You are on page 1of 13

ARTICLE IN PRESS

International Journal of Pressure Vessels and Piping 84 (2007) 749761


www.elsevier.com/locate/ijpvp

Structural integrity evaluation for corrosion in spherical pressure vessels


P. Tantichattanont, S.M.R. Adluri, R. Seshadri
Faculty of Engineering and Applied Science, Memorial University, St. Johns, NL, Canada A1B 3X5
Received 31 May 2005; received in revised form 1 December 2006; accepted 20 December 2006

Abstract
Corrosion is typical of the damage that occurs in ageing pressure vessels and pipelines used in industrial processes as a result of reactive
products inside or harsh environmental conditions on the outside. Structural integrity of such components needs to be evaluated
periodically to establish the continued suitability of the vessels under operating conditions. The present paper develops a method for
Level 2 (as categorized by API 579) tness-for-service (FFS) evaluation of spherical pressure vessels with localized corrosion. The decay
lengths for spherical shells subject to different sizes of corroded areas are calculated based on elastic effects in shells so as to identify the
reference volume participating in plastic action. Lower bound remaining strength factors of spherical pressure vessels containing
corrosion damage are formulated by the application of Muras variational formulation and the ma-multiplier method. Three alternative
design recommendations are given. The effectiveness of the proposed methods is evaluated and demonstrated through illustrative
examples and comparison with Level 3 inelastic nite element analyses.
r 2007 Elsevier Ltd. All rights reserved.
Keywords: Structural integrity; Fitness-for-service; Corrosion; Spherical shell; Decay length; Remaining strength factor; Locally thinned area

1. Introduction
Corrosion is one form of commonly occurring damage in
pressure components used in industrial and other applications. Internal corrosion is caused by corrosive products
inside the pressure vessel. External corrosion can occur in
components exposed to hostile surrounding environments.
The ability to demonstrate structural integrity of an inservice component that has sustained some damage such as
corrosion is termed as tness-for-service (FFS) or integrity
assessment. Structural integrity assessments for components containing corroded areas are necessary to verify the
acceptance of continued service. For the purpose of
evaluation, corrosion spots are usually termed as locally
thinned areas (LTA). From a strength point of view, FFS
can be quantied using the remaining strength factor
(RSF). The RSF is dened as the ratio of the failure load of
the corroded component to that of the uncorroded
component. Failure here implies that a certain predened
limiting criterion is exceeded, and does not necessarily
Corresponding author. Tel.: +1 737 3800; fax: +1 709 4042.

E-mail address: adluri@engr.mun.ca (S.M.R. Adluri).


0308-0161/$ - see front matter r 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijpvp.2006.12.004

indicate physical collapse. Limit criteria for ductile


materials typical for pressure components are commonly
based on limiting stress, maximum strain or displacement.
This implies that the component is rendered unserviceable
long before a state of physical collapse is reached. If the
calculated RSF is greater than or equal to the allowable
RSF, the damaged component can be placed back into
service or allowed to continue as before. If the calculated
RSF is less than the allowable value, the component can be
repaired to accommodate the severity of the operating
environment, or the allowable operating pressure can be rerated until the next scheduled maintenance shutdown.
A majority of the studies in structural integrity assessments for pressurized components containing LTA concentrated on the evaluation of piping and to some extent
cylindrical vessels. The most widely used criteria for
assessment of corroded pipes are called effective area
methods. The methods include ASME B31G [1], Modied
B31G and PRC RSTRENG [2]. The standard procedures
for FFS evaluations in the oil and gas sector for
pressurized components are from API 579 [3], whose
assessment procedures are in turn based on the ASME
B31G and the RSTRENG criteria.

ARTICLE IN PRESS
750

P. Tantichattanont et al. / International Journal of Pressure Vessels and Piping 84 (2007) 749761

The effective area method was developed from a semiempirical fracture mechanics relationship by Maxey et al.
[4]. The method assumes that the strength loss due to
corrosion is proportional to the amount of metal loss
measured axially along the pipe. This method assumes that
the aw fails when the stress in the aw reaches the ow
stress, sow. The ow stress used in B31G is calculated as
sow 1.1sy, where, sy is the material yield stress. A higher
ow stress is suggested in Modied B31G as sow
sy+10 ksi to compensate for highly conservative results
embedded in ASME B31G. For low strength material
(such as API 5L grade A whose yield stress is 30 ksi), the
ow stress calculated in Modied B31G can be as high as
1.333sy. These stresses are achieved by assuming the
material to attain some strain hardening after yield.
Seshadri [4] proposed a Level 2 FFS assessment for
thermal hot spots in cylindrical shells based on elastic
analysis. The localized effect of discontinuities in cylindrical shells is discussed in detail and the concept of
reference volume is introduced as the kinematically
active portion that participates in plastic action. The
RSF is then calculated based on the ma-multiplier [5] and
the concept of integral mean of yield in Muras variational
formulation [6,7]. Indermohan and Seshadri [8] extended
the application of the assessment methodology proposed
by Seshadri [4] to evaluate corrosion in cylindrical shells. It
is concluded by them that the RSF obtained from the mamultiplier is conservative and comparable with the RSF
from inelastic FEA results. Ramkumar and Seshadri [9]
studied a thicker cylindrical shell with both internal and
external corrosion using similar model and yield criteria.
The conclusions are similar in both the studies [8,9]. It is
also shown that the RSF evaluated by using the ASME
B31G criterion underestimates the effect of corrosion
damage in some cases.
In the case of corrosion in non-cylindrical pressure
vessels, widespread research has not been carried out by
way of technical development or experimental investigation. The LTA assessment procedures for spherical shells
and formed heads in API 579 [3] are based on the
procedures used for cylinders. However, in the Level 2
assessment procedures, the Folias factor developed for
spherical shells is used.
Sims [10] studied local round thin areas (RTA) in
cylinders and spheres using nite element analysis (FEA)
with elastic-perfectly plastic material and a limiting
maximum strain of 2%. An empirical equation by way of
a curve t to the FEA results was developed to give a
conservative assessment of the RSF. The results indicate
that the RSF for cylindrical shell is slightly greater than
that for an equivalent spherical shell. If the Folias factor
for a sphere is used in the API 579 Level 2 assessment, the
same trend occurs. The same observation can also be
validated by the concept of reference volume used in the
present work.
Tantichattanont et al. [11,15] proposed a level 2 FFS
assessment scheme for spherical shells with hot spots. The

decay length for elastic bending effects in spheres was


investigated and applied to the calculation of reference
volume. It was shown that the decay length for a sphere is
much smaller than that for a cylinder with the same inner
radius and thickness. This indicates a reduction in the RSF
of the damaged sphere compared with that of a similar
cylinder. A lower bound RSF approximation for hot spots
in spherical shells was proposed based on the ma-multiplier
and the classical lower bound multiplier. It was also
observed that the recommended RSF for hot spots is likely
to be shape independent. The same argument can be used
for other types of geometries to obtain a simplied level 2
estimation of RSF. The proposed RSF showed higher
conservatism for very small hot spot sizes. The concepts
used in the above study for hot spots (including the results
for decay length) will be used in the current paper to
estimate the RSF for corroded spherical vessels.
2. Corrosion in spherical pressure vessels
When thickness loss due to corrosion occurs in a
pressure component, the LTA would be subject to a larger
deformation due to a reduction in stiffness compared to the
adjacent area. Relatively large bending moments could be
generated at the discontinuities. If the thickness difference
between the two zones is very high, the effect of the
uncorroded zone can be regarded as close to xed end
support where the edge rotations are almost fully
restrained. As the thickness difference decreases, the
uncorroded zone may not be much stiffer than the
corroded zone and the effects of edge bending moments
become less pronounced. The effect of these bending
moments is also related to the size of the corroded zone. If
the corroded zone is very small, the bending effect is
minimal and the stretching action dominates the behaviour
of the damaged zone. As the size becomes larger, bending
increasingly dominates until a certain limit is reached and
drops off once again at very large damage zones.
The effects of these bending actions result in higher
deformation inside the corroded area. Excessive plastic
deformation in a corroded spot can lead to failure over a
localized region of the component. This localized failure is
facilitated by the participation of the surrounding region
that aids in the mobilization of plastic action. The total
volume that participates in such mobilization can be called
the reference volume. Tantichattanont et al. [11,15] used
this concept of reference volume in estimating the RSF for
components with hotspots. The reference volume is equal
to the sum of the (equivalent) volume of damage (such as
the corroded zone) and the volume of the surrounding shell
inuenced by the edge effects. This is the total volume that
is expected to signicantly participate in any plastic action
leading to failure. The inuence zone of the edge effects is
bounded by the decay length of forces or moments at the
edge of the damaged zone. In the present study, decay
lengths have been estimated using elastic solutions for line
forces normal to the surface and edge moments. The decay

ARTICLE IN PRESS
P. Tantichattanont et al. / International Journal of Pressure Vessels and Piping 84 (2007) 749761

length is a characteristic length corresponding to the


component geometry, i.e., shape of the shell and its
dimensions. Larger decay length generally indicates
better energy dissipation of the structure and leads to
higher loading capacity whereas smaller decay length
suggests severe local effects due to the applied forces.
Once the decay length associated with the component
geometry is obtained, the reference volume can be
calculated from the size of the damaged zone. The decay
length is also used to identify the possible interaction of
multiple local damages. If there are two separate damage
zones whose decay lengths (reference volumes) do not
intersect, the two damage zones are completely independent. Otherwise, the interaction effects must be considered.
It must be pointed out that the reference volume simply
identies the region that participates in plastic action. The
failure will still be due to primary stresses even if the
reference volume is identied using bending or other
similar effects.
3. Theoretical considerations
3.1. Decay length for spherical shells
The effects of the localized edge forces have a peak near
the edge point and decay drastically away from that point
in an oscillating pattern. Decay length is dened as the
distance from the applied force (or moment) to the point
where the effect of the force (or moment) is almost
completely dissipated and becomes negligible. In a real
corroded area, the decay length (or angle) is inuenced by
normal loads, edge moments as well as the membrane
forces at the boundary of the corroded spot. The actual
decay angle is likely to be in between the decay angles
obtained for different cases.
The decay length for edge moment can be found by using
the bending moment distribution across the shell due to a
line moment at the edge. For spherical shells with constant
thickness loaded symmetrically, this can be determined
as [12]

Ri
p
M j p Celc sin lc  g
and M y nM j ,
4
l 2
(1)
where, Ri is the inner radius of the sphere, c aj, a is
the angle from the crown to the edge of the sphere, j is the
angle measured from the crown of the sphere (Fig. 1), C
and g are constants, and l is a shell characteristic function
obtained from
l4 31  n2 R=h2 ,

(2)

where, n is Poissons ratio, R is the radius and h is the


thickness
The constant C is the magnication factor for the
functions in Eq. (1), whereas, g corresponds to the phase
shift of the sine function and directly affects the decay
length. The constants C and g for spherical shells with

751

Edge effect distribution

Shell

Applied
internal pressure

 = 

CL
Fig. 1. Edge effect distribution along the shell.

built-in edges subjected to uniform pressure p can be


determined from the end conditions and substitution of
c 0 into Eq. (1). The constant g for xed edges can be
found as


1
1
.
(3)
g tan
1 cos2 a
The angle g has values ranging from 26.51 to 45.01 for a less
than 901.
For spherical shells with edge moments where the
rotations are restrained and the displacements are allowed,
the constant g is equal to zero. If radial displacements and
rotations at the edges of spherical shells are allowed to
move freely, g is equal to 451. In each case, the effects of
edge bending moments can be reasonably considered
negligible after the rst half sine wave of Eq. (1). The
angle c1 at which zero moments rst occur is


1 3p
g .
(4)
c1
l 4
If the conservative value of g equal to zero is applied, the
decay angle based on bending effects cd can be calculated
for n 0.3 as
"
r
 2 #1=4
3p 3p 1
h
h

.
cd
2:140
4l
4 1:47 R
R

(5)

Note that the actual g and hence cd will have a slightly


larger value than above depending on the relative
stiffnesses of the corroded and uncorroded parts. In the
extreme case, the upper limit for the coefcient of 2.14 in
Eq. (5) is 2.85, an increase of 33%.
A similar derivation for decay angle can be carried out if
the shell is subject to radial displacements at the edge. It
can be shown that this decay length is slightly larger than
that due to edge moments. Conservatively, the angle given
by Eq. (5) is adopted in the present study.

ARTICLE IN PRESS
P. Tantichattanont et al. / International Journal of Pressure Vessels and Piping 84 (2007) 749761

752

3.2. Conditions for the dominance of membrane action in


spherical shells

b
rc

The decay length as calculated above is principally based


on bending effects acting on spherical shells. When a
spherical shell is subjected to a pure stretching membrane
action, the decay length is not the same. Owing to the
nature of shell behaviour, the decay length due to
membrane action is much larger than that due to bending
action. Results from nite element analysis (as will be
discussed in Section 4.2) illustrate that for corrosion spots
of very small sizes, the effects of stretching membrane
action dominate the behaviour of the damaged (and
surrounding) area. There is very little bulging. Instead,
the component tries to open up, thereby stretching the
LTA. For this behaviour, the decay angle and hence the
reference volume is much larger. For larger sizes of
corroded area, as mentioned earlier, the effects of bending
action compared to those of membrane action are
increased in dening the reference volume. Eventually, at
a certain size of corrosion, a crossover from dominance
of the stretching effects to dominance of the bending effects
(bulging of the shell) occurs. The size of corrosion dening
the crossover is discussed below using an approximate
analysis.
Consider the limit state when plastic hinges are formed
around the circular corroded area as shown in Fig. 2. If the
corroded spot is not too large, the curved shell can be
approximately considered to be a circular at plate
(Fig. 3a). For a segment of a unit perimeter cut out of
the circular plate (Fig. 3b), the plastic moment can be
calculated by assuming that the uncorroded zone is much
more rigid than the corroded part.
The xed end moments that initiate plastic ow at the
edge of the corroded area can be approximately computed
by equating the plastic moment capacity of the cross
section to the moments from applied internal pressure
(Fig. 3b) as

s=1

rc

Mp
F=p(A)

Fig. 3. (a) Top view of a circular corroded area. (b) Unit perimeter
segment.

where, Mp is the plastic moment of the section, sy is


material yield stress, hc is the remaining thickness after
corrosion, p* is the applied internal pressure, DA is the area
of the segment under consideration, and rc is the radius of
the nearly at corroded area. The area DA for a unit length
of perimeter is computed as
DA
1
rc

or DA .
2
prc 2prc
2

(7)

Eqs. (6) and (7) lead to the expression for the crossover
applied pressure, p*
 
3
hc
n
p sy
.
(8)
2
rc
For a given radius of corroded area rc, the pressure p* can
be calculated. It is postulated that if the design pressure is
smaller than p*, membrane action dominates the behaviour
of the damage and the decay angle is expected to be larger
than cd obtained from Eq. (5), whereas if the design
pressure is larger than p*, bending action dominates the
damage behaviour and cd can be applied in the RSF
calculation. In other words, for a known design pressure of
a component, the size of a damaged spot dening the
threshold to dominance between membrane and bending
effects can be computed. This will be discussed further in
Section 4.3.
4. Structural integrity considerations

sy h2c
rc
Mp
pn DA ,
4
3

(6)

Bulge
rc
internal
pressure, p

Plastic hinge line


Ro

4.1. Remaining strength factors


In the current paper, three RSF based on Muras
extended variational principle, RSFU, RSFa and RSFL,
are used to assess spherical pressure vessels containing
corrosion damage. They can be derived using the following
considerations.
The expression for the integral mean of yield criterion
employing the concept of reference volume can be written
as
Z
h
i
m0 f s0ij j0 2 dV .
(9)
VR

Fig. 2. Formation of plastic hinges along the edge of a circular corroded


area.

The superscript 0 refers to statically admissible stress


distributions that satisfy equilibrium conditions and m0 is
the plastic ow parameter from the associated ow rule.
The deviatoric stress s0ij corresponds to the impending limit

ARTICLE IN PRESS
P. Tantichattanont et al. / International Journal of Pressure Vessels and Piping 84 (2007) 749761

state, where, s0ij m0d s0ij and m0d is the limit load multiplier.
The deviatoric stress s 0ij equilibrates the applied set of loads.
The parameter j0 is a point function that takes on a value
of zero if s0ij is at yield and remains positive below yield.
The Tresca and von-Mises yield criteria can be
expressed, respectively, as


f s0ij m0d se  sy  0 and
f s0ij m0d se 2  s2y  0,

10

where se is the relevant equivalent stress and sy is the yield


stress.
For a component containing corroded areas, the integral
mean of yield using the von Mises criterion can be
expressed by integration of Eq. (9) as,
m0d s0eI 2  s2y  V I m0d s0eD 2  s2y  V D 0,

(11)

753

The third RSFL based on the classical lower bound


multiplier mL is
RSFL

mL
.
m0u

(17)

These three RSFS (RSFU, RSFa and RSFL) as dened in


Eqs. (14), (16), (17), respectively, will be used to calculate
the recommended RSF in the proposed FFS evaluation
procedures for spherical pressure vessels with corroded
areas. Tantichattanont et al. [13] suggested that the RSFs
for pressure vessels with hot spots are non-dimensional and
shape independent. The RSF for any pressure vessel was
shown to be only a function of the geometrical and
material property ratios for a given situation and are
independent of the actual applied load, etc. A Similar
conclusion can be deduced for pressure vessels containing
corroded areas (with some minor modication in material
property ratios based on Muras integral mean of yield
criterion corresponding to corrosion damage). The RSF
for corroded pressure vessels can be expressed in terms of
property ratiospas
set out below. Let

rV VI/VD, rs seI/
seD, and ra 1 rV =1 r2s rV . Then,

where the sufx I refers to the uncorroded (initial state)


region of the reference volume and sufx D refers to the
corroded (damaged state) area, seI is the equivalent stress
in the original pressure component and seD is the
equivalent stress of the LTA.
Rearrangement of Eq. (11) leads to
s
s2y V R
.
(12)
m0d
s2eI V I s2eD V D

Z ra =rs ,

(18a)

RSFU ra ,

(18b)

The Tresca criterion can be applied in a similar manner


and the equation for m0d for corrosion can be written as

RSFa

sy V R
m0d 0
.
seI V I s0eD V D

(13)

Three types of RSF can be derived using the above. The


upper bound RSFU is obtained from the m0d as
RSFU

m0d
,
m0u

(14)

where m0u sy =seI is the upper bound multiplier for the


undamaged vessel and m0d is determined on the integral
mean of yield basis and yield criterion using Eq. (12) or
(13).
The second RSF is calculated based on the improved
lower bound multiplier called ma proposed by Seshadri and
Mangalaramanan [7]. The ma-multiplier developed from
variational concepts in plasticity by using two linear-elastic
FEA is given by the expression
q
2
p
p 3
2
2
2Z

ZZ

1
1

2

ZZ

1

2
5,
p

ma 2m0d 4
Z2 2  5Z 2 2 5
(15)
where, Z m0d =mL . The classical lower bound multiplier
mL is given by m0L sy =seD . Note that Z is independent of
the equivalent stress se. The RSFa is then expressed as
ma
(16)
RSFa 0 .
mu

q
2
p
p 3
2Z 2 ZZ  12 1 2  ZZ  1 2
5,
p
p
2ra 4
Z 2 2  5Z2 2 5
18c

RSFL rs .

(18d)

Note that since internal pressure is considered as the


only primary load, the equivalent stresses in the corroded
and uncorroded areas (seD and seI) are calculated as
PRi/2h and PRc/2hc, respectively, where, Rc is the inner
radius of the corroded segment of the component and hc is
the remaining thickness of the corroded area. Hence, the
ratio rs here equals (hc/h)/(Ri/Rc) which depends only on
the geometry of the component and the damaged area and
not the actual applied load. It can also be observed that
since the inner radius Rc of a pressure vessel containing
external corrosion is the same as the uncorroded inner
radius Ri, rs will only be the ratio of the remaining
thickness and the uncorroded thickness. However, Rc is
slightly larger than Ri for pressure components with
internal corrosion. Thus, rs and consequently RSFL for
internal corrosion will be slightly less than that for external
corrosion for the same thickness loss. In thin shells, this
difference is not prominent and hence the more conservative RSF calculation assuming internal corrosion will be
presented in the current study.

ARTICLE IN PRESS
754

P. Tantichattanont et al. / International Journal of Pressure Vessels and Piping 84 (2007) 749761

=0
Corrosion spot Volume (VD )

hc
Reference Volume (VR = VD+VI )

Adjacent Volume (VI )


h

a

R.

d
=

Fig. 4. Equivalent circular corroded spot and reference volume.

4.2. Reference volume

such local damages can be calculated in various ways as


discussed below.

In the present paper, an equivalent circular shape with


constant thickness loss is used to represent an irregular
shape of corrosion in a spherical shell (Fig. 4).
The volume of the corrosion spot can be computed as
V D pR2o hc 1  cos ja ,

(19)

where Ro is the outside radius of the shell and ja is an


included angle of a corrosion spot.
The adjacent volume surrounding the corrosion where
the shell might participate in plastic action can be
calculated using the uncorroded or original wall thickness
h, and the decay angle cd dened by Eq. (5) as
V I pR2o hcos ja  cos ja cd .

(20)

The reference volume is then expressed as VR VD+VI.


4.3. Recommendations for RSF calculation
In the following discussion, three different kinds of RSFs
are proposed. Each RSF recommendation contains three
parts, each applicable to different sizes of corroded areas.
The decay length for the shell can be used to dene when
the damaged area can be called local. Assume that the
damaged area is large enough such that the edge effects are
completely dissipated near its centre. As far as this central
zone is concerned, the actual size of the damage ceases to
matter, i.e., the failure of this zone will not be inuenced by
any further increase in the size of the damage. For this case,
the angle ja is larger than cd and the RSF is calculated as
if the damage is spread throughout the vessel. This is given
by RSFL in Eq. (18d) and remains a constant with any
increase in the size of the damage. This observation can be
shown to be consistently applicable for different sizes of
vessels and damaged areas using nonlinear FEA. If on
the other hand, the size of the damage is small enough
such that the decay lengths from either side overlap
(jaocd), it can be called local damage. The RSF for

4.3.1. First recommendation for RSF calculation


The RSFa provides a lower bound approximation for the
component with small local corroded spots where decay
lengths from either side overlap signicantly. For large
spots, RSFL is applicable as explained earlier. A linear
interpolation between RSFa and RSFL can be suggested
for intermediate size local corroded spots. The transition
angle from small to intermediate corroded spots is
suggested as
p
(21)
cL .
4l
Fig. 4 shows this recommendation. Other recommendations are possible. Hence this is identied as the rst
recommended RSFr1. It has three parts corresponding to
the sizes of corroded spots identied by the angle ja. For
corroded spots smaller than cL, RSFa is used. Corroded
spots larger than cd as stated in Eq. (5) are considered
global and RSFL is recommended. For corroded spots of
intermediate sizes, the RSFr1 is obtained by linear
interpolation of RSFa at the angle cL, and RSFL (Fig. 5).
In summary, the rst recommendation for the RSFr1
for spherical pressure vessels containing a corroded area
can be calculated as below:
For ja pcL
For cL oja pcd

RSFr1
RSFr1

RSFa ;


1:5  0:637lja RSFa at cL


0:637lja  0:5 RSFL ;

For ja 4cd

RSFr1

RSFL ;
(22)

Although the rst proposed approach provides a lower


bound approximation for the problem of interest, the
results for corroded areas of very small sizes can be shown
to be fairly conservative compared with FEA results. As
discussed earlier, for spherical shells with highly localized
corroded spots, the effects of stretching membrane action

ARTICLE IN PRESS
P. Tantichattanont et al. / International Journal of Pressure Vessels and Piping 84 (2007) 749761
Corr

oded

RSF
RSF

Corr

RSF
spot

size

Linear
variation

oded

1.0

755

spot

size

RSF

Linear variation
RSFL

RSFL
Shell

Shell

L
d

CL

Fig. 5. First recommendation for remaining strength factor RSFr1 for


various sizes of corroded areas.

govern the shell behaviour and the decay angle calculated


from Eq. (5) is sometimes signicantly underestimated.
Since a small decay angle indicates an adverse effect, this
leads to an overestimation of the effect on the components
containing small corroded areas. Usually, very small
corrosion spots do not pose a signicant problem and
can be ignored. Thus, if the conservatism present in the use
of RSFr1 for very small corrosion spots is acceptable, it
can be adopted for regular use. However, if it is not
acceptable, a second recommendation in the form of
RSFr2 can be proposed. This is based on an approximate
calculation of the cross over limit between bending and
stretching decay angles.
4.3.2. Second recommendation for RSF calculation
As discussed earlier, the pressure p* can be calculated for
any given size of corrosion area (Eq. (8)). This pressure is a
function of material yield stress, remaining thickness and
the size of the corroded spot. Comparing p* with the design
pressure pd, we can nd out whether the corrosion area is
small enough such that stretching dominates the real decay
length. The size of the corroded area below which this
happens is given by
s
3sy
h
c
jn f n
,
(23)
Ro 2pd
where j rh Ro is the included angle of corroded spot at
the crossover, rh is the radius of the corroded spot at the
crossover and f* is an empirical safety factor included to
account for the approximation introduce in deriving p*. In
the current paper, the value of f* is chosen conservatively as
0.5. It can be shown that a slightly larger value of the factor
does not signicantly alter the RSF calculation.
The second recommendation, RSFr2, can then be
determined as shown in Fig. 6. First, the RSFa at angle
j* is calculated by using Eq. (16). For small corroded spots

CL

Fig. 6. Second recommendation for remaining strength factor RSFr2 for


various sizes of corroded areas.

where, jaoj*, RSFr2 is computed by linear interpolation


of the known RSF equaling unity at ja 0 (no damage)
and the RSFa at angle j*. For intermediate sized corroded
spots where, j*ojaocd, a linear transition from the RSFa
at angle j* to RSFL at angle cd is applied. For large
corroded spots with ja4cd, RSFL is used as before.
4.3.3. Third recommendation for RSF calculation
It can be seen that both the foregoing recommendations
above have a phenomenological and intuitive basis and are
grounded in a variational formulation in the form of the ma
calculation, etc. They are both lower bound estimates for
the range of parameters considered in the present study.
However, both the recommendations employ linear transitions and are slightly abrupt at the changeover points. A
third recommendation is provided if the user wishes to
avoid these issues. It uses RSFL for large corroded areas as
in the case of the other two recommendations. For smaller
corrosion areas, a parabolic transition is employed between
the limits of RSFr3 1 at ja 0 and RSFr3 RSFL at
ja cd (as shown in Fig. 7). It must be noted that this
recommendation is simpler than the previous two. However, it is also less grounded in theoretical considerations
compared to the rst two recommendations.
The function for RSFr3 can be expressed as below:
For ja pcd ;

RSFr3

L
1RSF
j2a
c2
d

L
 21RSF
jd 1:
c

(24)

For ja pcd ;

RSFr3

RSFL :

4.4. Distortion due to bulging


An additional limit criterion for the shell strength is
based on excessive deformation due to bulging inside the
corroded area of the component. Bulging is the local

ARTICLE IN PRESS
756

P. Tantichattanont et al. / International Journal of Pressure Vessels and Piping 84 (2007) 749761

inward or outward deviation from the original spherical


geometry. Assuming that the bulge is also spherical in
shape, the strain due to bulging (Fig. 7) and the included
angle [6] of the bulge can be shown to be


Rb jb  Ro ja
Ro sin ja
b
and jb arcsin
,
(25)
Ro ja
Rb
where eb is the maximum membrane strain in the bulge. By
limiting eb to the maximum value permitted for the average
strain through the thickness (as dened in the limit criteria
for the component), we can nd Rb and jb from Eq. (25).
Radial displacement of the assumed spherical bulge db is
then given by
q
db Rb  do  R2b  Ro sin ja 2 ,
(26)
where do (1cos ja)Ro.
The out-of-roundness ratio is computed as the ratio of
the maximum diameter of the sphere due to bulging
Dmax( Do+db), and the original diameter of the sphere
Do, as shown in Fig. 8. If the out-of-roundness ratio
Corr

RSF
1.0

oded

spot

size

Parabolic
variation

exceeds a certain recommended value, RSFr should be


taken as RSFL. It may be noted that the calculation above
ignores the rotation restraining effect of the surrounding
region and assumes a hinge support for the corroded zone.
Hence, the bulge displacements calculated from Eq. (26),
although very simple to calculate, tend to be considerably
overestimated.
5. Illustrative example and discussion
5.1. Evaluation of RSF
The following example is given to demonstrate the
proposed Level 2 integrity assessment method in a
spherical shell under internal pressure. The values given
in the parentheses are in imperial system.
ASTM material
Shell inside radius (Ri)
Operating pressure
Design pressure (pd)
Operating temperature
Design temperature
Corrosion allowance (CA)
Joint efciency (Ej)
Material yield stress (sy)

:
:
:
:
:
:
:
:
:

SA 516 Grade 55
0.556 m (21.9 in.)
1.72 MPa (250 psig)
2.70 MPa (390 psig)
32.2 1C (90 1F)
37.8 1C (100 1F)
0.0016 m (1/16 in.)
1.0
206.8 MPa (30 ksi)

RSFL

5.1.1. Required thickness calculations


Design thickness, hd, for the spherical shell can be
determined as [13]

Shell

hd

CL

Fig. 7. Third recommendation for remaining strength factor RSFr3 for


various sizes of corroded areas.

pd Ri
0:00795 m 0:313 in:.
2SE j  0:2pd

The required shell thickness is h hd+CA 0.00953 m


(0.375 in.).
Therefore, a 0.00953 m (3/8 in.) thickness is specied.
Note that, when the rate of corrosion is predictable the
corrosion allowance can be determined by the desired life
of the vessel.

Bulge
b

Shell

b R b

Bulge
Ro
Dmax

Do

CL

Fig. 8. Idealized bulging geometry.

ARTICLE IN PRESS
P. Tantichattanont et al. / International Journal of Pressure Vessels and Piping 84 (2007) 749761

The allowable RSF can be expressed as RSF* hd/h


0.835. Note that this RSF* is based on design requirements.
Outside radius of the sphere, Ro 0.566 m (22.275 in.).
In this case, the R/h ratio is 58.9 and for n 0.3 using
Eq. (2), l 9.865.
The decay angle of spherical shells is calculated from
Eq. (4), cd 13.71.
The decay length measured outside of the sphere is
sd cd  Ro 0.135 m (5.320 in.).
The decay length along the meridian
p of cylindrical shells
suggested by Seshadri [6] is 2:5 Rh while the circumferential decay length is 6.10(R3h)1/4. For a cylinder with the
same diameter and thickness as the example sphere, the
meridional and circumferential decay lengths are 0.183 m
(7.195 in.) and 1.235 m (48.635 in.), which are much larger
than the spherical decay length. This is in line with the
discussion earlier stating that corrosion effects are much
more localized in spheres with very small decay lengths
compared to those in cylinders.
5.1.2. Recommended RSFr
For the following demonstration, the depth of thickness
loss inside the corroded area is assumed to be equal to
h/6 0.0016 m (1/16 in). For internal corrosion, the inner
radius of the shell inside the corroded spot Rc becomes
0.558 m (21.963 in) and the remaining thickness hc is
0.794 m (5/16 in).
The dimensions of the corroded spot are dened with
ja 8.01.
From Eq. (21), cL 4.61.
Since cLojapcd, this is the case of an intermediate
corroded spot. RSFr1 is calculated from the second of
Eq. (22) and RSFr3 is calculated from the rst of
Eq. (24).
The angle for the reference volume is jR ja+
cd 21.71.
From Eq. (19), corroded spot volume (VD)
1.554  104 m3 (9.481 in3).
From Eq. (20), adjacent volume (VI) 11.050  104 m3
(67.433 in3).
Kinematically active reference volume (VR) VD+
VI 12.604  104 m3 (76.914 in3).
The yield stress of steel at a temperature of 37.8 1C
(100 1F) is 207 MPa (30  103 psi).
The primary stress of the undamaged shell is calculated
as seI pdRi/2h 78.9 MPa (11449 psi).
The primary membrane stress inside the corroded spot is
seD pdRC/2hc 95.0 MPa (13778 psi).
Comparison of seI and seD with sy, indicates that there is
no general yielding through membrane action. It must be
noted that seI and seD need not be computed to estimate
RSF values although they serve a useful purpose to
compare with sy.

Using the above, m0u sy seI 2.620 and mL

sy seD 2.177.
From Eq. (17), RSFL 0.831.

757

The RSFU and RSFa calculated based on the Tresca and


von Mises yield criteria are as follows:
Tresca criterion:
From Eq. (13), m0d 2.556.
From Eq. (14), RSFU 0.976.
Z m0d =mL 1.174. From Eq. (15), ma 2.000.
From Eq. (16), RSFa 0.919.
Von Mises criterion:
From Eq. (12), m0d 2.551.
From Eq. (14), RSFU 0.976.Z m0d =mL 1.139.
From Eq. (15), ma 2.406.
From Eq. (16), RSFa 0.918.
Similarly, RSFa calculated using the angle ja cL is
RSFa at the angle cL 0.926.
From the second of Eq. (22), RSFr1 0.887.
From Eq. (23), the angle j* 4.41. Thus, RSFa at the
angle j* 0.926. Linear interpolation from RSFa at the
angle j* and RSFL at the angle cd gives RSFr2 0.887.
From the rst of Eq. (24), RSFr3 0.857.
If the maximum strain in the bulge eb is 1%, from
Eqs. (26), Rb 12 in. (0.305 m) and jb 15.01.
From Eq. (27), db 0.191 in. (4.84 mm). This corresponding value from inelastic analysis is 2.08 mm.
(0.082 in.).
Thus, rout-of-roundness 1.004p1.01.

5.2. Comparison with nonlinear (level 3) analysis


In order to validate the proposed RSF evaluation
methods for spherical pressure vessels with corroded spots,
several nonlinear FEA have been carried out using ANSYS
[14] to determine the inelastic strength. Since the corrosion
damage is approximated as circular, an axisymmetric
model has been used. This can greatly reduce the modeling
and analysis time compared to that of an equivalent 3-D
model. An 8-node ANSYS element (PLANE82) has been
employed. It has compatible displacement shapes and is
well suited to model curved boundaries. Elasticplastic
material behaviour with material yield stress 207 MPa
(30 ksi) is used. The model adopts a plastic tangent
modulus equal to 1.38  108 Pa (20  103 psi) which is
small compared with the elastic modulus of 202.0 
109 Pa (29.3  106 psi). The limit criterion for failure is
chosen to be 1% von-Mises membrane strain. This is
consistent with API 579.
The inelastic RSFi is calculated from the ratio of the
internal pressure that causes 1% membrane strain in the
corroded region to the limit pressure (not design pressure)
of the undamaged sphere. Nonlinear FEA is the most
rigorous assessment method with the current technology
and can be classied as a Level 3 assessment provided that
material model is appropriately chosen. For the example
given above, RSFi is 0.949. It can be seen that all three
recommendations of the current study are below the
inelastic result. Thus, the recommended RSF are acceptable and conservative.

ARTICLE IN PRESS
P. Tantichattanont et al. / International Journal of Pressure Vessels and Piping 84 (2007) 749761

758

Table 1
Remaining Strength Factors (RSF) for corroded spherical shell with
thickness loss of h/6 and R/h 58.9
ja (1)

RSFU

RSFr1

RSFr2

RSFr3

RSFinternal
inelastic

3
5
8
12
15
20
25

0.994
0.986
0.974
0.958
0.948
0.933
0.922

0.928
0.919
0.887
0.843
0.831
0.831
0.831

0.949
0.919
0.887
0.843
0.831
0.831
0.831

0.932
0.896
0.857
0.843
0.831
0.831
0.831

0.993
0.983
0.949
0.896
0.866
0.842
0.834

RSFexternal
inelastic
0.975
0.939
0.891
0.865
0.849
0.835
0.832

6. Results and discussion


The analysis is further applied to the same sphere with
different corroded spot sizes and remaining thicknesses for
two different types of corrosion, internal and external, to
evaluate the effectiveness of the proposed Level 2 methods.
The RSF for the spherical shell with a remaining corroded
thickness of 5h/6 using the von-Mises criterion are as
shown in Table 1.
A different sphere problem with R/h ratio of 20 is
also studied to investigate the effects of shell curvature.
Figs. 911 compare the recommended RSFr1, RSFr2,
RSFr3 and the inelastic RSFi obtained from FEA of
similar spherical shells (with inner radius 21.9 in.) with R/h

1.000

0.900

RSF

0.800

0.700

0.600

0.500

0.400
0.00

0.05

0.10

0.15
0.20
0.25
0.30
Included angle of corroded area a

0.35

0.40

0.45

1.000

0.900

RSF

0.800

0.700

0.600

0.500

0.400
0.00

0.10

0.20
0.30
0.40
Included angle of corroded area a

0.50

0.60

Fig. 9. Comparison of RSFr1 and RSFi of spherical shell with internal and external corrosion for (a) R/h ratio 58.9 and (b) R/h ratio 20.

ARTICLE IN PRESS
P. Tantichattanont et al. / International Journal of Pressure Vessels and Piping 84 (2007) 749761

759

1.000

0.900

RSF

0.800

0.700

0.600

0.500

0.400
0.00

0.05

0.10

0.15
0.20
0.25
0.30
Included angle of corroded area a

0.35

0.40

0.45

1.000

0.900

RSF

0.800

0.700

0.600

0.500

0.400
0.00

0.10

0.20
0.30
0.40
Included angle of corroded area a

0.50

0.60

Fig. 10. Comparison of RSFr2and RSFi of spherical shell with internal and external corrosion for (a) R/h ratio 58.9 and (b) R/h ratio 20.

ratios equal to 58.9 and 20 for both internal and external


corrosion with remaining corroded thicknesses of 5h/6, 2h/
3 and h/2. The RSF are plotted against the included angle
of the corroded area. It must be noted that for the purpose
of plotting the curve for RSFr2, the design pressure must
be specied in order to calculate the crossover angle j*
where pd p*. For the graphs as shown in Fig. 10, the
design pressures applied in Eq. (24) are computed by using
the maximum internal pressure that can be sustained by the
shell, i.e., the design pressures in these gures are backcalculated from the formula for design thickness.
It can be seen from Figs. 810 that the recommended
RSF from all the proposed methods are shown to offer
good lower bound approximations for all cases of thickness
loss and size of corroded area.

As explained above, the three recommendations are all


acceptably low compared to the results of the Level 3
inelastic analysis. The rst two recommendations have a
good grounding in theory and phenomenological understanding but require slightly more calculation. The third
recommendation, while less rigorous, is also simpler to use.
The users can choose one or the other recommendation
depending on their need or preference.
6.1. Use of flow stress
Note that in general, steels used in pressure vessels can
sometimes have a signicant amount of strain hardening.
In level 2 procedures currently in use (such as modied
B31G), the strain hardening aspect is accounted for by the

ARTICLE IN PRESS
P. Tantichattanont et al. / International Journal of Pressure Vessels and Piping 84 (2007) 749761

760

1.000

0.900

RSF

0.800

0.700

0.600

0.500

0.400
0.00

0.05

0.10

0.15
0.20
0.25
0.30
Included angle of corroded area a

0.35

0.40

0.45

1.000

0.900

RSF

0.800

0.700

0.600

0.500

0.400
0.00

0.10

0.20
0.30
0.40
Included angle of corroded area a

0.50

0.60

Fig. 11. Comparison of RSFr3 and RSFi of spherical shell with internal and external corrosion for (a) R/h ratio 58.9 and (b) R/h ratio 20.

use of a ow stress in place of the yield stress. This implies


that we use an elastic perfectly plastic material which yields
at a ow stress instead of at the material yield point. The
calibration for RSF in the present paper has been carried
out by adhering to this concept. It assumes a material
model almost equivalent to elastic perfectly plastic
behaviour. If we use the ow stress in the calculations
proposed above, it will automatically account for any
strain hardening present in the material. This assumption
has been veried for all the three recommended RSF
procedures presented above. Extensive validation of this
and several other aspects of the current study are provided
in [15].

7. Conclusions
In the current paper, three approaches for Level 2 FFS
assessments of spherical pressure vessels containing corroded areas are proposed. A decay angle for spherical
shells based on edge effects is dened as the limit for the
size of what can be called local corrosion damage.
Various aspects of these edge effects have been discussed
using shell theory and simplied analysis. The concept of
reference volume is used in conjunction with a variational
formulation to assess limit strength. The RSF are
recommended depending on the size of corroded area.
For corroded spots larger than the local limit, the

ARTICLE IN PRESS
P. Tantichattanont et al. / International Journal of Pressure Vessels and Piping 84 (2007) 749761

recommended RSF is based on the lower bound multiplier.


For very small corroded spots, the recommendations
involve using the ma-multiplier and integral mean of yield
criterion or a transition between two limit points. For
intermediate size corroded spots, a linear or parabolic
transition is shown to work well. The rst recommendation
is more conservative than the others for very small
corroded spots but is also more theoretically grounded in
an overall sense. The second recommendation involves a
simplied analysis to identify the cross-over between
stretching and bending decay lengths. It uses a linear
transition between applicable points and uses the mamultiplier for the intermediate limit. The third recommendation involves a simple parabolic transition between the
zero corrosion case and the lower bound RSF limit for
large corroded areas. All the three proposed recommendations are veried to provide good lower bound estimation
compared to inelastic FEA results. The users can choose
one of the three recommendations depending on their need
or preference for rigour and the amount of computation.
Acknowledgement
The nancial support provided by the Terra Nova
Owner Group (A Petro-Canada operated Project), Natural
Sciences and Engineering Research Council of Canada,
and the Canada Research Chair program, which made the
research possible is gratefully acknowledged.
References
[1] ASME. Manual for determining the remaining strength of corroded
pipelines. American National Standards Institute (ANSI)/ American
Society of Mechanical Engineers (ASME) B31G. 1984.
[2] Kiefner JF, Vieth PH. A modied criterion for evaluating the
remaining strength of corroded pipe (with RSTRENG). American
Gas Assoc, Catalogue L51609, PR3-805. December 1989.

761

[3] American Petroleum Institute (API). API Recommended practice


579tness for service. Washington, DC. 2000.
[4] Seshadri R. Integrity assessment of pressure components with local
hot spots. ASME J pressure Vessel Technol 2005;127(2):13742.
[5] Seshadri R, Mangalaramanan SP. Lower bound limit loads using
variational concepts: the mamethod. Int J Pressure Vessels Piping
1997;71:93106.
[6] Mura T, Lee SL. Application of variational principles to limit
analysis. Quart Appl Math 1963;21(3):243348.
[7] Mura T, Rimawi WH, Lee SL. Extended theorems of limit analysis.
Quart Appl Math 1965;23:1719.
[8] Indermohan H, Seshadri R. Fitness-for-service methodology based
on variational principles in plasticity. J Pressure Vessel Technol
2005;127(1):927.
[9] Ramkumar B, Seshadri R. Fitness for service assessment of corroded
pipelines based on variational principles in plasticity. J Pipeline
Integrity 2005;2:99116.
[10] Sims JR, Hantz BF, Kuehn KE. A basis for the tness for service
evaluation of thin areas in pressure vessels and storage tanks.
Pressure Vessel Fract Fatigue Life Manage, ASME PVP 1992;
233:518.
[11] Tantichattanont P, Adluri SMR, Seshadri R. Fitness-for-service
assessment of spherical pressure vessels with hot spots. Int J Pressure
Vessels Piping, in press, doi:10.1016/j.ijpvp.2006.12.003.
[12] Timoshenko S, Woinowsky-Krieger S. Theor Plates Shells. 2nd ed.
New York: McGraw-Hill, Inc.; 1970 [chapter 16].
[13] Bednar HH. Pressure vessel design handbook. 2nd ed. NY: Van
Nostran Reinhold; 1985.
[14] ANSYS. University research version, 8.1. SASIP, Inc; 2004.
[15] Tantichattanont P. Fitness-for-service assessment for thermal hot
spots and corrosion in Pressure vessels, PhD Thesis, Memorial
University, St. Johns, NL, Canada. 2006.

You might also like