Tij - May-June 2008
Tij - May-June 2008
Tij - May-June 2008
JUNE
2008
_____
Vol 14
Issue 3
Alberta Boiler p. 4
Safety Association
Issues Rev 5
European p. 5
Commission
Launches Study
Developing p. 6
Methodology of
Pulsed Thermal NDT
2008 p. 16
International
Pipeline Conference
You Don’t p. 17
Get Something
For Nothing
2009 API p. 21
Inspector Summit
(ISSN 1082-6955)
❑ YES
I would like to subscribe to the INSPECTIONEERING JOURNAL
Single subscriptions cost $250.00 U.S. per year (6 issues). 10% discount for
2 yr. subscriptions. Contact us for details on multiple subscription pricing.
Non-U.S. subscribers add .50¢ per year for postage.
___________________________________________________________________________________________
_
Subscriber Name (Recipient)
___________________________________________________________________________________________
_
IND E X - Company Name
____________________________________________________________________________________________
Alberta Boiler Safety.................... 4-5 Address
Association Issues Rev 5
___________________________________________________________________________________________
_
City State/Province Zip/Code
____________________________________________________________________________________________
Email URL
Developing Methodolgy of.......... 6-15
Pulsed Thermal NDT ❑ Renewal ❑ New Subscription
❑ Check Enclosed ❑ Bill Me (payment due in 30 days)
___________________________________________________________
You Don’t Get Something ............... 17 Credit Card No. Expiration Date
For Nothing
_________________________________________________________________________________
Cardholder Signature Date
2009 API Inspector Summit............ 21
Please make checks payable to The Inspectioneering Journal
5315 FM 1960 West, Suite B 237
Houston, TX 77069
Telephone: (281) 397-7075 Fax: (281) 397-9996
Email: tij@gte.net Website: www.inspectioneering.com
The latest revision of this code (AB-506) is dated January small oil and gas processing facilities and commercial and
28, 2008. For many readers in the province of Alberta other applications.
Canada these rules will impact you directly. Others may
The information in AB-506 and other referenced ABSA policy
see effects or feel indirect effects as you jurisdictions may
documents was developed, and is updated periodically,
look to ABSA for direction or strongly consider their actions
based on ongoing consultation with Alberta pressure
for establishing other jurisdictions’ rules.
equipment owners and other stakeholders and information
The ABSA, as for many other jurisdictions, rules make from codes, standards and other published information.
frequent references and rely on API inspection codes such This process is designed to ensure that policy documents
as API 510 the pressure vessel inspection code, API 570 issued by ABSA, as the Alberta pressure equipment
the piping inspection code, API RP 580 the Risk-Based jurisdiction, reflect current best industry practices that are
Inspection recommended practice. suitable for all industry sectors.
It is important to note that often ABSA, as well as many other The sections of the National Board Inspection Code NB-
jurisdictions, “builds fences around” or imposes additional 23 and API-510 Pressure Vessel Inspection Code that are
requirements around the use of the referenced codes and referenced, shall be used in conjunction with AB- 506 to
rules, as demonstrated below in excerpts from AB-506. determine the inspection requirements and inspections
Operators should be well aware of and in compliance with intervals for pressure equipment.
all requirements of their respective jurisdiction.
While the principles for establishing inspection intervals
Per the purpose section of the document: in API-510 and NB-23 Codes are similar, API Codes were
“This AB-506 Pressure Equipment Inspection and developed for pressure equipment used by the petroleum
Servicing Requirements document (ISRD), has been issued and chemical process industries and NBIC is intended for
by the Alberta pressure equipment safety Administrator, to pressure equipment that is not covered by API Codes.
specify inspection and servicing requirements for pressure It should be noted that these codes are not adopted as
equipment under the Safety Codes Act. It covers the regulations in Alberta and that they contain a caution
requirements for determining inspection practices and that if their use is in conflict with jurisdictional regulatory
establishes maximum inspection/servicing intervals for requirements, the jurisdictional requirements shall prevail.
pressure equipment and pressure relief devices. In Alberta, there are a number of exceptions and additional
The application of AB-506 under an Owners Certificate of requirements that must be met in order to comply with the
authorization permit issued under the Pressure Equipment Safety Codes Act and (PESR).
Safety regulation (PESR), and for risk based inspection The grading system that is described in this procedure
programs provided in accordance with AB-505 Risk-Based is based on the Institute of Petroleum Pressure Vessel
Inspection Requirements for Pressure Equipment is also Examination Code of Practice grading system. It was
addressed.” selected as an adjunct to the above Codes to provide
From the General section of this document: an objective method for setting inspection intervals that
“Pressure equipment for process applications installed is suited for all sectors of Alberta s pressure equipment
in Alberta covers a broad range of facilities from major industry.”
petrochemical plants, pulp mills, and power utilities to
Editor’s note: The following type of development work, often paves the way to development of better NDE flaw or
damage detection, sizing and characterization capabilities. Because it could supply accurate information about actual flaws
in a test sample, without destructive testing, it could help ensure the production of better samples for industry qualification
demonstration testing by knowing more accurately the defect contents and features. Another area is for development of
simulation systems and equipment for flaw detection, characterization and sizing programs, for qualification demonstration
testing without shipping around large, heavy samples to test NDE candidates. Obviously, better NDE technology and use
of technology should result, as well.
There is also some good information on the importance of signal to noise ratios for interpretation of NDE information.
∆T , τ (∆T ), C
m m m and τ (C )
m
The values of and which appeared in our experiment have been found by considering two
points for each defect: a central (defect) point and a point placed close to a defect but still regarded as non-defect. The temporal
evolutions of ∆T and C are shown in Fig. 3 along with the corresponding maximum values. In accordance to the TNDT theory, it
is seen that maximums of C occur later than the maximums of ∆T . Oppositely, if both maximums would occur within heating, their
order will be reversed.
Three-dimensional theoretical temperature distributions were calculated with the ThermoCalc-6L program from Innovation, Ltd.,
Russia. Round-shape defects were substituted with parallelepipeds having the same lateral area (Fig. 3a) and ∆T and C values were
determined for each defect by analyzing the corresponding temporal evolutions (Fig. 3b). The comparison between the theoretical and
∆Tm , τ m (∆T ), Cm and τ m (C ) is presented in Table 1.
experimental values of
The average divergence between the theory and the experiment (neglecting extreme values) is 32% by ∆T and 19% by τ m (∆T )
Cm and 21% by τ m (C ) that is explained by: 1) uneven heating, i.e. varying value of Q, and 2) noisy character
, respectively 41% by
of ∆T and C evolutions.
In practice, optimal detection parameters are determined by analyzing temporal evolution of a signal-to-noise ratio SNR which
adheres to a particular defect and can be calculated if two areas are chosen: defect and non-defect:
Td − Tnd
SNR = , (1)
σ nd
T ,T σ
where d nd are the mean temperatures in chosen areas, and nd is the standard deviation of temperature in a non-defect area. The
example of the SNR (τ ) evolutions for Defect #5 and #8 is shown in Fig. 4. It is clear that optimal observation times correspond to
times when SNR values are maximal. These values are subjective because they depend on how the areas are defined. Typically, a defect
area should cover a ‘visible’ part of the corresponding defect and a non-defect area may cover a whole sample or an area comparable to
the defect area by size and shape. It is important that the SNR concept reflects distribution of pixel amplitudes but does not explain the
heuristic nature of how a trained operator identifies defects on a noisy background. In fact, in some cases, a defect which is characterized
by a lower SNR value can be easier detected by an experienced operator due to other informative features, such as shape, pixel coupling,
signal temporal behavior etc.
SNR is a good tool for optimizing test procedures and data processing algorithms. Some examples are shown in Table 2. First of
all, it is well seen that a higher heating power ensures higher SNR values probably because, in the particular experiments, the additive
noise dominated over multiplicative (surprisingly, the phasegrams proved to be inefficient under low-power heating). Shuttering thermal
radiation after heating has been useful due to cutting reflected radiation. Finally, it has been found that different defects may require
different processing algorithms, e.g. Defect #8 is better seen in the image of Fourier magnitude (as well as in the ‘best’ source image)
rather than in the phasegram, while the phase treatment has proven to be optimal in identifying Defect #5.
Q ~ 20 kW/m2(halogen lamps)
Shutter, sample monitored under angle 30o
16.7 20.3
Image of phase
1.7 53.9
Image of magnitude
12.8 57.8
Best source image
Q ~ 20 kW/m2 (halogen lamps)
No shutter, sample monitored under angle 30o
7.3 13.4
Image of phase
6.4 108.2
Image of magnitude
10.5 66.1
Best source image
Q ~1.6 kW/m2 (halogen lamps)
Shutter, sample monitored perpendicularly
0.1 0.7
Image of phase
1.9 13.3
Image of magnitude
0.6 11.2
Best source image
Q ~ 0.6 kW/m2 (air fan)
Shutter, sample rotated after heating
0.3 0.6
Image of phase
0.4 6.1
Image of magnitude
0.02 4.4
Best source image
In this study, the data treatment was fulfilled by using the ThermoFit Pro program from Innovation, Ltd. The results are presented in
Fig. 5 and Table 3. A correlogram is the image where each pixel value represents a correlation coefficient between a current pixel and a
pixel accepted as a reference [4]. Polynomial fitting allows approximation of T (τ ) evolutions with some polynomial functions [5]. As
a result, even a very long sequence can be substituted with few images of polynomial coefficients. The advantages of this technique are
the following: 1) images of coefficients may reveal defects better than the source images, 2) useful information can be concentrated only
in few images, 3) a source sequence can be restored to smooth the T (τ ) evolution; such evolutions can be further processed with some
techniques which can be hardly applied to noisy functions, e.g. derivation. Principle component analysis (PCA) is a useful statistical
procedure which is becoming increasingly popular in NDT [6]. It results in few images of significant signal components which reflect
peculiarities of T (τ ) evolutions. Phasegrams are results of the 1D Fourier processing in time obtained for a chosen temporal frequency
[7]. The most informative are phasegrams obtained at low frequencies because low-frequency thermal waves penetrate deeper into
materials. Since the Fourier treatment takes into account overall features of T (τ ) evolutions, pulse phase thermography (PPT) is often
regarded as a processing technique No. 1. This technique requires no specific knowledge on analyzed functions and can be applied to
arbitrary signal evolutions. Finally, dynamic thermal tomography (DTT) is a special data treatment technique which is based on the
τ m is
fact that, in one-sided TNDT, deeper defects produce lower signals at longer times, therefore, selecting particular intervals of
equivalent to ‘slicing’ a material by separate layers [8].
A map of defects is a binary image which appears as a result of ‘thresholding’ any source image. This concept can be easily
illustrated by considering an image histogram. First, the user has to define defect and non-defect areas (see squares in the image of Fig.
6) which produce two corresponding populations characterized by a particular SNR value according to Eq. (1). Choosing such areas is a
subjective process but effective while comparing different images at a fixed configuration of defect and non-defect areas. In a histogram,
these populations can be well-separated or cross each other. In the first case, ‘thresholding’ is trivial but, in the second case, the location
Pf .a. Pc.d . (see the histogram in Fig. 6). In
of a threshold defines the probability of false alarm and the probability of correct detection
P P
such way, a particular configuration of defect and non-defects areas results in a single SNR value but plentiful pairs of f .a. and c.d . .
Some binary images are presented in Fig. 7. The first two images are obtained by ‘thresholding’ the best source image (all 8 defects
are covered by defect areas and the rest of the sample is regarded as non-defect to produce SNR=2.5). Another pair of binary images
was obtained by a phasegram (SNR=13.6). Let us consider Defect #4. It is not seen in the source binary image (Fig. 7a) because of
Pf .a. Pf .a.
the low =5% value accepted. However, this defect turns to be visible if to accept =30% (Fig. 7b). In the phasegram binary
Pf .a.
image, Defect #4 is clearly seen at both =5% and 30% values. It is worth noting that enhancing correct detection can be achieved
Pf .a.
by relaxing requirements to
Fig. 6. Identifying defect and non-defect areas in the bakelite reference sample
Pf .a. Pc.d . =78%)
(ThermoFit Pro, Statistics option, best source image, SNR=2.5, =30%,
Several defect identification procedures have been proposed in pulsed TNDT [9-12]. Two defect characterization algorithms
conventionally called 1D and 3D are implemented in the ThermoFit Pro. The 1D algorithm assumes defects to be laterally infinite, thus
only two defect parameters (defect depth l and defect thickness d) can be evaluated. The input parameters are material thermal properties
and a heating time. The program calculates a maximum running contrast and a time of its appearance over a defect chosen by a user and
then produces images of defect depth and thickness.
The results of applying this algorithm to all 8 defects in the reference sample are presented in Table 4. The accuracy in evaluating
defect depth is from 20 to 30% on average. Note that defect thickness estimates are fairly inaccurate due to the fact that the above-
described algorithm is valid for thin hidden defects, such as delaminations, rather than for bottom-holes.
Table 5 contains characterization results obtained with the 3D algorithm. The principal feature of this algorithm is that the user has to
evaluate maximum and minimum lateral size D of an identified defect by using a kind of a threshold algorithm (the example is shown
in Fig. 6). Except small-size defects, the accuracy of determining D has been from 5 to 30% in our case, while errors in the determining
defect depth l have been typically less than 10%, i.e. better than in the case of the 1D algorithm (defect thickness estimates are also
invalid in this case).
In this study, we have summarized the concept of TNDT including the simulation of finite-size defects in solid materials, optimization of
test procedures and advanced data treatment based on maximizing a signal-to-noise ratio and applying 1D and 3D defect characterization
algorithms. The experimental results have been obtained on a bakelite reference sample which contains bottom-hole defect surrogates of
different depth and thickness. The proposed approach is quite general thus allowing optimizing pulsed TNDT findings.
References
1. Vavilov V.P., D.D. Burleigh and Demin V.G. Advanced modeling of thermal 7. Maldague X., Marinetti S. Pulse phase infrared thermography.-J. Appl. Phys.,
NDT problems: from buried landmines to defects in composites.-Proc. SPIE 1996, Vol. 79, p. 2694-2698.
“Thermosense XXIV”, Vol. 4710, pp. 507-521. 8. Vavilov V., Maldague X. Dynamic thermal tomography: new promise in the
2. Vavilov V.P. and Maldague X. Optimization of heating protocol in thermal IR thermography of solids.-Proc. SPIE, “Thermosense-XIV”, 1992, Vol. 1682, p.
NDT: back to the basics.-Intern. E. & Instr., pp.132-138. 194-206.
3. Vavilov V.P. Evaluating the efficiency of data processing algorithms in 9. Vavilov V., Grinzato E., Bison P.G., Marinetti S. Thermal characterization
transient thermal NDT.- Proc. SPIE “Thermosense XXVI”, Vol. 5405, 2004, pp. and tomography of carbon fibre reinforced plastics using individual identification
336-347. technique.-Mater. Evaluation, May 1996, Vol. 54, No.6, p. 604-611.
4. ThermoFit Pro operation manual, Innovation Ltd., Russia, 2007, 72 p. 10. Krapez J.-C., Maldague X., Cielo P. Thermographic NDE: Data inversion
5. Grinzato E., Bison P.G., Marinetti S. and Vavilov V. Thermal NDE enhanced procedure (Part II: 2D analysis and experimental results).-Res. in NDE, 1991. No.
by 3D numerical modeling applied to works of art.-In: Proc. 15th World Conf. on 2, p. 101-124.
NDT, Rome (Italy), 15-21 Oct. 2000 (available only on CD), 9 p. 11. Winfree W.P., Zalameda J.N. Thermographic determination of delaminations
6. Hermosilla-Lara S., Joubert P.-I., Placko D. et al. Enhancement of open-cracks depth in composites. // Proc. SPIE “Thermosense-XXV”, 2003, Vol. 5073, p. 363-
detection using a principal component analysis/wavelet technique in photothermal 373.
nondestructive testing.-Abstr. Intern. Conf. Quant. Infrared Thermography 12. Almond D.P., Saintey S., Lau S.K. Edge effects and defect sizing by transient
QIRT’02, Dubrovnik, Croatia, Sept. 24-27, 2002, p. 12-13. thermography.-”Proc. Quant. Infr. Thermography QIRT’94”, Eurotherm Seminar
#42, Sorrento, Italy, August 23-26, 1994, p. 247-252.
The significant problems we face cannot be solved from the same level of thinking at which they were created.
Albert Einstein
Do we know how to free ourselves do anything to raise us to a new level of thinking to help us
from the current level of thinking to overcome our problems? Would we really even know it? Do
rise to a higher level? It will take we have the courage to tell management or others what we
this for us to overcome many of really think? Do we know how to do it constructively without
our current challenges in the arena losing the substance of the message? Do we have the courage
of equipment reliability. We are to say, “I made a mistake”?
responsible for this critical role as
As George Santayana wrote in his work, The Life of Reason,
the mission for many of us, when
Vol. 1 Reason in Common Sense, “Those who cannot
it comes to equipment integrity and
remember the past are condemned to repeat it.” So what are
reliability, goes something like this:
you doing differently? How are you improving, individually,
personally, professionally? How are you migrating that to your
Achieve regulatory and corporate
fixed equipment reliability management program? I am a firm
compliance, and ensure reliable
Gregory C. Alvarado believer that personal growth is necessary before we can
use of equipment (including
Chief Editor contribute significantly to other areas of our lives. One of the
piping) for finite run times,
first places to start is with honesty, i.e. honesty with ourselves
Inspectioneering Journal while measuring, managing and and with others. Check our motives. Are they noble? Are we
minimizing risks and eliminating
making this decision or peddling our influence for selfish or
non-value adding activities and costs.
noble reasons?
Risk would include health and safety and environmental
It isn’t easy, but few worthy endeavors are. Let’s start with,
elements, as well as business.
perhaps one of the biggest challenges……
Items that may impede rising to a new level might be fear,
greed, culture, politics, busyness, “turf wars”, pride, focus on
Culture – Inspectors and Fixed Equipment
short term profits, lack of appreciation for long term thinking
Reliability Engineers
and actions, prejudice, not researching the facts, ignorance In the world of RBI and fixed equipment reliability, there are
(this can breed a state of unconscious incompetence), having various team members but there are basically two roles that
the wrong people in the position, etc. Just throwing money at who are most engaged on a day to day basis, that of the;
the challenges or still “doing it on the cheap” are not thinking
at a higher level either. Industry has been doing this for quite • Fixed equipment reliability engineer, who may be
some time, most unsuccessfully as evidenced by continued a degreed engineer or a senior inspector, who via
industry losses due to equipment failure1. experience, certifications and education, is an engineer
in his/her own right.
Just because we place a new wrapper on a piece of software, • Then we have the inspector, who provides information
add few buzzers and whistles or claim, “this software has an to the fixed equipment reliability engineer and either
SAP link”, or to say it does everything (but having a shallow performs or is responsible for the oversight and quality
technical basis), we must ask ourselves, did that really of inspections. This information is used to “fine tune”
us understand how equipment fails and appreciate 1. Cost of the software and implementation costs are
how sensitive NDE methods need to be, or don’t need about the same, in general.
to be, to find the damage or degradation that would 2. Resource requirements to implement are about the same.
“go critical” before the next inspection. 3. The ability to understand and use the software requires
• Using FFS to make run, repair, and replace decisions. thought for the quantitative tools.
It surprises me how many people are making 4. Results using qualitative tools are roughly equal to
unnecessary replacements and repairs and leaving a traditional API 510 and 570 approaches, i.e. very few
lot of money on the table in the process by not using results, albeit a few, will provide significantly longer
recommended practices like API RP 579. inspection intervals than the traditional approaches.
• Corrosion systemization and circuitization of piping This may leave some with a sour taste in their mouths
systems. The circuits that result from this exercise eventually because non-value adding activities are not
should be used for your RBI and inspection database being identified and ultimately RBI did not free them
management systems. This will allow you to track CML up to do more valuable things. In these cases it is just
thickness readings as associated with true corrosion another thing that they already did not have time to do.
circuits. What a novel idea? As a corporate or team I have a couple of theories of why this happens:
learning exercise, this leads to better understanding, a. Sellers do not want to expose themselves to
especially when coupled with RBI, a FFS perspective, risk they cannot back up technically.
damage mechanisms reviews and MOE’s, of the b. It costs too much money and resources are
dynamics and vulnerabilities associated with piping scant to build better predictive models.
failure, in advance to produce a more proficient 5. Qualitative tools output static risk matrices
and productive piping management program. If you 6. Some quantitative tools output static risk matrices,
question the worth of such an approach check out a and dynamic metrics, some with dynamic modeling,
few case histories in the first reference at the end of showing risk gradients over time and accompanying
this article. probability of failure trending over time
• And more… 7. Qualitative tools by their very rarely, if ever, possess
the ability to know the amount of scatter in their
Note that I have not gone into already established practices predictions, hence their not being much different in
like positive materials identification programs, welding benefit from using traditional 510/570 approaches
inspection programs, QA/QC programs, inspection database other than providing an opportunity to perform on-
management programs and others. They are important. I often stream versus internal inspections to “skirt” regulatory
look at all these programs as interdependent “layers” of a good requirements in a non-RBI environment.
fixed equipment reliability program. None of them by itself 8. Quantitative tools with their working process, should
provides 100% assurance that we will avoid failures. Together, provide the information to understand and measure
they make a formidable team to prevent that “bad actor’ from the amount of scatter in most any prediction,
passing through all the layers. understanding the assumptions, if any were made in
the process.
It is critical when making assumptions in the RBI process,
for either qualitative or quantitative technologies, that highly
experienced people, both with the technology, understanding
the sensitivities of the input data and impact on the model
Sunoco, Inc., an independent refiner and marketer and experienced with the specific type of unit, are consulted
of petroleum in Toledo, has an immediate need to ensure that assumptions are reasonably conservative. The
owner operator must be involved in this process, review and
for an Inspection Manager. approve the assumptions.
The selected applicant will manage a staff of
inspector’s to ensure the mechanical integrity of the Implementing and maintaining an effective RBI program will
refinery and provide technical leadership in solving require thought, holistic thinking and reasoning, honesty,
synergizing with complimentary initiatives, resources (internal
complex problems involving refinery equipment. and possibly external) and investment in software.
Candidates must hold a BS in Engineering and ten Ultimately the lagging parameters of % availability, % reliability,
years of refinery experience or 20 years inspection loss or improvement of profit or production opportunities due
to equipment reliability will be the judge of our success at
experience with ten or more year’s refinery inspection improving fixed equipment reliability programs. It will require
background. Knowledge of API 5/10/570/653. rising to a higher level of thinking.
Supervisory experience required. 1
The 100 Largest Losses 1972-2001, Large Property Damage
Losses in the Hydrocarbon-Chemical Industries, 20th Edition, 2003,
For confidential consideration, please submit your Marsh’s Risk Consulting Practice, Marsh and McLennan Co.
resume through our website career center at 2
New Forces at Work in Refining – Industry Views of Critical Business
www.sunocoinc.com and Operations Trends, D.J. Peterson and Sergej Mahnovski.
EOE/M/F/D/V Prepared for the National Energy Technology Laboratory, United
States Department of Energy, ISBN 0-8330-3436-7, 2003
Please use the order form in this journal, or you may download an order
form from http://www.Inspectioneering.com/mailorderform2.htm or by
visiting the journal web site at www.Inspectioneering.com
Make sure you subscribe directly with the Inspectioneering Journal for your
2009 subscriptions. Please feel free to contact me with any comments or
feedback. It is appreciated. Greg Alvarado, Chief Editor
VCESageTM - The benchmark program for equipment review and analysis of structural integrity,
code compliance, re-rating and remaining life evaluations
VCEIntelliJointTM - The “total joint solution” that incorporates knowledge in all the areas required
to effectively solve the root cause of leakage
VCEDamage MechanismsTM - The quick reference guide to identify the potential damage
mechanisms that can cause costly equipment failure