipeline and Hazardous Material Safety Administration
(“PHMSA”), “Pipeline Safety: Safety of Gas Transmission and Gathering Pipelines, 49 CFR
Parts 191 and 192, Docket No. PHMSA-2011-0023,” posted on March 17, 2016.
2
The term grandfathered usually indicates the pipeline has never undergone a hydrotest, but
is using previous operating pressure to establish MAOP.
3
49CFR§192.3 Definitions, defines the MAOP as “the maximum pressure at which a
pipeline or segment of a pipeline may be operated under this part.”
4
NPRM, p. 30.
The term Engineering Critical Assessment and analysis (identified as “ECA”) is not
specifically defined in the proposed rulemaking. The rulemaking does explain in a proposed
new section permitting determination of MAOP (§192.624(c)), using Method 3: Engineering
Critical Assessment, that “An ECA is an analytical procedure based on fracture mechanics
principles, relevant material properties (mechanical and fracture resistance properties),
operating history, operational environment, in-service degradation, possible failure
mechanisms, initial and final defect sizes, and usage of future operating and maintenance
procedures to determine the maximum tolerable sizes for imperfections.”5 Industry standards
further explain, though also don’t define the term, as “ECA is a rigorous evaluation of the
data that reassesses the criticality of the anomaly and adjusts the projected growth rates based
on site-specific parameters.”6 In more layman’s terms, I see ECA as an analysis process
utilizing fracture mechanics principles to determine if a pipeline is structurally sound enough
to meet the service requirements for which it was intended, for a specific period of time.
One perspective on the history of ECA development as it relates to pipelines might prove
helpful. The science of fracture mechanics was developed and significantly advanced during
World War II when steel Liberty ships started to fail catastrophically (breaking in half). For
pipelines, the science of fracture mechanics applications tended to progress and follow the
initial development and use of inline inspection (“ILI”), or smart pig tools in the mid to late
1980’s that were first focused on pipe deformation (usually the less complicated to assess)
and also on general corrosion threats for obvious reasons. As pipe
location/deformation/caliper tools and general corrosion ILI tools (as well as computers)
improved, the engineering approaches and fracture mechanics science advanced, subject to
various limitations and/or boundary conditions. As general corrosion evaluations were more
5
Ibid., p 473.
6
Incorporated by reference into Federal pipeline safety regulation, “ASME B31.8S –
Managing System Integrity of Gas Pipelines – 2004,” p. 24.
In the 1990’s and 2000’s, attempts were made to try and develop cracking ILI tools to
address various vintage manufacturing (e.g., vintage ERW, FW) and transportation cracking
threats characteristic of single crack or “crack-like” anomalies causing pipeline rupture
failures. In the early 2000’s ILI tool evolution attempted to detect crack colonies or “crack-
field” threats that manifest neither as quite pure corrosion nor purely cracking. Crack-field
colonies are associated with stress corrosion cracking (SCC) and corrosion fatigue (another
form of environmentally assisted cracking) and require more complex ECAs than
conventional single crack or crack-like threats, though they tend to fail more like cracks.
Single crack-like anomalies utilize ECAs that estimate crack growth rates through a
combination of corrosion crack and pressure cycling. Crack-field growth rates, such as what
occurs in significant SCC, are more difficult to estimate/predict for ECAs as the presence of
multiple cracks that can interact or link unpredictably via corrosion and/or pressure cycling
may not be linear or predictable.8 Unfortunately, despite many ILI claimed advances, ILI
cracking tools today still have many limitations, though I understand and encourage
continued advancement in their application. Tool tolerances must be incorporated in IM
programs and their associated ECAs as clarified in the NPRM.
III Description of what an ECA should include to adequately assess the safety of a pipeline
There are three primary components in performing a proper ECA for a pipeline:
1) The threats that could cause a pipeline segment to fail by rupture need to be properly
identified.
2) Assessments must accurately measure each at-risk anomaly.
3) Fracture mechanics approaches performed on the identified at-risk threat must
incorporate assumptions that are truly conservative.
7
ASME B31G-1991, “Manual for Determining the Remaining Strength of Corroded
Pipelines: a Supplement to B31, Code for Pressure Pipeline.”
8
Some forms of SCC are not significant, even not readily observable by the naked eye.
These non-significant forms of SCC can usually be remediated by careful “buffing” to
remove the colony, if such remediation activity doesn’t overly penetrate the pipe wall.
Depending on the type of threat (deformation, general corrosion, cracking), its location on
the pipe (near welds, such as girth or axial seam welds), and the orientation (circumferential
vs. axial), the pipe can be evaluated as to its likelihood to fail within some period of time,
hopefully well past the next inspection via ECAs. I discuss the three main categories of
threats in further detail in the next section. In estimating time to failure by rupture using
ECA approaches, some assumptions must be made as to the rate of the threat’s deterioration.
This can be problematic if average industry growth rates are utilized, and if such averages are
much lower than actual growth rates experienced in the field at a specific pipeline location.
For interactive threats, such as SCC within general corrosion wall loss, predicted growth
rates can be too low and time to failure overly optimistic if they are evaluated as separate
threats. SCC also has an additional complication if significant colonies of cracks, rather than
a single crack, are present, as colonies of cracks can link (either lengthen or increase depth,
or both) over time making rate of growth and time to failure prediction models highly
unreliable.
Recent pipeline ruptures have involved too many instances where ECA approaches were
utilized and the time to failure ECA predictions were off by many decades. These failures
occurred at pressures well below the Predicted Failure Pressure, or PFP, clearly
demonstrating the associated ECAs were anything but conservative.10 Major problems in
these approaches involved a combination of poor engineering applications including:
9
NPRM §192.3 Definitions, pp. 431 – 439.
10
PFP is usually stated as a ratio of the predicted failure pressure divided by the MAOP.
An example of the first deficiency is the use of B31G for wall losses greater than 80 percent
that ignores the safety factors incorporated for this standard, developed from empirical
testing. The second deficiency can be shown in crack evaluation ECAs demonstrated by
assuming pipe toughness properties that don’t capture the extremely low values that can
accelerate growth rates and shorten time to failure associated with many vintage
manufacturing cracks. Growth rates may be underestimated for significant SCC colonies
where multiple cracks can quickly grow nonlinearly when multiple cracks link. This is
especially difficult if SCC is located in areas of severe general corrosion wall loss, placing
the total remaining wall loss not only from crack but also from general corrosion,
dangerously thin. The true pipe strength must be based on the actual remaining pipe wall
thickness reduced by both threats interacting.
PHMSA, in its proposed ECA NPRM approach, has clearly incorporated into regulation the
obligation to utilize certain important limiting conditions and/or values that should avoid
confusion when applying ECA approaches such as:
1. for general corrosion, the requirement to not use ASME B31G in wall thickness
losses greater than 80 percent, and
2. for vintage manufacturing crack threats “a lower toughness threshold of 5.0 ft-lb for
body cracks; 1.0 ft-lb for ERW seam bond line defects such as cold weld, lack of
fusion, and selective seam weld corrosion defects.”11
11
NPRM, pp. 482 – 483.
12
Ibid., p. 482.
Threats to pipelines tend to fall into three major fields or categories of pipe damage that are
briefly discussed below:
1) Deformation
2) Corrosion
3) Cracking
4a) Deformation
Certain activities around or on a pipeline during its various lifecycle stages can cause
dents, gouges, wrinkles or a pipeline to go out of round, weakening the pipe or its welds,
or introduce a combination threat, a dent with a stress concentrator, such as dents with
gouges. The latter, because of its unpredictability to reliably determine time to failure, is
not permitted in pipelines and require that “The operator shall {field} examine these
indications within a period not to exceed 5 days following determination of the
condition.”13 Federal pipeline safety regulations go one step further classifying as an
immediate repair condition “A dent that has any indication of metal loss, cracking or a
stress riser.”14 Wrinkles are also no longer permitted on newer pipelines, but may be
present on older systems.15
Deformation features reduce the pipe’s ability to hold certain pressures because the
containment structure has been changed by physical force. The science of fracture
mechanics focused on pressure vessel containment, with the exception of dents with
stress concentrators or dents around certain welds where the metal may have been
impacted in or near a weld such as in the heat affected zone, is well established. Reliable
high-resolution deformation ILI tools have been available for many years. The
advancements in ILI tools in this area and the development of fracture mechanics for this
category of threats yield a high degree of confidence on deformation damage ECA
assessments from either field or ILI measurement on pipelines. The exception to this
statement involves ECAs attempting to predict time to failure for dents with major stress
concentrators or dents near welds, as the science has yet to be proven to be reliable or
conservative for these specific dent threats.
13
American Society of Mechanical Engineers (“ASME”), “ASME B31.8S-2004 Managing
System Integrity of Gas Pipelines,” p. 23.
14
49CFR§192.933(d)(1)(ii).
15
49CFR§195.212(a) Bending of Pipe.
Given the limitations imposed in industry standards and now clarified in the proposed
pipeline safety regulations, I would also place a high degree of confidence in ECAs
associated with general corrosion from either field direct assessment or high resolution
ILI tool assessments. This confidence is reduced as wall thinning approaches 80 percent
loss, so a prudent pipeline operator will take special precautions on ECAs nearing this
limit.
4c) Cracking
Cracking became more of a public concern in the late 1980’s as various tragic vintage
pipe manufacturing crack ruptures occurred causing the public to become more aware of
this rupture risk that was well known in the industry. The 1990’s saw the first
development and field application of transverse ILI crack tools focused on manufacturing
crack detection that are axially aligned with the flow of the pipeline. In addition, failures
in the 1990s and 2000s, both in gas and liquid pipelines from SCC, and transportation
induced cracking of larger diameter thinner wall pipe, both very different forms of crack
16
ASME B31G-1991 (Reaffirmed 2004), “ Manual for Determining the Remaining Strength
of Corroded Pipelines: a Supplement to B31, Code for Pressure Piping.”
17
PRCI PR 3-805 (R-STRENG).
A subset of corrosion, selective corrosion such as selective seam weld corrosion, or SSC,
as well as SCC, and/or corrosion fatigue, must be handled differently from general
corrosion. For SSC, this difference is largely due to lower pipe toughness associated with
the possible “crevice seam” along/near the seam weld that tends to follow along the
axially aligned heat affected zone where corrosion may focus, generating a “crack like”
interactive anomaly. I have previously discussed the difficulties in estimating
conservative SCC corrosion rates in significant SCC colonies where individual SCC
cracks can link, both in depth and in length, to cause pipe rupture.
There is also another form of cracking associated with poor girth welding during pipeline
construction/installation that is causing pipeline rupture. In my opinion, no ILI
inspection tool is currently capable of reliably ascertaining girth weld cracks despite
many claims. This is one reason why some pipeline operators nondestructively inspect
all girth welds during construction even though federal minimum pipeline safety
regulation do not require such 100 % assessment. Because of the different hoop stresses
imposed by pipeline pressure, hydrotesting of pipeline, even to 125 percent of MAOP as
historically required in federal pipeline regulations, can leave very large cracks in girth
welds. These cracks can survive through many years of pipeline operation only to fail
when placed under different lateral stresses, such as surface loading or earth movement,
or pipeline change in service such as reverse flow. There may be other technologies
beyond ILI that can more reliably identify girth weld cracks, but associated ECAs not
using ILI, such as direct field measurement, should be able to be reliably performed on
girth weld cracks if they are recognized and accurately measured.
Given the limited developments and challenges associated with ILI cracking, either axial
or circumferential (girth weld) ILI identification and characterization, I would place a
low probability on ECA to reliably estimate time to failure for most cracks if girth weld
crack assessment is based on ILI, or if there are severe or significant colonies of axial
cracks that are often associated with SCC. Serious advancements in ILI development and
associated fracture mechanics efforts and pipe properties are required in this area before
ECA can be reliable for these two threat categories. Any claimed technical advances in
this area should be made transparent and public.
The IM regulatory process to avoid transmission pipeline rupture failure is data driven.
Without reliable and accurate data, coupled with sound engineering approaches within the
boundary conditions that may be included or imposed in industry standards, IM is just an
illusion of safety, based on poor risk assessment or incomplete safety management
applications. This author appreciates the need to advance technology and engineering
fracture mechanics efforts to avoid ruptures, and some of the challenges or hurdles to assure
transmission pipeline integrity utilizing ECAs. ECAs, however, must be based on sound
science and engineering principles that do not involve unrealistic assumptions, oftentimes
guesses, as to true pipe properties or anomaly characterizations (i.e., type, size, alignment,
and location) that can quickly lead to pipeline rupture failure.
Despite continued claims of various advances in ILI crack technology, identification, and
measurement, tolerances for this category of threats are highly dependent on the type of
crack or cracks, the ILI technical determination approach (i.e., ultrasonic, transverse mag
flux, EMAT), the crack(s) orientation/complexity, and location. ILI crack tool tolerances can
cover a wide range of values, both for POD and POI. For ILI tools, POD is the probability a
feature will be detected by an ILI tool, while POI is the probability of identification, the
probability that the type of anomaly once detected will be correctly identified (i.e., stated
usually with qualifiers and limitations). In addition to POD and POI, another important set
of parameters when using ILI tools is their accuracy and precision, usually stated as a
tolerance, that a pipeline operator should verify for their specific pipeline. The NPRM is
requiring that such tolerances be incorporated in ECAs when used.18
For cracks, regarding time to failure ECA predictions, the other major variable that can
confuse ECA time to failure estimates is the crack growth rate, a value that can vary
considerably, well beyond a safety factor of 2, depending on the type of crack and its
sensitivity to corrosion growth as well as pressure cycling growth, and other factors that can
accelerate time to rupture failure (such as lateral stress on girth welds). Not all cracks are
sensitive to corrosion attack, nor are all cracks sensitive to pressure cycling, so an
understanding of the crack growth rate used in ECAs is critically important as even fifty
percent time to failure benchmarks can be far off if too low a crack growth rate is assumed.
This can be especially problematic on gas pipelines that historically have not seen aggressive
pressure cycling, but by the nature of a modification or new operation, may be seeing
18
NPRM section 192.493 requiring compliance with three specific industry standards
referenced, API STD 1163-2005, NACE SP0102-2010, and ANSI/ASNT ILI-PQ-2005 when
conducting in-line inspection of pipelines.
Lastly, the NPRM mentions that some commenters have identified that ILI has severe
technical limitations as to its ability to clearly identify certain cracks. My experience is that
ILI is still in development concerning proper identification of cracks in girth welds, dents
with stress concentrators, and significant SSC. I believe PHMSA understands the highly
unreliable nature of ILI in this area and is encouraging the continued advancement in ILI, but
also other technologies, as well as engineering approaches that may more reliably field
determine the condition of such anomalies for continued service.
The approach in the NPRM to allow various methods to validate MAOP is most likely driven
by the information that PHMSA has garnered over recent years from improved annual
reporting concerning MAOP determinations. I would expect that PHMSA has discovered
that there is more transmission pipeline mileage operating under the grandfathered clause of
pipeline safety regulations, §192.619(c), than first expected. The various proposed methods
to validate MAOP in the NPRM make sense. Permitting ECA methods where appropriate,
encourage and take advantage of important technical advances concerning pipe integrity
evaluation. ECA use, however, needs to incorporate important qualifiers that should be
included in specific regulation to reduce possible ECA misuse via poor assumptions and
unsound engineering that I have observed all too often in past IM programs, while
encouraging technology improvements. I understand the need to foster such considerations
especially in the further use/development of ILI applications. ILI, if properly chosen and
applied, can tell operators more about the condition of a pipeline than any other assessment
method, provided that the ILI tool capabilities and limitations are understood and their results
prudently and timely utilized.
To keep the door open for further advancements in the NPRM MAOP determination
approaches, PHMSA has wisely permitted the use of “other technologies” provided that such
efforts are reviewed and approved by PHMSA before their use. PHMSA also recognizes that
smaller diameter pipelines (many associated with intrastate pipelines) have significantly
smaller actual rupture impact areas than larger diameter higher stress transmission pipelines.
While one doesn’t want rupture even on these smaller pipelines, such smaller diameter
pipelines also may not be able to accommodate ILI tools. Other assessment approaches are
warranted to validate a pipeline’s integrity for a specific MAOP.
While strongly implying which threats need to consider ECA approaches defined in the
NPRM, the proposed rule is less clear as to how certain specific threats are to be evaluated so
that an ECA may be appropriate. While I believe PHMSA’s intent is well meaning, this lack
of clarity could leave ECAs open to less than sound engineering approaches, especially those
based on assumptions that don’t really apply to a specific field location. Based on many
decades of observation, PHMSA may be attempting to advance new technology for certain
threats (such as ILI or “other technologies”). The problem with such an effort is that the
public may not be made aware of the uncertainties and risks associated with such “advances”
that could result in pipeline rupture, as such engineering attempts have not been properly nor
publically vetted. In many cases ILI will be the appropriate assessment method, in other
cases it will not. While I understand and appreciate PHMSA’s intent to foster ILI
development and other technology advancements, the fact remains that certain risks that can
and have resulted in pipeline rupture, have yet to be reliably determined by ILI and their
associated ECA methods. The ILI has not proven reliable, or if the ILI has advanced, the
associated ECA method may not be reliable or truly conservative.
19
49CFR§192.243 Nondestructive testing.
In conclusion, despite my early aversion to ECA approaches based on too many pipeline
rupture investigations that clearly demonstrated many flaws in various ECA approaches that
claimed transmission pipeline rupture would only occur well above PFP, PHMSA’s ECA
approach in the NPRM is well thought out, based on sound research that has been made
public, and an apparent understanding of past failures reported to the pipeline safety agency.
With the exception of the three recommendations noted above, I thus support PHMSA’s
approach using ECA, subject to the condition that efforts do not delete the clarity submitted
for MAOP methods for ECAs in the rulemaking process defined in the NPRM. Should
PHMSA be placed in a situation to approve ECAs related to the three crack threats identified
above, such approval should be conditioned on the requirement that the determinations and
associated ECAs be made public to assure a proper and scientific vetting of the approach, and
that it is indeed based on solid science and realistic engineering assumptions that are truly
conservative.