Journal of Loss Prevention in the Process Industries 65 (2020) 104125

Contents lists available at ScienceDirect

Journal of Loss Prevention in the Process Industries

journal homepage: http://www.elsevier.com/locate/jlp

Effects of geometry of a local thin area defect on remaining strength and

fitness of pressure equipment
Yen-Ju Lu, Chen-Hua Wang *
Department of Safety Health, and Environmental Engineering, National Kaohsiung University of Science and Technology (NKUST), No. 415, Jiangong Rd., Sanmin Dist.,
Kaohsiung City, 80778, Taiwan, ROC

A R T I C L E I N F O A B S T R A C T

Keywords: As an important equipment for production, pressure vessels are important equipment in the factory and can be
Local thin area (LTA) dangerous when there is structural damage. In recent years, the development of applicability assessment tech­
Remaining strength factor (RSF) nology has enabled various defects to be properly evaluated. The local thinning defect (LTA) basically uses the
Fitness for service (FFS)
RSF as an indicator to measure the eligibility of the device. In the evaluation process using API 579 Level 1 and
Pressure equipment
Level 2, different LTAs have the same residual strength results, so the availability evaluation results may have
doubts. However, when using API 579 Level 1 and Level 2 for defect evaluation, it was found that two defects of
the same length but different widths had the same residual strength. Therefore, this study explores its causes in
depth.
In addition, the practice of treating the defect thickness profile in CTP mode in API 579 Level 2 is considered to
be relatively conservative and hence be safer. However, reordering the defect profile in fact smooth out the stress
concentration effect and offers an overly optimistic estimate on the fitness of the device.
First of all, the experiment started off with a set of real defect data, which was taken from measuring a
localized thin area on a cylindrical pressure vessel. By changing the width of the LTA, a list of defects were
generated and finite element analysis was performed. What follows is to compare results of finite element models
with defects modeled with true thickness profile, critical thickness profile(CTP), and parabolic profile.
For a square LTA, results from API 579 Level 2 approach is quite close to those from finite element analysis.
However, the variation increases as the aspect ration of width to length of LTA becomes large. Contrary to the
API 579 Level 2, the width dimension of LTA does have influence RSF, and should be taken into account. Also, in
API 579, a rearranged CTP was used instead of the true profile of LTA on a belief that results from using CTP is
more conservative, which later was found to be incorrect. Hence, the true thickness profile is preferred in most
cases.

to assess. Therefore, a reliable method for assessing the remaining

strength of equipment is required (Janelle, 2005). The length, width,
1. Introduction depth, and profile of a metal loss area are key when evaluating strength
at the defect. Depth is more crucial when determining the ultimate
Pressure vessels are key to production and potentially hazardous due strength, as demonstrated in the bursting strength studies conducted by
to the number of chemicals they contain (Dupen, 2016). During opera­ Cai and Jiang (2017), which is also why much research opted for a
tion, flaws can appear on the interior and exterior surfaces of vessel simplified geometry such as parabolic curve for the defect profile for
walls that might cause critical and unwanted events (Dumitrescu et al., stress analysis.
2018). Minimum required thickness (tmin) is calculated in accordance with
Metal loss (thickness loss) is a common type of equipment defect (Cai design criteria including design conditions (pressure and temperature),
et al., 2018; Cai et al., 2017) and the main reason for structural failure. geometry, and allowable stress (ASME, 2019a; ASME, 2019b). Corrosion
These defects can be roughly divided into two types: general and allowance, as a safety margin for future operation, is then added to tmin
localized metal loss. Localized metal loss is the most hazardous because to obtain the design thickness. Finally, the nominal thickness (tnom) is
it is difficult to detect and measure, making equipment strength difficult

* Corresponding author.
E-mail address: [email protected] (C.-H. Wang).

https://doi.org/10.1016/j.jlp.2020.104125
Received 6 June 2019; Received in revised form 26 February 2020; Accepted 30 March 2020
Available online 14 April 2020
0950-4230/© 2020 Elsevier Ltd. All rights reserved.
Y.-J. Lu and C.-H. Wang Journal of Loss Prevention in the Process Industries 65 (2020) 104125

Abbreviations and acronyms LTA local thin area

si longitudinal extent or length increment of metal loss
RSFa allowable remaining strength factor LD1 the longitudinal extent of the LTA model
Ai area of metal loss based on si including the effect of FCA LD2 the circumferential extent of the LTA model
c circumferential extent or length of the region of local metal σ eq;max maximum equivalent stress in the LTA
loss σ eq;mean maximum equivalent stress distant from the defect area
Ec circumferential weld joint efficiency tmin minimal required thickness
CA corrosion allowance L
tmin minimum required thickness based on the longitudinal
CTP critical thickness profile membrane stress for a cylinder
trd uniform thickness away from the local metal loss location tmm minimal measured thickness
established by thickness measurements at the time of the tnom nominal thickness
assessment MAWP maximum allowable working pressure of the undamaged
Lmsd distance to the nearest major structural discontinuity component
σf flow stress PTP parabolic thickness profile
Mt Folias factor based on the longitudinal extent of the LTA LUC the plastic collapse load of the undamaged equipment
Mit Folias factor based on the longitudinal extent of the LTA for LDC the plastic collapse load of the damaged equipment
a through-wall defect for the current subsection being RSF remaining strength factor
evaluated Rt remaining thickness ratio
Q the factor used to determine the length for thickness Ls recommended spacing of thickness readings
averaging MAWPr reduced maximum allowable working pressure of the
FCA future corrosion allowance damaged component
FEM finite element method MAWPrr reduced maximum allowable working pressure if the
tc future corroded wall thickness for areas away from the circumferential effect is included
damaged area λ shell parameter
D inside diameter of the cylinder TTP True thickness profile
s longitudinal extent or length of the region of local metal LOSS the amount of uniform metal loss away from the local
loss metal loss location at the time of the assessment
EL longitudinal weld joint efficiency

selected from commercially available specifications and is in general

LDC
greater than the design thickness (M.M. Hossain and R. Seshadri, 2010; RSF ¼ (1)
LUC
ASME, 2019a,b).
The design thickness is often used as the threshold indicating the Being FFS means that equipment will have sufficient strength until its
point at which equipment ceases to be safe for further service, and the next scheduled inspection (Kaida, T.et al., 2010).
remaining life of the equipment is thus determined using the time The API 579 analysis can be divided into three levels from simple to
required for the thickness to reach tmin. However, this approach is overly sophisticated. However, in Level 3, the analysis mostly involves nu­
simple because it only applies to cases of general thinning. Therefore, a merical techniques (Janelle and J. L., 2005; Witte et al., 2011). The local
more reliable and accurate method is required to correctly predict the thin area (LTA) represents a localized metal loss defect.
actual life of the equipment. Analyzing the effects of an LTA on equipment involves two tasks:
The American Petroleum Institute (API) Recommended Practice 579 managing the defect profile and obtaining the RSF. In Level 1, the defect
(API 579) was first published in 2000 and widely adopted immediately area is treated as a flat rectangular region cut out from the shell, whereas
(Janelle, 2005). In 2007, the second version of API 579 became a joint in Level 2, the defect area is rearranged and must be first assessed from
industrial standard with ASME (Attia et al., 2016). According to C. M. the longitudinal side view, followed by the circumferential side if
Holtam et al. (2011) and Tahara, T. et al. (2015), the API 579 has necessary.
become the nearly de facto approach for analyzing the remaining For both Level 1 and 2 assessments, the LTA decision is based on the
strength of equipment with defects. To ensure equipment status and longitudinal RSF, and the circumferential to longitudinal aspect ratio of
prevent unnecessary resource waste, assessment of equipment fitness the defect must be lower than 2.0. The minimum measured thickness
through the timely collection of available data and validation of previ­ (tmm) in the LTA must be larger than the minimum required thickness in
ous work are essential (Chalmers and Stewart, 2018; Al-Dojayli, M. the longitudinal direction (tLmin). This practice may sometimes be erratic
et al., 2018; Anderson and T. L., and Osage, D. A., 2000). Takuyo Kaida because two defects that coincidentally have nearly identical longitu­
(2008) noted that the API 579 can be used to evaluate the reliability of dinal lengths but differ in circumferential width can have equal RSFs.
components with local thin area defects. It provides simple and practical The RSF is crucial in assessing the fitness of a structure with a defect
approaches for assessing the effect of structural defects on equipment and is highly dependent on the characteristics (i.e., shape and geometry)
strength (Szary, T., 2008; Dipl.-Ing. and Tomasz Szary, 2006). of the defect.
In assessing the FFS of equipment with a local thin area defect, the Moreover, in contrast to the accepted assumption that using rear­
remaining strength factor (RSF) must be greater than the allowable RSF ranged critical thinness profile (CTP) readings instead of the actual
(denoted RSFa) (Kaida, T.et al., 2010) rather than the conventional defect profile yields conservative results, we observed that using a
standard of tmin as the minimum acceptable thickness. The RSF was rearranged thickness profile may smooth out rough edges, in certain
proposed by Sims and Hantz (Sims, 1992). When the RSF is lower than circumstances, resulting in misjudgments of equipment fitness. Under­
RSFa, load adjustment is required if service is to continue or for calcu­ standing the effects of circumferential and longitudinal dimension on
lating the remaining life of the equipment (API 579-1/ASME FFS-1, remaining strength requires the actual LTA profile and detailed analysis.
2016; Szary, T, 2008). When equipment is repaired, reassessment of Although full-scale stress analysis of equipment is ideal because it
its strength is recommended (Yamamoto et al., 2013). considers as many conditions as possible, it is not preferred because it is

2
Y.-J. Lu and C.-H. Wang Journal of Loss Prevention in the Process Industries 65 (2020) 104125

time-intensive and unlikely to become general practice (Cai et al., investigated the following topics:
2017).
Numerous studies have addressed this topic. Kim et al. (2009) per­ 1. Effect of defect dimensions on RSF value: Six defect models with
formed the finite element method (FEM) analysis on an elbow with an identical depth and longitudinal length but different circumferential
internal LTA. Kim (2006) used a one-quarter symmetrical model with width were prepared. FFS Level 2 and the FEM were performed on
manmade thinning on the exterior of a straight line pipe. Xue (2015) the six defects, and their subsequent RSFs were compared. Addi­
investigated the creep damage behavior of a line pipe with an external tionally, to determine whether a change in defect size affects the
thinning defect. Tahara (2003); Tahara et al. (2015) performed an FEM actual RSF value because of stress concentration, the stress concen­
analysis and compared the results between one-quarter symmetrical and tration factors of each LTA at the PCL were calculated and compared.
full models. The results obtained using the FEM analysis were close to Furthermore, to exclude significant stress concentration induced by
those obtained from the whole-model stress analysis in real situations. changes in defect thickness profile, the RSF and K values of the de­
Therefore, FEM results under specific conditions can be accepted as true. fects’ parabolic thickness profiles were calculated for comparison.
Diniz et al. (2006) performed an FEM analysis on a corroded pipe to 2. Numerical effects of the thickness-profile-processing approach. The
obtain the burst pressure and validated the results by using an actual true thickness profile (TTP) and two defect simplification techniques
burst pressure test. The results of the two approaches corresponded with (CTP and parabolic thickness profile (PTP)) were applied alongside
each other, indicating that the FEM is suitable for studying defects. the FEM to calculate the RSF and K values of the defects and deter­
Most studies such as Lee et al., 2015; Jin-Weon Kim, 2008; Tan, W mine whether manipulation of defect profile affects the outcomes.
et al., 2012; Xu and Cheng, 2012; Hui and Li, 2010; Yoon, G et al., 2015.,
Blachut and Iflefel, 2011; Abdalla Filho et al., 2014; Chiodo and Rug­ The methodology can be summarized as in Fig. 1.
gieri, 2009; Kolios et al. (2014); Ma et al. (2013); Mokhtari and For clarity, the paper is organized as follows: Section 2 presents the
Melchers, 2018; Saffar et al. (2019) have used a simplified profile with a API 579 Part 5 Level 1 and Level 2 LTA analyses. Section 3 presents a
parabolic or rectangular cutout shape rather than the actual profile, procedure for analyzing the LTA RSF based on the FEM; included are a
which is difficult to use. description of the experimental objects, material setting, submodel
To save on computing time, some studies (Lee et al., 2015; Peng, J. technology, and convergence analysis. Section 4 presents the final sec­
et al., 2011; Duan and Shen, 2006; Bao, S. et al., 2019) have used a tion discusses four items: the results of the mesh convergence analysis,
simplified finite element model with the LTA modeled as symmetric the RSF, and K of the real defect, the RSF and K of the parabolic contour,
geometry and performed stress analysis on one-half or one-quarter of the and the influence of three thickness contours on the RSF. Section 5
model, whereas other studies have used simplified two-dimensional draws the findings of our research and provides a reference for future
(2-D) models. API 579 revisions.
Although the LTA is typically small compared with the equipment
body, it is the weakest area and determines the minimum equipment 2. FFS part 5 assessment of local metal loss
strength. Therefore, detailed stress analysis of a small region housing the
LTA defect is required. Conventionally, stress analysis is performed on a The API, in cooperation with the ASME, established the API 579 for
portion of the vessel containing the defect with proper boundary con­ assessing defects in pressure equipment and piping. API 579 provides a
ditions (Almeida et al., 2014). However, this is insufficient for swift and near-analytical solution for assessing common defect patterns
computing the plastic collapse load (PCL), because large deformations in pressure vessels, tanks, and pipelines. Further processing principles
extend outward beyond the finite element model boundary due to load are provided when equipment is deemed unfit for continued operation.
steps continuing to accumulate in the plastic zone. Hence, the optimal In FFS assessment, equipment must be considered able to retain
compromise is separating the process into two stages: using a main enough strength until its next inspection (Koçak, M., 2007).
model of the defect-free vessel and a submodel containing the LTA
defect. The SUBMODEL function in ANSYS is applicable for this purpose 2.1. Measurement and sizing of the LTA
if the boundary conditions are established correctly (Diamantoudis and
Kermanidis, 2005). The longitudinal length (s) and circumferential length (c) of the
Janelle and J. L. (2005) found the triaxial stress increases as the defect area determine the size of the LTA. To determine a defect’s
toughness of the equipment at the defect zone reduces. However, the thickness profile, thickness measurement data must be obtained using a
triaxiality disappears as the LTA is modeled as a 2-D shape and the set of evenly spaced grid points within the LTA. After obtaining all the
triaxiality disappears and hence the strength of equipment with a defect thickness data, the thickness profile of the entire defect area can be
will be underestimated. established, and this profile is essential for the overall assessment. Grid
In conclusion, the LTA is irregular (Cosham, A., and Hopkins, P., spacing can be established using the following equation:
2003) and asymmetric and is the weakest location in a piece of equip­ h pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi i
ment. The LTA model should not be replaced with a simplified shape LS ¼ min Q Dðtnom FCAÞ ; 2trd (2)
when the remaining strength of the equipment is being assessed.
Therefore, FEM analysis should be based on a three-dimensional (3D) where Q is a function of Rt and RSFa and is represented as
finite element model as close to the actual profile as possible for both the 8 "� �2 #0:5
>
vessel and LTA. >
< 1:123 1 Rt
1 Rt < RSFa
A finite-element-based submodel technique was employed in this Q¼ 1 Rt =RSFa (3)
>
>
study to calculate both the main model and that of an LTA. Finite :
50 Rt � RSFa
element analysis was performed for various LTA models with identical
thickness profiles and longitudinal lengths but different LTA circum­ An LTA and its grid are illustrated in Fig. 2.
ferential extents. The resulting stresses were then further processed to
obtain the RSF for comparison as well as the stress concentration factor 2.2. Level 1 assessment
(K). In addition, two commonly used simplified LTA types, namely those
with critical and parabolic thickness profiles, were employed to deter­ In Level 1 and 2 assessments, the profile of the LTA is replaced with a
mine the RSF and K. simplified representation in both the longitudinal and circumferential
An FFS test was then performed on a cylindrical pressure vessel with directions rather than the actual 3D region being used. In the Level 1
an internal LTA far from any structural discontinuity. This study calculation, instead of using the whole-plane data, one minimal

3
Y.-J. Lu and C.-H. Wang Journal of Loss Prevention in the Process Industries 65 (2020) 104125

Fig. 1. Flowchart of the analysis method.

operation only when it passes the assessment criteria in both the lon­
gitudinal and circumferential directions, as illustrated in Fig. 3.

2.3. Level 2 assessment

What differentiates Level 2 from Level 1 assessment lies in RSF where

two-dimensional thickness profile both longitudinally and circum­
ferentially formed by the multi-point minimum thickness value under
each inspection plane for the representative contour of the defect area.
This is the so-called critical thickness profile, CTP, of the LTA. For cy­
lindrical and conical pressure components, the process is begun by
rearranging the longitudinal defect profiles by placing the smallest value
in the middle and sequentially placing the remaining values outward.
Therefore, tmm is in the middle, and si is the width of the cross-hatched
area as in Fig. 3. The computing procedure is begun from the center

Fig. 3. Level 1 FFS assessment of an equivalent rectangular defect, shown by

the solid lines.
tmm FCA
Rt ¼ � 0:2 (4)
tC

tmm FCA � 2:5mmð0:1inchesÞ ðfor vessel&tanksÞ

(5)
Fig. 2. LTA and its grid area on the surface of a piece of equipment. tmm FCA � 1:3mmð0:05inchesÞ ðfor pipingÞ
pffiffiffiffiffiffiffi
Lmsd � 1:8 Dtc (6)
measured thickness data point is used for each direction. In the case of a
cylindrical shell, the Level 1 assessment is begun by calculating the Rt
minimal measured thickness reading in the longitudinal plane, after RSF ¼ 1
(7)
1 Mt
ð1 Rt Þ
which the minimal measured thickness reading in the circumferential
plane is calculated. Equipment with an LTA is suitable for continued

4
Y.-J. Lu and C.-H. Wang Journal of Loss Prevention in the Process Industries 65 (2020) 104125

by calculating the ratio of the corroded region to the whole si segment as 3. Analysis planning
one RSF. From the center outward, a list of RSFs is then obtained with
segment lengths increasing by si in each iteration, as illustrated in Fig. 4. Three experiments were performed in order to verify how changes in
The smallest value in the list is then adopted as the final RSF. LTA circumferential length affect the RSF and to realize the effects of the
The RSF formula is expressed as follows: thickness profiles on the analysis results. These include (i) effects of the
2 � � 3 width of the TTP defect on the RSF and K; (ii) effects of the width of the
1 Ai
Aio
PTP defect on the RSF and K; and (iii) Effects of TTP, CTP, and PTP on
6 7
RSF ¼ min64 RSF i ¼ � �7
5 Aio ¼ si tc (8) the RSF and K.
1
1 Mi Ai Ai In the first experiment, this defect was fixed at a constant longitu­
dinal length and depth profile based on actual thickness data. The
t o

The RSF values obtained in Level 1 or Level 2 assessment must be circumferential length was then adjusted, resulting in six local defects
compared with RSFa, which is 0.90 according to the API 579. with different circumferential widths. The FEM was employed to
If the strength of the defect area is insufficient to sustain internal calculate the RSF values of the local defects, which were compared with
pressure before the next inspection, then the equipment fails to pass the those obtained using FFS Level 2. In addition, stress concentration fac­
criteria. If the equipment requires continual service, it should be rerated tors at the LTA were calculated for subsequent comparison. Results were
according to the following equation: then compared with those obtained using FFS Level 2 since in Level 2
they all have the same RSF value. In the second experiment, the focus is
MAWPr ¼ MAWP
RSF
(9) on analyzing effects when the width of the PTP defect is changed on the
RSFa RSF and K. The solid models of the PTP were then formed by curve
If the longitudinal RSF meets the criteria, the defect is most likely fitting with the previous six TTP models. The final experiments discus­
acceptable because stress in the longitudinal cross-section of the cylin­ sing the effects of different thickness profiles on the RSF and K, the TTP
drical structure is nearly twice that in the circumferential cross-section, and two simplified profiles (CTP and PTP) of LTA models were all
which is why the circumferential strength in the longitudinal cross- analyzed with FEM.
section should be assessed first, after which the longitudinal strength The FEM was performed using ANSYS Workbench v.15.0 (Lawrence
in the circumferential cross-section can be assessed if necessary. If the and K. L., 2011). DesignModeler was used to establish the geometrical
circumferential extent c of the defect meets the condition model, and the stress analysis process was performed on Workbench.
� �
EL
c � 2s (10) 3.1. Equipment parameters and the FEM model
EC

and the circumferential strength meets the criterion, verifying the The main equipment model analyzed in this study was the thin cy­
remaining longitudinal strength is unnecessary. lindrical shell of a pressure vessel. Parameters of the main model are
Furthermore, if the circumferential extent c of the defect is large and listed in Table 1.
exceeds 2s(EL/EC), further evaluation of strength in the longitudinal The pressure vessel was modeled as a cylindrical shell with an
L
direction is required becausetmin � tmm FCA. If the defect fails the elliptical head at both ends. The vessel was vertically placed and fixed at
evaluation of longitudinal strength, the pressure vessel must be further the bottom.
de-rated according to the following equation: The entire vessel (6096 mm in length and 762 mm in diameter) was
� � then converted into a finite element model of 10,171 elements. The main
MAWPrr ¼ MAWPr �
tmm FCA
(11) model used a SHELL 181 element type in ANSYS (Cai et al., 2017).
L
tmin

where MAWPrr is the twice-reduced maximum allowable working 3.2. LTA model dimensions and parameters and FEM preprocessing
pressure.
For a spherical structure, the process needs to be performed just once The LTA profile is composed of a 10 � 10 grid area covering the
because the stress in the longitudinal direction is approximately equal to defect area with a total of 11 � 11 ¼ 121 thickness readings, as displayed
that in the circumferential direction. in Table 2. In this case, the LTA profile was obtained from an actual LTA
on the pressure vessel described in Table 1.
The circumferential and longitudinal spacing between adjacent
measurement points were both 31.75 mm (1.25 inches). Additional
parameters used in FFS included LOSS ¼ 2.54 mm, FCA ¼ 2.54 mm for a
two-year inspection interval, and Lmsd ¼ 1524 mm.
As previously mentioned, five additional models based on the same
set of thickness data were generated, each with a different circumfer­
ential length as shown in Fig. 5. Dimensions of all the six LTAs are listed
in Table 3.

Table 1
Equipment parameters.
main model

parameter Unit

Nominal thickness mm 31.75

Inner diameter mm 762
Equipment high mm 6096
tmin mm 25.40
Design Pressure MPa 3.93
o
Design Temperature C 343
Fig. 4. Subdivision process for determining the RSF (API 579-1/ASME
Material specification SA-516 Grade 70
FFS-1, 2016).

5
Y.-J. Lu and C.-H. Wang Journal of Loss Prevention in the Process Industries 65 (2020) 104125

Table 2
Measurement thickness data in the LTA zone.
mm C1 C2 C3 C4 C5 C6 C7 C8 C9 C10 C11

M1 29.21 29.21 29.21 29.21 29.21 29.21 29.21 29.21 29.21 29.21 29.21
M2 29.21 25.40 25.91 26.67 24.89 25.40 24.64 24.13 24.64 25.40 29.21
M3 29.21 24.64 24.89 23.37 22.86 23.37 24.13 23.37 24.13 24.38 29.21
M4 29.21 25.40 24.13 22.10 22.10 22.61 23.62 22.86 23.37 25.40 29.21
M5 29.21 24.38 23.62 21.59 21.84 21.84 22.86 22.10 22.86 24.38 29.21
M6 29.21 25.91 23.37 21.34 22.10 22.35 21.08 22.61 23.37 25.91 29.21
M7 29.21 24.89 23.88 21.84 22.61 22.10 21.59 22.86 22.86 24.89 29.21
M8 29.21 25.40 24.13 23.88 21.84 22.61 23.37 23.88 23.37 25.40 29.21
M9 29.21 24.38 23.37 23.37 22.10 23.37 24.64 22.86 24.13 24.38 29.21
M10 29.21 25.40 24.64 24.13 23.37 24.89 25.40 26.67 25.40 25.40 29.21
M11 29.21 29.21 29.21 29.21 29.21 29.21 29.21 29.21 29.21 29.21 29.21

Fig. 5. Illustration of the LTA models.

Table 3
Dimensions of defect models with various circumferential widths.
Model # unit LTA-I LTA-II LTA-III LTA-IV LTA-V LTA-VI

Defect area (S � C) mm � mm 317.5 � 508 317.5 � 381 317.5 � 317.5 317.5 � 254 317.5 � 190.5 317.5 � 127
Illustration

If all six LTA models were put through the FFS Level 2 analysis, all The submodel dimensions were carefully established to be a few
would have the same RSF of 0.90, in this special case. According to API times larger than the defect, as illustrated in Fig. 7, where LD1 is
579 FFS level 2 criterion for the LTA defect, the pressure vessel is right at 2550.16 mm and LD2 is 990.6 mm.
the brink of failure, two years from now. Therefore, the RSF must be An internal view of the solid models is as shown in Table 4.
accurate so that a proper maintenance strategy can be made.
In general belief, these six LTAs should have different RSFs, and the 3.3. Comparing rearranged and simplified defect models with the actual
LTA with the largest volumetric metal loss (LTA-I in this case) should profile model
have the smallest RSF, and hence be the weakest.
A suitable means of resolving the problem is to subject the actual The TTP within the LTA is typically jagged, not smooth and the
thickness profiles of all six LTA models in Table 3 to rigorous finite weakest point of the structure is usually within the LTA. However, in
element analysis so that the actual strengths can be compared. previous studies, the LTA was often modeled using simplified geometry,
In FFS, the RSF is used to represent a vessel’s strength. The RSF is the such as a parabola (PTP)previous studies, the LTA was often modeled
PCL of the damaged component divided by the PCL of the undamaged using simplified geometry, such as a parabola (PTP). In the Level 2
component. The PCLs of both the undamaged vessel and vessel with analysis of API 579, the 3-D LTA is replaced by two 2-D defects each
defects were required for calculating the RSF. Focusing on the FEM governed by a CTP; one longitudinal and one circumferential. This study
model of the LTA alone is impossible without including the main model. investigated the effects of a CTP and PTP on the RSF and possible stress
PCL computation requires a continual iterative process of adding loads concentration phenomena.
and performing stress analyses. The CTPs of an LTA consisting of two individual profiles (circum­
The local defects were modeled and analyzed in an FEM tool called a ferential and longitudinal) are displayed in Table 5. In this particular
submodel, in which a portion of the vessel containing the defect is cut study, all six LTA models share the same longitudinal and circumfer­
out from the rest of the vessel and modeled differently (Diamantoudis ential CTPs except for the circumferential spacings among them.
and Kermanidis, 2005; Dmitrieva et al., 2018) as shown in Fig. 6. The solid models of the parabola shaped LTA(PTP) were then formed

6
Y.-J. Lu and C.-H. Wang Journal of Loss Prevention in the Process Industries 65 (2020) 104125

plastic zone, thus requiring a nonlinear material model. The maximum

tolerable stress for PCL calculation in the FEM was established as the
flow stress of the vessel material (ASME, 2019a,b).
Among popular material models, the MPC model was selected
instead of the Ramberg–Osgood model because it provides a more
realistic stress-strain relationship and is more suitable for nonlinear
analysis (Kamaya, M., 2014). The MPC model is also preferred by the
API 579 (API 579-1/ASME FFS-1, 2016).
During finite element analysis, the nonlinear MPC stress-strain curve
was modeled as a trilinear stress-strain relationship, as depicted in
Fig. 10. The first line extends from the origin to the yielding point, and
the second line extends from the yielding point to the flow stress point,
marked A in Fig. 10.
The equipment was made of SA 516 Grade 70 carbon steel, and the
design temperature was 343 � C. The material properties are presented in
Table 6. The flow stress (σf) is defined as the mean of the yield and ul­
timate strength (API 579-1/ASME FFS-1, 2016; ASME, 2019a,b;
Almeida et al., 2014).

3.6. Boundary conditions

3.6.1. Main model load steps

There is no single, one-step approach to calculating the PCL. Loads
Fig. 6. Relationship between the main model and defect submodel. are incrementally applied to the finite element model for stress analysis,
and the PCL is the load at which the von Mises stress of the structure
reaches the maximum tolerable stress.
The initial stage of the load-displacement curve is linear. At this
stage, a large load increase ΔP can be used to reduce the calculation
time. As the load is increased and reaches the ultimate load of the
structure, continued excessively large increases cause an inaccurate
result. Therefore, when approaching the burst pressure, a small ΔP must
be used to ensure calculation convergence and curve smoothness.
The load step increment was reduced as the pressure approached the
PCL, which is expressed as (12) (Takuyo KAIDA, 2009) and equals
13.0375 MPa in the aforementioned cylindrical pressure equipment.
LUC ¼ ð2tc = DÞ � σf (12)

3.6.2. Submodel analysis and setting the boundary conditions

After the computation cycle on the main model was complete, the
computation on the submodel began at each load step, with the
displacement values at the cutting boundary passing from the main
model to the submodel.
This process proceeded iteratively until the maximum stress in the
submodel reached the flow stress, and the computing process concluded
as described in Fig. 11.

3.7. LTA stress concentration

Fig. 7. Submodel with the LTA defect at the center. The LTA is narrow and wedge-like, and hence stress concentration
effect should not be neglected. The stress concentration factor K is
by curve fitting with a parabolic curve. expressed as follows (Pilkey, 2008):
σeq;max
K¼ (13)
3.4. Mesh submodel σ eq;mean

To improve the accuracy of the FEM, the submodel had to be divided 3.8. Convergence analysis
into elements starting from its borders and gradually intensifying until
the defect center was reached. In this study, each LTA submodel con­ To ensure numerical convergence, a relative error method was used
tained 14,400 elements, whereas the same area in the main model to ensure that the element size was suitable and the PCL value con­
contained only 682 elements, as illustrated in Fig. 8 and Fig. 9. The verges. In this study, the convergence criterion was set to be within 1%
submodel used a SOLID 186 element type in ANSYS. discrepancy.

3.5. The material model for PCL calculation

According to the ASME VIII Division 2 design by analysis re­

quirements, the PCL must be calculated for the elastic and into the

7
Y.-J. Lu and C.-H. Wang Journal of Loss Prevention in the Process Industries 65 (2020) 104125

Table 4
Detailed internal view of the LTA solid models.
No. defect No. Defect

LTA-I LTA-IV

LTA-II LTA-V

LTA-III LTA-VI

Table 5
Longitudinal and circumferential CTP thickness values.
Unit 1 2 3 4 5 6 7 8 9 10 11

Longitudinal mm 29.21 24.13 22.86 22.10 21.59 21.08 21.59 21.84 22.10 23.37 29.21
CTP
Circumferential mm 29.21 24.38 23.37 21.34 21.84 21.84 21.08 22.10 22.86 24.38 29.21
CTP

4. Results and discussion the RSF. However, in FFS Level 2, all six defects had an RSF of 0.9. In
contrast to the common belief that the FFS technique provides a con­
4.1. Mesh independence test servative estimation of the remaining strength of pressure equipment, in
four of the six cases, the level approach overestimated the RSF by a small
The mesh independence test was first performed for the submodel margin.
containing the LTA, and the independence test results are illustrated in Table 8 presents the K values calculated using the FEM and PCLs of
Fig. 12. The horizontal axis is the number of elements in the defect the various defect submodels.
submodel; the vertical axis on the left-hand side is the relative error in Unsurprisingly, a longer and narrower defect such as LTA-VI had a
the two calculations, whereas that on the right-hand side is the LDC of higher stress concentration factor of 1.16, as indicated in Table 8.
each iteration. Rearranging the thickness profile in the CTP strengthened this effect.
To obtain the optimal PCL while maintaining computing efficiency, Therefore, caution must be taken when assessing a slender local metal
we tested several element models for the submodel. The experiment was loss defect because both geometry and stress concentration effects cause
begun using a 1600-element model and then progressed to submodels the computed estimation to deviate greatly from the real condition of the
with 10,000, 14,400, and 19,600 elements (Fig. 12). The relative errors vessel.
in LDC were all within 0.2% of each other. Finally, the 14,400 elements
model was selected for use throughout the study.
4.3. Effects when the width of the PTP defect is changed on the RSF and K

4.2. Effects when the width of the TTP defect is changed on the RSF and K To confirm that the aforementioned findings were not caused by a
substantial stress concentration induced by changes in the defect
The FEM was applied to determine the RSF and K values of six TTP thickness profile, PTPs were used to calculate the RSF and K, the values
defects (LTA-I–VI) to demonstrate the effects of defect shape. of which are presented in Table 9 and Table 10, respectively.
The plastic collapse load and RSF of the cylindrical pressure vessel The results confirmed that the defects formed the profile of an LTA
with each defect are listed in Table 7. along the parabola based on the TTPs of LTA-I to LTA-VI, which were
The data in Table 7 reveal that the larger the defect, the larger was subsequently encoded as LTA-Ip to LTA-VIp, respectively.

8
Y.-J. Lu and C.-H. Wang Journal of Loss Prevention in the Process Industries 65 (2020) 104125

Fig. 10. Trilinear stress-strain curve.

Table 6
Fig. 8. The portion of the main model with an LTA in the center for conversion Material properties (ASME, 2019a,b; ASME BPVC Section II - Materials, 2019).
into the finite element submodel.
SA 510 Grade 70 Material properties

Elastic modulus 178 GPa

Tangent modulus 3.145 GPa
Poisson’s ratio 0.3
Yield stress(σy) 260 MPa
Tensile Stress(σT) 485 MPa

4.4. Effects of TTP, CTP, and parabola on the RSF and K

As the number of measurement points increases, the internal contour

of a defect becomes a concern. The square-shaped LTA-III model was
selected to investigate how the internal contour of the LTA zone affected
the calculations. Three defect profile types—namely the TTP, CTP, and
PTP—were established for LTA-III, and the FEM was employed to
calculate the RSF and K for these profiles, with the results displayed in
Table 11.
Table 11 indicates that when the same defect was subjected to the
FEM, the RSF obtained using the TTP was lower than that obtained using
the CTP and PTP. However, the K value obtained using the TTP was
greater than that obtained using the other two profile types. Although
the K difference was only approximately 1%, if K were converted into
the PCL, the difference would be as high as 0.1 MPa.
Fig. 9. Finite element submodel containing the LTA.
The reasoning for the lower K values for the CTP and PTP than for the
TTP is as follows. Because the CTP was a rearranged version of the TTP,
As indicated in Table 9, the changes in RSF values were equal to the originally angular thickness profile in the TTP was rounded and
those for LTA-III. As a defect narrowed, the RSF decreased but K much smoother in the CTP. This mitigated the stress, enabling the CTP to
increased. This finding suggests that stress concentration was not caused sustain a greater PCL than the TTP could, hence the higher RSF. The PTP
by considerable changes in the defect thickness profile. To further had the same properties as the CTP. Fig. 13 compares the three longi­
confirm that the length-to-width ratio alone is crucial for estimating the tudinal thickness profiles.
RSF, the LTA profile was replaced with a parabolically shaped, circular FFS Level 2 performs well when used to analyze a defect of local
defect with smooth transitions in geometry changes. The most slender metal loss. Care must be taken when the RSF is close to the minimum
defect still had the lowest RSF of 0.883, and the contour inside the defect allowable remaining strength because managing the LTA is crucial, and
did not affect the trend. However, the same RSF of 0.9 was obtained for in some cases, the TTP is recommended.
all six defects when LTA Level 2 analysis was employed. For an LTA with a square-shaped TTP, the RSF obtained when
The RSF is related to both the length and width of a defect. When FFS applying FFS Level 2 was closest to that when using the FEM. As the
Level 2 is applied to a cylindrical object, only the LTA’s longitudinal defect’s circumferential width was decreased, the RSF obtained using
defect is considered. Therefore, if other LTAs have identical longitudinal the FEM decreased, whereas that obtained using FFS Level 2 remained
length and depth but differ in terms of circumferential length, they may the same. Thus, when evaluating a longitudinally long and narrow LTA,
yield the same RSF when using a Level 2 assessment. FFS Level 2 tended to overestimate.
Instead of the true defect, FFS Level 2 uses a rearranged thickness
profile of a defect profile that is deeper in the center and radiates out

9
Y.-J. Lu and C.-H. Wang Journal of Loss Prevention in the Process Industries 65 (2020) 104125

Fig. 11. The procedure of plastic collapse analysis for the main model and submodel.

Fig. 12. Mesh independence convergence test for LTA-III.

Table 7
PCL and RSF of LTA models based on the TTP.
FEM

mainmodel defect submodel

defect size (S � C) – 317.5 � 508 317.5 � 381 317.5 � 317.5 317.5 � 254 317.5 � 190.5 317.5 � 127
No. LTA-I LTA-II LTA-III LTA-IV LTA-V LTA-VI
defect model TIP TTP TTP TTP TTP TTP
Collapse load (MPa) 13.238 12.135 11.997 11.929 11.859 11.514 11.238
RSF 1.000 0.917 0.906 0.901 0.896 0.870 0.849

Table 8
K values of LTA models based on the TTP.
defect size (S � C) 317.5 � 508 317.5 � 381 317.5 � 317.5 317.5 � 254 317.5 � 190.5 317.5 � 127

No. LTA-I LTA-II LTA-III LTA-IV LTA-V LTA-VI

defect model TTP TTP TTP TTP TTP TTP
σeq,max(MPa) 372.496 372.496 372.496 372.496 372.496 372.496
σ eq,mean(MPa) 346.708 342.768 341.586 338.828 328.978 321.099
K 1.074 1.087 1.090 1.099 1.132 1.160

10
Y.-J. Lu and C.-H. Wang Journal of Loss Prevention in the Process Industries 65 (2020) 104125

Table 9
PCL and RSF of LTA models based on the PTP.
FEM

main model defect submodel

defect size (S � C) – 317.5 � 508 317.5 � 381 317.5 � 317.5 317.5 � 254 317.5 � 190.5 317.5 � 127
No. LTA-Ip LTA-IIp LTA-IIIp LTA-IVp LTA-Vp LTA-VIp
defect model PTP PTP PTP PTP PTP PTP
Collapse load (MPa) 13.238 12.100 11.987 11.962 11.860 11.790 11.687
RSF 1.000 0.914 0.906 0.904 0.896 0.891 0.883

Table 10
K values of LTA models based on PTP.
defect size (S � C) 317.5 � 508 317.5 � 381 317.5 � 317.5 317.5 � 254 317.5 � 190.5 317.5 � 127

No. LTA-Ip LTA-IIp LTA-IIIp LTA-IVp LTA-Vp LTA-VIp

defect model PTP PTP PTP PTP PTP PTP
σeq,max (MPa) 372.496 372.496 372.496 372.496 372.496 372.496
σ eq,mean (MPa) 345.723 342.571 341.586 338.857 336.857 333.903
K 1.077 1.087 1.090 1.099 1.106 1.116

is effective if the RSF is not close to the threshold value of 0.9. When the
Table 11
LTA is nearly square-shaped, the results from Level 2 and 3 analyses
RSF and K for three thickness profile models.
agree favorably. When the LTA deviates from the square shape, the
FEM actual RSF starts to exhibit a clear difference from that of Level 2
main model defect submodel analysis.
defect size (S � C) – 12.5 � 12.5 12.5 � 12.5 12.5 � 12.5 Generally, a wider defect is believed to have a lower RSF because it
No. LTA-m LTA-m LTA-IIIp has a larger volumetric loss in the vessel wall. However, our study
defect model TIP CTP PIP demonstrated that narrower defects are actually more dangerous
Collapse load (MPa) 13.238 11.929 11.962 11.962 because of the stress concentration effect. The high stress inside the LTA
RSF 1.000 0.901 0.904 0.904
may be due to large differences between adjacent thickness measure­
K 1.000 1.099 1.090 1.090
ment points. Further RSF computations for LTAs with parabola-shaped
profiles were designed to suppress the effect caused by uneven grid
spacing, and the results indicated that narrow, slender LTAs have lower
RSFs.
The weakest point of equipment with a localized defect is within the
LTA. Research should focus on managing LTA geometry in detail, and
the LTA profile should not be simplified. The actual thickness profile of
the defect is preferred for accurate stress analysis.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.

org/10.1016/j.jlp.2020.104125.

Declaration of interests

The authors declare that they have no known competing financial

Fig. 13. Comparing the three thickness profiles. interests or personal relationships that could have appeared to influence
the work reported in this paper.
toward the borders. This rearrangement smooths the originally angular
thickness profile of the true shape of the defect, thereby alleviating References
stress. This slightly increases the RSF compared with that based on the
original thickness profile. Rather than a more conservative (smaller) Abdalla Filho, J.E., Machado, R.D., Bertin, R.J., Valentini, M.D., 2014. On the failure
pressure of pipelines containing wall reduction and isolated pit corrosion defects.
RSF, in some cases, FFS Level 2 yields a higher RSF. Comput. Struct. 132, 22–33.
Al-Dojayli, M., Chomyn, K., Ghorbani, H., Barriault, P., 2018. Fitness-for-Service
5. Conclusion assessment and Re-rating of flawed alumina feeding vessels. In: TMS Annual Meeting
& Exhibition. Springer, Cham, pp. 49–55.
Almeida, J.I.L., Rodrigues, M.C., Mishina, K.D., 2014. V. Modeling and level 3 fitness-for-
Aging pressure equipment often has structural defects caused by service assessment of a cylindrical pressure vessel with general metal loss in
mechanisms such as corrosion after many years of service, and the ur­ conjunction with the numerical thermal transient analysis [J]. Int. J. Mech. Appl. 4,
80–93.
gent need for a standard technique for assessing remaining strength can
Anderson, T.L., Osage, D.A., 2000. API 579: a comprehensive fitness-for-service guide.
be observed from the popularity of the API-ASME FFS standard. In this Int. J. Pres. Ves. Pip. 77 (14–15), 953–963.
study, all LTA analyses were performed strictly according to API 579. API 579-1/ASME FFS-1, 2016. Fitness-for-service. American Petroleum Institute,
Hence, our study focused on a peculiar characteristic observed when Washington D.C., USA.
ASME, 2019. Boiler and Pressure Vessel Code Section VIII-Rules for Construction of
analyzing an LTA on a cylindrical pressure vessel. Pressure Vessels. Division 1. ASME International, New York. https://www.asme.
In most cases of local metal loss defects, the API 579 Level 2 method org/.

11
Y.-J. Lu and C.-H. Wang Journal of Loss Prevention in the Process Industries 65 (2020) 104125

ASME, 2019. Boiler and Pressure Vessel Code Section VIII-Rules for Construction of Kaida, T., Mogami, Y., Izumi, S., Sakai, S., 2010. The reliability investigation on the
Pressure Vessels. Division 2. ASME International, New York. https://www.asme. assessment criteria for local metal loss based on API 579-1/ASME FFS-1. In: ASME
org/. 2010 Pressure Vessels and Piping Division/K-PVP Conference. American Society of
ASME BPVC Section II - Materials, 2019. Part D - properties (metric). American Society of Mechanical Engineers Digital Collection, pp. 29–36.
Mechanical Engineering. Kamaya, M., 2014. Stress–strain curve estimation procedures for stainless steels based on
Attia, M.S., Megahed, M.M., Darwish, M.A., Sundram, S., 2016. Assessment of corrosion yield and ultimate strengths. Eng. Fract. Mech. 127, 194–210. https://doi.org/
damage acceptance criteria in API579-ASME/1 code. Int. J. Mech. Mater. Des. 12 10.1016/j.engfracmech.2014.06.008.
(1), 141–151. Kim, Yun-Jae, 2006. Net-section limit load approach for failure strength estimates of
Bao, S., Liu, Y., Mao, J., Ge, R., Li, X., 2019. Numerical and experimental investigation on pipes with local wall thinning. Int. J. Pres. Ves. Pip. 83, 546–555.
limit load of elbow with local thinning area. Int. J. Pres. Ves. Pip. 172, 414–422. Kim, J.W., Lee, S.H., Park, C.Y., 2009. Experimental evaluation of the effect of local wall
Blachut, J., Iflefel, I., 2011. Analysis of pipes containing plain and gouged dents. Strain. thinning on the failure pressure of elbows. Nucl. Eng. Des. 239 (12), 2737–2746.
Cai, Jie, Jiang, Xiaoli, 2017. Residual ultimate strength of offshore metallic pipelines Kolios, A., Srikanth, S., Salonitis, K., 2014. Numerical simulation of material strength
with structural damage–a literature review. Ships and Offshore Structures. deterioration due to pitting corrosion. Procedia CIRP 13, 230–236.
Cai, J., Jiang, X., Lodewijks, G., 2017. Residual ultimate strength of offshore metallic Koçak, M., 2007. FITNET fitness-for-service procedure: an overview. Weld. World 51
pipelines with structural damage–a literature review. Ships Offshore Struct. 12 (8), (5–6), 94–105.
1037–1055. Lawrence, K.L., 2011. ANSYS Tutorial Release 13. SDC publications.
Cai, J., Jiang, X., Lodewijks, G., Pei, Z., Wu, W., 2018. Residual ultimate strength of Lee, G.H., Pouraria, H., Seo, J.K., Paik, J.K., 2015. Burst strength behaviour of an aging
damaged seamless metallic pipelines with metal loss. Mar. Struct. 58, 242–253. subsea gas pipeline elbow in different external and internal corrosion-damaged
Chalmers, F., Stewart, T., 2018. Using the right integrity management tool for the job. In: positions. Int. J. Naval Architect. Ocean Eng. 7 (3), 435–451.
SPE International Oilfield Corrosion Conference and Exhibition. Society of Ma, B., Shuai, J., Liu, D., Xu, K., 2013. Assessment on failure pressure of high strength
Petroleum Engineers. pipeline with corrosion defects. Eng. Fail. Anal. 32, 209–219.
Chiodo, M.S., Ruggieri, C., 2009. Failure assessments of corroded pipelines with axial Mokhtari, M., Melchers, R.E., 2018. A new approach to assess the remaining strength of
defects using stress-based criteria: numerical studies and verification analyses. Int. J. corroded steel pipes. Eng. Fail. Anal. 93, 144–156.
Pres. Ves. Pip. 86 (2–3), 164–176. Peng, J., Zhou, C.Y., Xue, J.L., Dai, Q., He, X.H., 2011. Safety assessment of pipes with
Cosham, A., Hopkins, P., 2003. The assessment of corrosion in pipelines–Guidance in the multiple local wall thinning defects under pressure and bending moment. Nucl. Eng.
pipeline defect assessment manual (PDAM). International Colloquium Reliability of Des. 241 (8), 2758–2765.
High Pressure Steel Pipelines, pp. 1–30. Pilkey, 2008. Peterson’s Stress Concentration Factors, second ed. John Wiley, New York.
Diamantoudis, A.T., Kermanidis, T., 2005. Design by analysis versus design by formula of Saffar, A., Darvizeh, A., Ansari, R., Kazemi, A., Alitavoli, M., 2019. Prediction of failure
high strength steel pressure vessels: a comparative study. Int. J. Pres. Ves. Pip. 82 pressure in pipelines with localized defects repaired by composite patches. J. Fail.
(1), 43–50. Anal. Prev. 1–14.
Diniz, J.L.C., Vieira, R.D., Castro, J.T., Benjamin, A.C., Freire, J.L.F., 2006. Stress and Sims, J.R., 1992. A basis for the fitness for service evaluation of thin areas in pressure
strain analysis of pipelines with localized metal loss. Exp. Mech. 46 (6), 765–775. vessels and storage tanks. ASME PVP.
Dipl, -Ing, Szary, Tomasz, 2006. The Finite Element Method Analysis for Assessing the Szary, T., 2008. The Finite Element Method Analysis for Assessing the Remaining
Remaining Strength of Corroded Oil Field Casing and Tubing. Fakult€ at für Strength of Corroded Oil Field Casing and Tubing. Doctoral dissertation,
Geowissenschaften, Geotechnik und Bergbau. Universit€atsbibliothek der TU BAF.
Dmitrieva, A.S., Samigullin, G.H., Lyagova, A.A., 2018. Evaluation of stress-strain state Tahara, T., 2003. Fitness-for-Service Assessment for Pressure Equipment in Japan. ASME
of steel cylindrical tank with crack defect using ANSYS software. In: Topical Issues of 2003 Pressure Vessels and Piping Conference. American Society of Mechanical
Rational Use of Natural Resources: Proceedings of the International Forum-Contest Engineers.
of Young Researchers. CRC Press, St. Petersburg, Russia, p. 97. April 18-20, 2018. Tahara, T., Kaida, T., Niimura, M., Sakai, S., 2015. Study on LTA measurement for FFS
Duan, Z.X., Shen, S.M., 2006. Analysis and experiments on the plastic limit pressure of assessment. In: ASME 2015 Pressure Vessels and Piping Conference. American
elbows. Int. J. Pres. Ves. Pip. 83 (10), 707–713. Society of Mechanical Engineers. V01AT01A002-V01AT01A002.
Dumitrescu, A., Zecheru, G., Diniţ� a, A., 2018. Characterization of volumetric surface Takuyo, K.A.I.D.A., 2009. Fitness-For-Service Assessment for Pressure Equipment in
defects. In: Non-destructive Testing and Repair of Pipelines. Springer, Cham, Chemical Plants, vols. 2009-I. SUMITOMO KAGAKU.
pp. 117–135. Tan, W., Zhang, J., Niu, W., Wang, Z., Wang, S., Xiao, Y., 2012. The simulation and
Dupen, B., 2016. Applied Strength of Materials for Engineering Technology. https://core. assessment of compressed natural gas storage well with defects. J. Pressure Vessel
ac.uk. Technol. 134 (6), 061501.
Holtam, C.M., Baxter, D.P., Ashcroft, I.A., Thomson, R.C., 2011. A survey of fitness-for- Witte, C.C., Grossman, B., Ribeiro, D.M., 2011. Avoiding disaster: evolution in integrity
service trends in industry. J. Pressure Vessel Technol. 133, 1–10 (February). and maintenance management. In: 30th International Conference on Ocean.
Hossain, M.M., Seshadri, R., 2010. Simplified fitness-for-service assessment of pressure Offshore and Arctic Engineering, Rotterdam, The Netherlands, pp. 19–24.
vessels and piping systems containing thermal hot spots and corrosion damage. Int. Xu, L.Y., Cheng, Y.F., 2012. Reliability and failure pressure prediction of various grades
J. Pres. Ves. Pip. 87, 381–388. of pipeline steel in the presence of corrosion defects and pre-strain. Int. J. Pres. Ves.
Hui, H., Li, P., 2010. Plastic limit load analysis for steam generator tubes with local wall- Pip. 89, 75–84.
thinning. Nucl. Eng. Des. 240 (10), 2512–2520. Xue, Jilin, 2015. Plastic limit load of Grade 91 steel pipe containing local wall thinning
Janelle, J.L., 2005. An Overview and Validation of the Fitness-For-Service Assessment defect at high temperature. Eng. Fail. Anal. 57.
Procedures for Local Thin Areas (Doctoral Dissertation. University of Akron. Yamamoto, Eiichi, et al., 2013. Guidelines for repair welding of pressure equipment in
Jin-Weon Kim, 2008. Effect of local wall thinning on the collapse behavior of pipe elbows refineries and chemical plants. J. Pressure Vessel Technol. 135.
subjected to a combined internal pressure and in-plane bending load. Nucl. Eng. Des. Yoon, G., Osuna, D., Kwon, O., 2015. Level 3 fitness for service assessment of CUI
238, 1275–1285. damage in a deisobutanizer tower. In: The Twenty-Fifth International Ocean and
Kaida, Takuyo, 2008. Experimental and Numerical Validation of Fitness-for-Service Polar Engineering Conference. International Society of Offshore and Polar Engineers.
Assessment for Cylindrical and Spherical Pressure Vessel With Local Metal Loss.
ASME Pressure Vessels and Piping Conference 48302.

12