Proceedings of PVP2008

2008 ASME Pressure Vessels and Piping Division Conference

July 27-31, 2008, Chicago, Illinois, USA

Paper Number: PVP2008-61030

3-D Stress Intensity Factors for Arrays of Inner Radial Lunular or Crescentic
Cracks in Thin and Thick-Walled Spherical Pressure Vessels

By
M. Perl and V. Bernstein1
1

Pearlstone Center for Aeronautical Engineering Studies

Department of Mechanical Engineering
Ben-Gurion University of the Negev
Beer-Sheva 84105, Israel

ABSTRACT parameters: the geometry of the sphere- η, the number of cracks

Some spherical pressure vessels are manufactured by
methods such as the Integrated Hydro-Bulge Forming (IHBF)
method, where the sphere is composed of a series of double
curved petals welded along their meridional lines. Such vessels
are susceptible to multiple radial cracking along the welds. For
fatigue life assessment and fracture endurance of such vessels
one needs to evaluate the Stress Intensity Factors (SIF)
distribution along the fronts of these cracks. However, to date,
only one 3-D solution for the SIF for a circumferential crack in
a thick sphere is available, as well as 2-D SIFs for one through
the thickness crack in thin spherical shells. In the present paper,
mode I SIF distributions for a wide range of lunular and
crescentic cracks are evaluated. The 3-D analysis is performed,
via the FE method employing singular elements along the crack
front, for five geometries representing thin, moderately thick,
and thick spherical pressure vessels with outer to inner radius
ratios of η=Ro/Ri=1.01, 1.05, 1.1, 1.7, and 2.0. SIFs are
evaluated for arrays containing n=1-20 cracks; for a wide range
of crack depth to wall thickness ratio, a/t, from 0.025 to 0.95;
and for various ellipticities of the crack, i.e., the ratio of crack
depth to semi crack length, a/c, from 0.2 to 1.5. The obtained
results clearly indicate that the SIFs are considerably affected
by the three-dimensionality of the problem and by the following
in the array-n, the depth of the cracks-a/t, and their ellipticity-
a/c.
INTRODUCTION
Pressure vessels, mostly cylindrical, are widely
used in many industries. However, due to their optimal
specific strength (strength/weight) and their ease of
packaging, spherical pressure vessels are commonly used for
example as: gas tanks on LNG (liquefied natural gas)
carriers, propellant/oxidizer/pneumatic tanks on spacecrafts
and aircrafts, pressurized storage tanks for chemical
substances, cookers for the food industry, and as metal or
concrete containment structures in nuclear plants.
Furthermore, whenever extremely high pressure occurs,
such as in high explosion containment tanks or in the
apparatus used to manufacture artificial diamonds and other
crystals, spherical pressure vessels is the only feasible
solution. Some of these pressure vessels are manufactured
by methods such as the Integrated Hydro-Bulge Forming
(IHBF) method [1], where the sphere is composed of a
series of double curved petals welded along their meridional
lines see Fig. 1. These pressure vessels are susceptible to
cracking due to one or more of the following factors: cyclic

1
Aaron Fish Professor of Mechanical Engineering-Fracture Mechanics, and graduate
student, respectively.

1 Copyright © 2008 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 08/26/2018 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 pressurization-depressurization, the existence of a heat-
effected zone near the welds, tensile residual stresses within
this region, and the presence of corrosive agents. As a result,
one or more radial cracks develop from the inner surface of
the vessel on the welding planes.
THREE-DIMENSIONAL ANALYSIS
The three dimensional analysis is performed on an
elastic sphere of inner radius Ri, outer radius RO, and wall
thickness t. The sphere contains an array of n identical,
inner, radial, lunular or crescentic cracks of length 2c and
depth a (see Fig. 2). The cracks are on equally-spaced
meridional planes and are symmetric with respect to the
equatorial plane as described in Fig. 2a. A Poisson ratio, υ,
equal to 0.3 is assumed in the analysis representing a steel
sphere.

FINITE ELEMENT MODEL

Due to the various symmetries of the geometrical
configuration, only part of the sphere, in the form of half a
lune, must be analyzed. The planes of symmetry
encompassing the half lune are: the meridional planes,
θ = 0˚, θ=π/n˚ and the equatorial plane φ=0˚ (see Fig.2a).

Fig. 1 A Liquefied Petroleum Gas (LPG) spherical

tank formed by the Integrated Hydro-Bulge
Forming method (IHBF) [1] .
In order to assess the static fracture endurance,
crack growth rate, and the total fatigue life of such a
pressure vessel it is necessary to determine the prevailing
mode I stress intensity factor (SIF)-KI for one or multiple
radial cracks. Only little attention was given so far to this
problem. At first, two-dimensional SIFs for one through-
the-thickness crack for a thin spherical shell were
evaluated by Folias [2], Erdogan and Kibler [3], and Tada
et al. [4] (see page 486). These solutions can only provide
an upper limit to the SIFs. Thus, when one or more radial
cracks emanate from the inner surface of a thick-walled
pressurized sphere, the 2-D SIFs might provide very
conservative estimates for the fracture endurance and the
total fatigue life of the vessel. Presently, the only available
3-D SIFs are for an internal circumferential crack in a thick
pressurized sphere presented by Atsumi and Shindo [5].
This solution is of limited scope.
It is therefore the purpose of the present study to
perform a full three-dimensional analysis of cracked thin
and thick-walled spherical pressure vessels by the finite
element method. Three-dimensional Mode I SIF
distributions along the crack fronts of a variety of both
lunular2 and crescentic3 radial crack array configurations
are to be evaluated for several geometries of the spherical
pressure vessel.

2
A lunular crack is defined as a planar, part-through crack, whose shape is Fig. 2 The multicracked spherical vessel: (a) segment of
enclosed by two circular arcs of different radii, one concave and one convex, which the sphere containing one crack; the parametric
intersect at two points, having an ellipticity of a/c=1 (Fig. 2b).
3 angle ψ defining the points on the crack fronts of
A Crescentic crack is defined as a planar, part-through crack whose shape is
enclosed by two intersecting arches, the concave one which is elliptical, and the
a lunular crack (b), a slender crescentic crack (c),
convex one which is circular, having an ellipticity of a/c≠1(Fig. 2c-d). and a transverse crescentic crack (d).

2 Copyright © 2008 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 08/26/2018 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 Pressure p, is applied to the inner surface of the sphere,
r=Ri, and the internal pressure is assumed to fully penetrate
the cracks cavity acting on both crack surfaces. This model
is solved using the 2007 version of the standard ANSYS
11.0 FE code [6]. To accommodate the singular stress field
in the vicinity of the crack front this area is covered with a
layer of 20-node isoparametric brick elements collapsed to
wedges, forming singular elements at the crack front [7].
On top of this layer at least four additional layers
consisting of 20-node isoparametric brick elements are
meshed. The rest of the model is meshed with both brick
and 10-node tetrahedron elements. Near the crack front the
elements are chosen to be small and their size is gradually
increased when moving away from it. SIFs are extracted
using the crack-face displacement extrapolation procedure
built into ANSYS [6]. The distribution of the SIF is
discrete and is calculated every 1°-3° in the range ψ= 0° to
90°, depending on the crack geometry. For ψ< 0°, the SIF
is calculated for all the cracks at intervals smaller than 2°,
depending on the angle ψ0, which changes from case to
case.
To ensure the accuracy of the results a large number
of convergence tests were performed. The stress intensity
factors at ψ= 0° is used as the convergence criteria. As a
result of these trials it is anticipated that for meshes of more
than 25,000 degrees of freedom (DOF) the level of error
will be less than 2%. Hence, in order to keep good accuracy
while using reasonable computer resources all meshes
herein contain 25,000 to 50,000 DOF. The elements shape
and aspect ratio for all meshes were automatically taken
care off by the software.
VALIDATION OF THE MODEL
The FE model is validated against the only available 3-D
solution by Atsumi and Shindo [5]. Therefore, a full 3-D
analysis of an inner circumferential crack prevailing in a
thick-walled spherical pressure vessel is performed. SIFs
are calculated for various relative crack depths of
a/t=0.025-0.6, and for two sphere geometries with outer to
inner radius ratios of η =Ro/Ri=1.7, and 2.0. The FE results
as well as those of Atsumi and Shindo [5] are presented in
Fig. 3.
The results are in excellent agreement for shallow
cracks, as could have been anticipated. As the crack gets
deeper the deviation between the results in [5] and the FE
results becomes larger, with a maximum of about 6%. As
the accuracy of the results is not addressed in Atsumi and
Shindo [5], and as the FE results are assumed to be a lower
bound for the "real" SIFs, it seems that the present analysis
yields very good results and validates the model.
3 Copyright © 2008 by ASME
Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 08/26/2018 Terms of Use: http://www.asme.org/about-asme/terms-of-use
Fig. 3 SIFs for an inner circumferential crack in a
pressurized spherical vessel
RESULTS AND DISCUSSION
In order to determine the influence of the various
parameters on the interaction between cracks in the array
and hence on the prevailing SIFs distributions, a
considerable number of cases are solved. SIFs for radial,
lunular or crescentic crack arrays containing n=1, 2, 4, 8,
16 and 20 cracks; with ellipticities a/c=0.2, 0.5, 1.0, and
1.5; and ratio of crack-depth to wall-thickness a/t=0.025,
0.01, 0.05, 0.1, 0.2, 0.3, 0.4, 06, 0.8, 0.9, and 0.95 are
solved4. In the first part of the analysis a typical sphere is
assumed with η=Ro/Ri=1.1, as this geometry is commonly
considered to be the border between a "thin" and a "thick"
spherical vessel.
Due to the symmetry of the problem, the
distribution of KI as a function of the parametric angle ψ is
given only in the range of ψ= ψ0-90° (see Figs. 2b-2d). The
angle ψ=ψ0 represents the cusp, the extreme point of
intersection between the crack plane and the internal
surface of the sphere, and, ψ=90° is the deepest point of
the crack. It is worthwhile noting that the value of ψ0 is
negative and its value varies from case to case depending
on the particular geometry of the crack.
All the SIFs are normalized with respect to K05
given by:
πa
K0 = σθθ (1)
Q
where σθθ , is the average hoop/meridional stress through
the sphere thickness given by:
R i2 pR i 1
σθθ = p = ⋅
RO − Ri
2 2
2 t (1 + t / 2R i ) (2)
and Q is the shape factor for an elliptical crack (see Raju
4
Due to lack of interest in certain cases not all possible combinations of these
parameters are solved for.
5
The definition of K0 in the case of a spherical pressure vessel is similar to
that for the case of a cylindrical pressure vessel, where the average hoop stress
through the cylinder's thickness is given by pRi /t.
 and Newman [8]). Q is given by the square of a complete Kmax/K0, and on the other hand, the crack front gets closer
elliptic integral of the second kind and is commonly [9] to the outer wall of the sphere increasing the finite body
approximated by: effect and thus Kmax/K0. For shallow cracks the first effect
1 . 65 is stronger than the second one, and therefore, as a/t
a
Q = 1 + 1 . 464   ; a c<1 increases, Kmax/K0 decreases. For cracks deeper, a/t>0.1-0.2
c (depending on n), this effect is reversed. In terms of the
1 . 65 (3) absolute value of Kmax, Fig. 8 emphasizes the opposing
c
Q = 1 + 1 . 464   ; a c>1 effect of a/t and n i.e., Kmax increases with crack depth and
a decreases as the number of cracks in the array becomes
The results for lunular cracks, a/c=1, and larger.
crescentic cracks, a/c≠1, will be presented separately. In
the latter case, differentiation is made between slender
cracks, a/c<1, and transverse cracks a/c>1

Lunular Cracks a/c=1

The variation of the SIF with the parametric
angle, ψ, along the fronts of lunular radial cracks of
relative depth a/t=0.025, 0.4, and 0.6 as a function of the
number of cracks in the array, n, are given in Figs 4-6. In
each of the lunular crack cases evaluated, including the
results for a/t=0.01, 0.05, 0.1, 0.2, and 0.3 not presented
herein, the SIF distribution along the crack front has the
same pattern: i.e., the maximum SIF occurs in the vicinity Fig. 4 The variation of KI /K0 along the fronts of
of the cusp, ψ= ψ0, and it decreases monotonically as ψ lunular radial cracks of relative depth of
tends to 90°. a/t=0.025, for arrays of n=1-20 cracks (η=1.1).
The highest SIF values occur for the case of two
cracks though all the results for n=1-8 hardly differ from
n=2. As the number of cracks in the array increase, the SIF
decreases. For very shallow cracks, (a/t=0.025, Fig. 4) the
results for n=1-20 are practically the same. This result is
important as shallow cracks play a major role in the
vessel's fatigue life. Unfortunately, these results can not be
compared to a half-penny shaped crack in a half- space [4]
(see page 410), because the critical SIF at ψ ≈ 0° is not
available in [4]. As the crack becomes much deeper
(a/t=0.4, Fig. 5, and a/t=0.6, Fig. 6) the SIFs become
distinct. The fact that even for the deepest crack in this Fig. 5 The variation of KI /K0 along the fronts of
case, a/t=0.6, the influence of n is less than 5% results from lunular radial cracks of relative depth of
the fact the vessel considered is only moderately thick a/t=0.4, for arrays of n=1-20 cracks (η=1.1).
η=1.1.
Fig. 7 shows the distribution of the normalized
SIF for arrays of n=1-8 cracks6 for various crack depths
a/t=0.025-0.6. As a/t increases, the normalized SIF
increases along the entire crack front. The maximum
increase occurs in the vicinity of the cusp, ψ= ψ0. As ψ
approaches 90 the effect of a/t decreases. Fig. 8 represents
the normalized maximum SIF, Kmax/K0, as a function of
crack depth for various crack arrays. For all crack arrays
(n=1-20), Kmax/K0 at first decreases with crack depth and
then increases. This is due to two inverse effects: as the
crack becomes deeper on the one hand the "opening
stress", σθθ , drops through the sphere's wall, thus reducing Fig. 6 The variation of KI /K0 along the fronts of
lunular radial cracks of relative depth of
6
a/t=0.6, for arrays of n=1-20 cracks (η=1.1).
In this case, the SIF distributions are practically identical for all these arrays.

4 Copyright © 2008 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 08/26/2018 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 13-14, the location of Kmax shifts as in the case of a/c=0.5.
While in the case of a/c=0.5 (see Fig. 12), Kmax/K0 moves
from ψ=90° to the vicinity of the cusp for deep cracks of
a/t≥0.6, in the case of more slender cracks, a/c=0.2 (see
Fig. 15), this transition occurs for much shallower cracks
a/t≥0.2.

Fig. 7 The variation of KI /K0 along the fronts of

lunular radial cracks of relative depth of
a/t=0.025 to 0.6 (n=1-8 cracks, η=1.1).

Fig. 9 The variation of KI /K0 along the fronts of

slender crescentic radial cracks (a/c=0.5)
of relative depth a/t=0.025, for arrays of
n=1- 20 cracks (η=1.1).

Fig. 8 The variation of the maximal relative SIF

Kmax /K0 along the fronts of lunular radial
cracks as a function of relative crack depth
a/t, for arrays of n=1-20 cracks.

Slender Crescentic Cracks a/c<1

Two geometries of slender crescentic crack are
considered with ellipticities of a/c=0.5 and a/c=0.2. SIF
distributions are evaluated for both cases for a variety of Fig. 10 The variation of KI /K0 along the fronts
crack depths a/t=0.025-0.6, and for arrays of up to n=20 of slender crescentic radial crack (a/c=0.5)
of relative depth a/t=0.4, for arrays of
cracks.
n=1-20 cracks (η=1.1).
Typical results for cracks with ellipticity, a/c=0.5,
and depths, a/t=0.025, 0.4, and 0.6, are presented in Figs 9-
11. The maximum SIF for the deep crack a/t=0.6 occurs in
the vicinity of the cusp ψ= ψ0, as in the case of lunular
cracks. However, for the shallower crescentic cracks
a/t=0.025 and 0.4, Kmax moves to ψ=90°, the deepest point
of the crack. The case of n=1-8 cracks is still the most
critical case for slender crescentic crack arrays. The
normalized SIFs for arrays of n=1-8 cracks for various
crack depths a/t=0.025-0.8 are presented in Fig. 12. As a/t
increases, the normalized SIF increases along the entire
crack front. Furthermore, while for cracks of depth a/t<0.6
Kmax occurs at ψ=90°, for deeper cracks, a/t≥0.6, Kmax shifts Fig. 11 The variation of KI /K0 along the fronts of
to the vicinity of cusp ψ= ψ0. slender crescentic radial cracks (a/c=0.5)
In the case of the more slender cracks i.e., of relative depth a/t=0.6, for arrays of
a/c=0.2, of depths, a/t=0.025, and 0.6, presented in Figs n=1-20 cracks (η=1.1).

5 Copyright © 2008 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 08/26/2018 Terms of Use: http://www.asme.org/about-asme/terms-of-use


Fig. 12 The variation of KI /K0 along the fronts of
slender crescentic radial cracks (a/c=0.5) as
a function of relative crack depth a/t, for
arrays of n=1-8 cracks (η=1.1).
Fig.13 The variation of KI /K0 along the fronts of
very slender crescentic radial cracks
(a/c=0.2) of relative depth a/t=0.025, for
arrays of n=1-20 cracks (η=1.1).
Fig.14 The variation of KI /K0 along the fronts of
very lender crescentic radial cracks
(a/c=0.2) of relative depth a/t=0.6, for
arrays of n=1-20 cracks (η=1.1).
Whereas the SIF distribution in the lunular case is
monotonic along the crack front (Figs. 4-6), The SIF for
the slender crescentic case reaches a minimum value
between ψ=0° and ψ=12°. It is also worthwhile noting, that
for very shallow slender crescentic cracks (a/t=0.025, and
a/c=0.2 and 0.5) the maximum SIF, at ψ=90°, tends to the
plane-strain solution i.e., K max / σ ∞ π a = 1.12
6 Copyright © 2008 by ASME
Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 08/26/2018 Terms of Use: http://www.asme.org/about-asme/terms-of-use
From Fig. 16 it can be seen that for slender
crescentic cracks, crack depth, a/t, and the number of
cracks in the array, n, have the same adverse effect on Kmax
as in the case of lunular cracks (Fig. 8); namely, Kmax
increases with crack depth and decreases as the number of
cracks in the array becomes larger. The same behavior
occurs for a/c=0.2 (these results are not presented herein
for the purpose of clarity)
Fig. 15 The variation of KI /K0 along the fronts of
slender crescentic radial cracks (a/c=0.2)
as a function of relative crack depth a/t, for
arrays of n=1-8 cracks (η=1.1).
Fig.16 The variation of the maximal relative SIF
Kmax /K0 along the fronts of slender radial
cracks (a/c=0.5) as a function of relative
crack depth a/t, for arrays of n=1-20
cracks (η=1.1).
Transverse Crescentic Cracks a/c>1
The variation of the SIF along the front of
transverse cracks, a/c=1.5, of relative depths a/t=0.025,
and 0.6, are presented in Figs 17-18. In all transverse-crack
cases solved, including those cases which are not presented
for reason of clarity, the maximum SIF7 occurs at around
ψ=0°. From the point on the SIFs distributions along the
crack from are similar to those of the lunular cracks (see
Figs. 4-6). The highest values of the SIF are still attained
7
The value of the SIF for points along the crack front between the
cusp, ψ= ψ0, and ψ=0° is constant for all practical purpose

Fig.17 The variation of KI /K0 along the fronts of
transverse crescentic radial cracks (a/c=1.5) of
relative depth a/t=0.025, for arrays of n=1-20
cracks (η=1.1).
Fig.18 The variation of KI /K0 along the fronts of
transverse crescentic radial cracks (a/c=1.5)
of relative depth a/t=0.6, for arrays of n=1-20
cracks (η=1.1).
the n=1-8 case, however, the influence of n for these cases
is minuscule. Fig. 19 represents the maximum SIF, Kmax/K0,
as a function of crack depth for various transverse
crescentic crack arrays. As in all the previous cases,
Kmax/K0 at first decreases with crack depth and then
increases. These results further accentuate the inverse
effects of the "opening stress" vs. the "finite body" effect
(see explanation to Fig. 8). In terms of the absolute value
of Kmax, Fig. 19 stresses once again the opposing effect of
a/t and n on Kmax.
From the fatigue point of view the most
interesting case is a very shallow crack, such as a/t=0.025,
that might represent the initial crack depth. For this crack
the SIFs are almost n–independent. Fig. 20 presents the
variation of the SIF along the crack front for n=1-20 for
various crack ellipticities. From this figure one can see the
gradual transition of the location of Kmax from ψ=90° for
slender crescentic cracks to the vicinity of ψ= ψ0 for
lunular cracks and transverse crescentic cracks. Moreover,
it appears that for a certain value of ellipticity around
a/c~0.7-0.8 the SIF is almost uniformly distributed along
the crack front with almost equal KI values at ψ=90° and
7 Copyright © 2008 by ASME
Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 08/26/2018 Terms of Use: http://www.asme.org/about-asme/terms-of-use
ψ= ψ0 (see Fig. 20 the curve for a/c=0.75). Therefore, one
might expect that cracks having a high variability of the
SIF along their crack front will tend to grow in such a way
that will minimize this variation i.e., no matter what their
initial shape is they will tend to change it towards an
ellipticity of around a/c~0.7-0.8.
Fig.19 The variation of the maximal relative SIF
Kmax/K0 along the fronts of transverse
crescentic radial cracks (a/c=1.5) as a
function of relative crack depth a/t, for
arrays of n=1-20 cracks (η=1.1).
Fig. 20 The variation of KI /K0 along the front of a
very shallow crack (a/t=0.025) for various
crack ellipticities a/c (n=1-20, η=1.1).
The Influence of the Sphere Geometry η= Ro/Ri on the
SIF
So far, a moderately thick-walled cracked sphere of
radii ratio η=Ro/Ri=1.1 was considered. In order to
quantify the effect of the sphere's geometry on the
prevailing SIFs, four additional configurations of thin and
thick spheres with η=1.01, 1.05, 1.7 and 2.0 are analyzed.
All the crack configurations considered for the original
sphere, η=1.1 are resolved for the four new sphere
geometries. Therefore, SIFs distributions for radial, lunular
or crescentic crack arrays, containing n=1 to 20 cracks;
with ellipticities a/c=0.2 to 1.5; and ratio of crack-depth to
wall-thickness a/t=0.025 to 0.95 are evaluated for the
spheres with η=1.01, 1.05, 1.7 and 2.0.
In order to evaluate the influence of the spheres
 geometry on the absolute values of the most critical SIFs,
the results for Kmax are re-normalized to K 00 = p R i ,
actually representing the absolute values of the SIFs.
Kmax/K00 vs. crack depth for the most critical crack
configurations, for these five geometries of the sphere
η=1.01, 1.05, 1.1, 1.7, and 2.0, are presented in Figs 21-24
for the ellipticities a/c=0.2, 0.5, 1.0 and 1.5 respectively.
For all crack configurations, the lower the relative
thickness of the sphere (η) the higher the values of Kmax/K00
i.e., the thin shell cases are more critical than those of the
thick ones.
Fig. 21 Kmax /K00 vs. a/t for the most critical slender
crescentic crack array of a/c=0.2 in spheres
of radii ratios of η=1.01, 1.05, 1.1, 1.7, and 2.0.
Fig. 22 Kmax /K00 vs. a/t for the most critical slender
crescentic crack array of a/c=0.5 in spheres
of radii ratios of η=1.01, 1.05, 1.1, 1.7, and 2.0.
Fig. 23 Kmax /K00 vs. a/t for the most critical lunular
crack array in spheres of radii ratios of
η=1.01, 1.05, 1.1, 1.7, and 2.0.
8 Copyright © 2008 by ASME
Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 08/26/2018 Terms of Use: http://www.asme.org/about-asme/terms-of-use
Fig. 24 Kmax /K00 vs. a/t for the most critical transverse
crescentic crack array of a/c=1.5 in spheres of
radii ratios of η=1.01, 1.05, 1.1, 1.7, and 2.0.
CONCLUDING REMARKS
The three-dimensional interaction effects between
cracks, in numerous configurations of arrays of radial,
lunular and crescentic, internal, surface cracks in thin and
thick spherical pressure vessel, were studied via the impact
on the prevailing SIFs. The influence of the number of
cracks in the array, crack ellipticity, and crack depth, as
well as the sphere geometry was determined.
The case of n=8 was found to be the most critical
for the thin spheres (of η=1.01, 1.05, and 1.1), though SIF
values for n=1-8 are practically the same. For the thicker
spheres (of η=1.7, and 2.0) the most critical configuration
is found to consist of four cracks.
In all cases analyzed as crack depth increases and
as its ellipticity decreases, Kmax becomes larger, yielding a
more critical situation. The location of the maximum SIF,
Kmax, along the crack front is either at ψ=90º or near the
cusp ψ=ψ0 and depends on crack ellipticity, a/c, and crack
depth, a/t.
The SIFs are considerably affected by the three -
dimensionality of both the sphere and the cracks. As the
sphere becomes thinner (η→1), and the crack becomes
shallower (a/t<<1), and more elongated (a/c<<1) the
difference between the 3-D SIF and the 2-D approximation
increases considerably. Therefore, crack growth rate based
on the 2-D model over estimates the realistic 3-D crack
growth rate, yielding a very conservative total fatigue life
estimate.
The interaction effect between coplanar, lunular
and crescentic surface cracks emanating from the inner
surface of a pressurized sphere is presently under
consideration and will be soon reported elsewhere.
 NOMENCLATURE [2] Folias E. S., 1965, "A Finite Line Crack in a
a - crack depth Spherical Shell," Int. J. of Fracture Mechanics, 1,
c - crack half length 1, pp. 20-46.
KI - Mode I SIF
Kmax - maximum SIF [3] Erdogan, F., and Kibler, J. J., 1969, "Cylindrical and
K0 - normalizing SIF [eq. (1)] Spherical Shells with Cracks," Int. J. of Fracture
K00 - normalizing SIF, K 00 = p R i Mechanics, 5, pp. 229-237
n - number of cracks in the array
Q - shape factor for lunular or crescentic crack [4] Tada, H., Paris, P. C., Irwin, G. R., 1985, The Stress
[eq. (3)] Analysis of Cracks Handbook, 2nd Ed., Paris
p - internal pressure Productions Incorporated (and Del Research
Ri - inner radius of the sphere Corporation), St. Louis, MO.
Ro - outer radius of the sphere
r, θ, φ- spherical coordinates [5] Atsumi, A., and Shindo, Y., 1983, "Axisymmetric
t - sphere wall thickness Singular Stresses in a Thick-Walled Spherical Shell
with an Internal Edge Crack," J. of Applied
Greek Symbols Mechanics, 50, pp. 37-42.
η - sphere's radii ratio Ro / Ri
ν - Poisson's ratio [6] ANSYS 11.0, 2007, Verification Manual, Swanson
Analysis Systems Inc.
σ rr - radial stress component
σ θθ - hoop stress component [7] Barsom, R. S., 1976, “On the Use of Isoparametric
Finite Elements in Linear Fracture Mechanics”,
σ ϕϕ - meridional stress component
International Journal of Numerical Methods in
σθθ - average hoop stress through the sphere's Engineering, 10, 1, pp. 25- 37.
thickness[eq.(2)]
ψ - parametric angle for lunular and crescentic cracks [8] Raju, I. S., and Newman, Jr., J. C., 1980, “Stress
(Fig. 2) Intensity Factors for Internal Surface Cracks
in Cylindrical Pressure Vessel”, ASME Journal of
REFERENCES Pressure Vessel Technology, 102, pp. 342-
346.
[1] Wang, Z. R., Kun Dai, 2000, "The Development of
Integral Hydro-Bulge Forming Process and its [9] Newman Jr., J. C., and Raju, I. S., 1979, “Analysis of
Numerical Simulation," J. of Mater. Process. Surface Cracks in Finite Plates Under
Technol., 102, pp.168-173. Tension and Bending Loads”, NASA TP-1578.

9 Copyright © 2008 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 08/26/2018 Terms of Use: http://www.asme.org/about-asme/terms-of-use