You are on page 1of 20

# * Corresponding author. Tel.: 0086-21-6384-2238.

## E-mail address: yurenhuc@online.sh.cn (Y. Hu).

Marine Structures 14 (2001) 311}330
Analysis on the ultimate longitudinal strength
of a bulk carrier by using a simpli"ed method
Yuren Hu*, Ainian Zhang, Jiulong Sun
School of Naval Architecture and Ocean Engineering, Shanghai Jiaotong University, 1954 Huashan Road,
Shanghai 200030, People's Republic of China
Shanghai Rules and Research Institute, China Classixcation Society, 1234 Pudong Ave., Shanghai 200135,
People's Republic of China
Received 16 May 2000; received in revised form 28 August 2000; accepted 21 September 2000
Abstract
The ultimate longitudinal strength of a typical bulk carrier is analyzed by using a simpli"ed
method. The moment}curvature curve, the ultimate bending moment and the location of the
instantaneous neutral axis at ultimate state are calculated for both hogging and sagging
conditions of the ship under vertical bending. The stress distribution over the hull cross-section
at ultimate state is also obtained. The ultimate strength of the ship hull under combined vertical
and horizontal bending moments is further investigated. An interaction curve is obtained
according to the results of a series of calculation for the hull subjected to bending conditions
with di!erent angles of curvature. It is found that the interaction curve is asymmetrical because
the hull cross-section is not symmetrical about the horizontal axis and the behavior of the
structural members under compression is di!erent from that under tension due to the non-
linearity caused by buckling. The angle of the resultant bending moment vector and that of the
curvature vector are di!erent in general cases. An interaction equation suitable for bulk carriers
is proposed based on the results of the analyzed ship. 2001 Elsevier Science Ltd. All rights
reserved.
Keywords: Ultimate longitudinal strength; Bulk carrier; Simpli"ed method; Combined vertical and
horizontal bending; Interaction equation
PII: S 0 9 5 1 - 8 3 3 9 ( 0 0 ) 0 0 0 6 3 - 0
Nomenclature
a length of element, plate or sti!ener
A
EG
cross-sectional area of element i
A

## cross-sectional area of sti!ener

b plate width
E Young's modulus
f deduction factor representing e!ect of plate buckling mode
G shear modulus
I
N"
polar moment of inertia of sti!ener cross-section about axis of rotation
I
S "
sectorial moment of inertia of sti!ener cross-section about axis of rotation
J torsional moment of inertia of sti!ener cross-section
k
P
spring sti!ness per unit length of rotational restraint provided by plate
M
N
plastic moment
M
S
ultimate longitudinal bending moment
M
W
"rst yield moment
M
7
vertical bending moment (about the Y-axis)
M
8
horizontal bending moment (about the Z-axis)
p lateral pressure
p
P
proportional limit of material
t plate thickness
t
U
web thickness of sti!ener
z
C
distance from elastic neutral axis to base line
z
S
distance from instantaneous neutral axes at ultimate state to base line
:

, :
`
parameters of interaction equation
[ plate slenderness
[
"
sectional property of sti!ener
o
N
initial de#ection of plate
c strain
c dimensionless strain with respect to yield strain of sti!ener
c
?
average strain of element
c
W
yield strain of material
c
S
average strain at ultimate state
p factor of residual tension stress block width
0 angle of curvature vector
magnitude of curvature vector

## curvature corresponding to ultimate state

z slenderness of panel
j parameter representing lateral pressure acting on panel
parameter representing initial de#ection of panel
j
C
e!ective radius of gyration of panel cross-section
j
S
radius of gyration of panel cross-section at ultimate state
o stress
312 Y. Hu et al. / Marine Structures 14 (2001) 311}330
o dimensionless stress with respect to yield stress of sti!ener
oPB
?
average stress of plate with e!ects of residual stresses and initial de#ections
o
?F
average stress of hard corner element
o
?N
average stress of panel element
o
?Q
average stress of sti!ener
o
AP
buckling stress of plate
o
W
yield stress of material
o
P
residual stress
o
2AP
inelastic tripping stress of sti!ener
o
2#
elastic tripping stress of sti!ener
angle of resultant bending moment vector
1. Introduction
The longitudinal strength is the most important criteria for structural design of ship
hulls, which is generally represented by the maximum bending moment that the hull
cross-section can withstand. The linear elastic theory has been employed to predict
the longitudinal strength of the ship hull for years. According to this theory, the
maximum bending moment that the hull cross-section can withstand is equal to the
bending moment corresponding to the "rst yield, that is, the bending moment when
the maximumstress on the hull cross-section reaches the yield stress of the material. In
design practice, an allowable stress is used instead of the yield stress, which corres-
ponds to a safety factor against yielding.
However, researches in the last 20 years have revealed that the linear elastic
theory is not adequate in estimating the longitudinal strength of the ship hull. It is
necessary to take into account the following factors: (1) various possible failure
modes including buckling, (2) progressive and interactive behavior of the failure of
structural members, (3) redistribution of the loads on the hull cross-section and (4)
residual strength of structural members after buckling and even after collapse. By
considering these factors, the maximum bending moment that the hull cross-section
can withstand is designated by the ultimate longitudinal strength, which represents
the maximum load-carrying capacity of the ship hull under longitudinal bending
[1}4]. Obviously, calculation of the ultimate longitudinal strength is a non-linear
problem in which both the non-linearity of material and the non-linearity of geometry
are involved.
There are three main methods to calculate the ultimate longitudinal strength of the
ship hull, namely, the non-linear "nite element method (NFEM) [5,6], the idealized
structural unit method (ISUM) [7,8] and the simpli"ed method (SM) [9]. In addition,
the empirical equations regressed from the results of theoretical calculation or derived
theoretically under certain assumptions are also of signi"cance in practical applica-
tion [10,11]. Among these methods, the simpli"ed method based on the discrete
analytical model of the hull cross-section has proved to be a simple and e!ective
Y. Hu et al. / Marine Structures 14 (2001) 311}330 313
method with adequate accuracy. It has drawn wide attention in the "eld of naval
architecture.
In this paper, the ultimate longitudinal strength of a typical bulk carrier is analyzed
by using the simpli"ed method. First, vertical bending is considered. The
moment}curvature curves, the values of the ultimate longitudinal bending moments
and the locations of the instantaneous neutral axes at ultimate states are obtained for
both hogging and sagging conditions. The stress distribution over the hull cross-
section at the ultimate state is also obtained and discussed. Secondly, the ultimate
strength under combined vertical and horizontal bending moments is considered. An
interaction curve is obtained according to the results of a series of calculation for the
hull subjected to bending conditions with di!erent angles of curvature vector. The
characteristics of the interaction curve are discussed in connection with the character-
istics of the hull structure of the bulk carrier. Finally, an interaction equation suitable
for bulk carriers is proposed.
2. Features of the method
2.1. Procedure of calculation
The simpli"ed method was "rst proposed by Smith [9] in 1970 s. This method
was further developed and applied by other researchers [12}20]. The authors
recently proposed a simpli"ed method and developed a corresponding com-
puter program, UStrength [21]. The program is employed in this paper to analyze
the ultimate longitudinal strength of a bulk carrier by using the simpli"ed
method.
The hull cross-section, generally the midship cross-section, is divided into
`panel elementsa and `hard-corner elementsa, to construct an analytical model in
the simpli"ed method. Calculation of the ultimate longitudinal strength by using
the simpli"ed method follows the following procedure. Let the curvature increase
by small increments. At each curvature, calculate the strain of every element according
to the assumption of plane cross-section. Then determine the stress of the elements
from the average stress}strain relationship of the elements. Take the moment of
stress on each element about the instantaneous neutral axis. The resultant moment
of all elements is the bending moment of the cross-section at the considered curvature.
After a series of calculation at di!erent curvatures, a moment}curvature curve
can be obtained and the moment corresponding to the point on the curve with a
zero gradient is the ultimate longitudinal bending moment of the cross-section (see
Fig. 3). In the above procedure, the location of the instantaneous neutral axis should
be determined by iteration or trial-and-error according to the condition that the sum
of the stresses on all elements of the cross-section equals zero.
The accuracy of the simpli"ed method depends largely on the accuracy of the
average stress}strain relationship of the elements. The main features of the average
stress}strain curves of the panel element and the hard-corner element in the method of
the present paper are described as follows.
314 Y. Hu et al. / Marine Structures 14 (2001) 311}330
Fig. 1. Panel element and hard-corner element.
2.2. Behavior of panel element
A panel element is de"ned as a combination of a sti!ener and the plate attached to
it, as shown in Fig. 1(a). In this panel element model, the plates on the left and right
sides of the sti!ener are assumed to be di!erent both in geometry and in material. This
is of signi"cance in some cases, for instance, where sti!eners are unequally spaced
and/or plates with di!erent thickness or with di!erent yield stress connect.
To study the behavior of the panel element, it is necessary "rst to study the
non-linear behavior of the plate, which is represented by the average stress}strain
curve. In the present method, the average stress}strain curves proposed by the authors
[22] for rectangular long and wide plates with the e!ects of residual stresses and initial
de#ections are employed. The average stress}strain curves are generated from the
design formula proposed by Faulkner [23] after introducing an e!ective slenderness
of the plate in terms of the average strain. The e!ects of residual stresses and initial
distortions were also considered. The assumption proposed by Valsgard [24] is
adopted in generating the average stress}strain curve for wide plates.
When a panel element is under compression, there are three failure modes, namely,
(1) yield of the sti!ener; (2) elastic or inelastic #exural buckling of the sti!ener}plate
combination as a beam-column, and (3) tripping of the sti!ener about its line of
attachment to the plate. The non-linear behavior of the panel element is represented
by the load-end shortening curve expressed as follows:
o
?N
(c
?
)"
o
?Q
(c
?
)A
Q
#
`
o PB
?
(c
?
/c
W
)o
W
b

#
`
o PB
?`
(c
?
/c
W`
)o
W`
b
`
t
`
A
Q
#
`
b

#
`
b
`
t
`
(1)
Y. Hu et al. / Marine Structures 14 (2001) 311}330 315
where o
?Q
is the dimensionless stress of the sti!ener, which is calculated according to the
failure mode. o PB
?
is the average stress of the plate, in which the e!ects of residual stresses
and initial de#ections are taken into account. A
Q
is the cross-sectional area of the
sti!ener. b and t are width and thickness of the plate. o
W
and c
W
are the dimensionless
yield stress and yield strain of the plate with respect to the yield stress and the yield
strain of the sti!ener, respectively, i.e. o
W
"o
W
/o
W1
and c
W
"c
W
/c
W1
. Subscripts 1 and
2 denote the plates on the left and right sides of the sti!ener, respectively.
For the failure mode of beam-column type #exural buckling, the load-end shorten-
ing curve of the panel element is derived by a method similar to that proposed by
Gordo and Guedes Soares [25]. After introducing a strain-governed e!ective slender-
ness of the panel element, the load-end shortening curve is generated from the theory
of tangent modulus and the Ostenfeld}Bleich parabola [26]. The e!ect of residual
stresses is considered in generating the average stress}strain relationship of the plate
part of the element. The e!ect of residual stresses on the sti!ener is inherently included
in the Ostenfeld}Bleich parabola. However, the Ostenfeld}Bleich parabola cannot
cope with the e!ects of initial de#ections and lateral pressures. So, if there is initial
de#ection or lateral pressure, the Perry}Robertson equation is employed to derive the
relationship can be expressed as
o
?1
(c
?
)"

c
?

1
2
1!j#
1#
z`
j
C
j
S

` c
S
c
?

1
4
1!j#
1#
z`
j
C
j
S

` c
S
c
?

`
!
1!j
z`
j
C
j
S

` c
S
c
?

, 0)c
?
)c
S
,
c
?

1
2
1!j#
1#
z`
j
C
j
S

` c
S
c
?

1
4
1!j#
1#
z`
j
C
j
S

` c
S
c
?

`
!
1!j
z`
j
C
j
S

` c
S
c
?

, c
?
*c
S
,
(2)
for the panel element bending towards the plate side, and
o
?1
(c
?
)"

c
?

1
2
o
WKGL
!j#
1#
z`
j
C
j
S

` c
S
c
?

1
4
o
WKGL
!j#
1#
z`
j
C
j
S

` c
S
c
?

`
!
o
WKGL
!j
z`
j
C
j
S

` c
S
c
?

, 0)c
?
(c
S
,
c
?

1
2
o
WKGL
!j#
1#
z`
j
C
j
S

` c
S
c
?

1
4
o
WKGL
!j#
1#
z`
j
C
j
S

` c
S
c
?

`
!
o
WKGL
!j
z`
j
C
j
S

` c
S
c
?

, c
?
*c
S
,
(3)
316 Y. Hu et al. / Marine Structures 14 (2001) 311}330
for the panel element bending towards the sti!ener side. In the above two equations,
and j are parameters related to the initial de#ection and the lateral pressure,
respectively [4]. z is the slenderness of the panel element, j
C
gyration of the panel element at given strain level. j
S
is the radius of gyration at the
ultimate state. c
S
"c
S
/c
W1
is the dimensionless average strain at the ultimate state,
o

"min(o
W
,o
W`
) is the dimensionless yield stress of the plate. Note that , j and
j
C
vary with the average strain level. The detailed derivation of Eqs. (2) and (3) can be
found in [26] and [27].
In ship hull structures, the sti!eners usually have thin-walled open cross-sections,
such as T, angle cross-sections. Due to the low torsional rigidity of the thin-walled
open cross-section, tripping is likely to occur prior to beam-column-type #exural
buckling. A method to estimate the elastic critical tripping stress with the e!ect of
lateral pressure is proposed recently by the authors [28]. Generally, the elastic critical
tripping stress is obtained by solving an eigenvalue problem. If the absolute value of
the lateral pressure is not very high, the following approximate equation can be
employed.
o
2 #
" min
K`
2
EI
S "
(m/a)`#GJ#[k
P
#2q[
"
(`m`!3)/12](a/m)`
I
N"
#k
P
f/o
AP
(a/m)`
(4)
where I
S "
and I
N "
are the sectorial and polar moments of inertia of the sti!ener
cross-section about the axis of rotation, respectively, J is the torsional moment of
inertia of the cross-section, [
"
is another sectional property whose de"nition can be
found in [28], k
P
is the spring sti!ness per unit length of the rotational restraint on the
axis of rotation provided by the plate, q is the lateral pressure, f is the deduction
factor representing the e!ect of the plate buckling mode. o
AP
is the buckling stress of
the plate.
The inelastic critical stress of tripping is obtained in a similar way to that used for
inelastic beam-column-type #exural buckling by using the tangent modulus theory,
that is
oN
2 AP
"
o
2 AP
o
W1
"1!p
P
(1!p
P
)
o
W1
o
2 #
, (5)
where p
P
is the proportional limit of the material.
The load shedding in the failure mode of sti!ener tripping is assumed to have the
following pattern:
o
?1
(c
?
)"o
2AP
c
2
c
?
"
o `
2AP
c
?
(6)
2.3. Behavior of hard-corner element
The hard-corner element is composed of several plates and possibly a sti!ener at the
corner, as shown in Fig. 1(b). Generally, it is thought that the hard-corner element can
fully withstand the load, that is to say, its load-end shortening curve is the same as the
stress}strain curve of the elastic-perfectly plastic material. However, it has been noted
Y. Hu et al. / Marine Structures 14 (2001) 311}330 317
Table 1
Principal dimensions of 34,000 ton bulk carrier
Length overall 200.00 m
Length between perpendiculars 191.00 m
Depth molded 14.90 m
Draft 10.68 m
Block coe\$cient 0.8619
Frame spacing 0.80 m
that though #exural buckling will not occur for the whole hard-corner element, the
plates composing the element are likely to buckle and thus cannot fully withstand
the load. For this reason, a hard-corner model is proposed by the authors to include
the e!ect of the buckling of the plates [26]. In this model, the average stress}strain
relationship of the plates is similar to that of plates in the panel element and the
average stress}strain relationship of the sti!ener is the same as that of the elastic-
perfectly plastic material. The load-end shortening curve of the hard-corner can be
expressed as
o
?F
(c
?
)"
o
?Q
(c
?
)A
Q
#
`
LN
G
o PB
?G
(c
?
/c
WG
)o
WG
b
G
t
G
A
Q
#
`
LN
G
b
G
t
G
. (7)
The main di!erence between Eqs. (1) and (7) is that o
?Q
(c
?
) in Eq. (1) is determined from
the load-end shortening relationship according to di!erent failure modes of the panel
element, while o
?Q
(c
?
) in Eq. (7) is determined from a stress}strain relationship the
same as that of the elastic-perfectly plastic material. That is to say, only one failure
mode of yielding is possible for the hard-corner element as a whole. The results from
this model have proven to be satisfactory [26].
3. Ultimate longitudinal strength of a bulk carrier
A 34000 ton bulk carrier is analyzed by using the UStrength program, which is
based on the simpli"ed method. The principal dimensions of the ship are listed in
Table 1. The midship cross-section is shown in Fig. 2. The dimensions of the
longitudinals are listed in Table 2. High-strength steel is used for structures at deck
and at upper part of the topside tank. The yield stress of HST is 315 MPa. Otherwise,
the yield stress of the material is 235 MPa.
The cross-section is divided into 98 panel elements and 34 hard-corner elements in
the calculation. The e!ects of residual stresses and initial de#ections are taken into
account. The weld residual stress is estimated by the following equation.
o
P
"
o
P
o
W
"
2pt
b!2pt
, (8)
318 Y. Hu et al. / Marine Structures 14 (2001) 311}330
Fig. 2. Midship cross-section of 34,000ton bulk carrier.
Y. Hu et al. / Marine Structures 14 (2001) 311}330 319
Table 2
Dimensions of longitudinals
Stif No. Type Dimensions (mm) Yield stress (MPa)
1 Angle bar 350;100;12/17 315
2 Angle bar 300;90;13/17 315
3 Angle bar 250;90;12/16 235
4 Angle bar 250;90;10/15 315
5 Flat bar 200;20 235
6 Flat bar 120;11 235
7 Flat bar 220;11 235
8 T bar 180;12#100;16 235
9 T bar 150;12#100;16 235
where p is the factor of residual tension stress block width. For ordinary plates in
as-weld ship structures, it is suggested by Faulkner that p"3.0}4.5 [29]. In the
calculation for the present bulk carrier, p"4.0 is chosen. The initial de#ection of
plates is estimated by the equation proposed by Faulkner [29], i.e.
o
M
N
"
o
N
t
"0.12[`

t
U
t
, t
U
)t, (9)
where [ is the slenderness of the plate and t
U
is the thickness of the sti!ener web.
First consider vertical bending of the ship hull. A series of calculation is performed
by increasing the curvature by small increments for both hogging (0"03) and sagging
(0"1803) conditions. The moment}curvature curve obtained from the calculation is
shown in Fig. 3. For comparison, the moment}curvature curve without the e!ects of
residual stresses, initial de#ections and lateral pressures is also plotted in the "gure.
The ultimate longitudinal bending moments of the ship, M
S
, under hogging and
sagging conditions can be identi"ed from the moment}curvature curve, which are
listed in Table 3 together with the curvatures corresponding to the ultimate states,
S
.
The "rst yield moment, M
W
, and the plastic moment, M
N
, are also listed in the table.
The ratios of the ultimate bending moments to the "rst yield moment and the
plastic moment are listed in the table too. The values without the e!ects of residual
stresses, initial de#ections and lateral pressures are also listed in the table. It can be
seen that the ultimate longitudinal bending moments will be over-estimated without
the e!ects of residual stresses, initial de#ections and lateral pressures. The relative
errors are 20.99% and 13.27% for hogging and sagging, respectively, for the present
ship.
Variation of the location of the instantaneous neutral axis during bending is shown
in Fig. 4. It can be seen from the "gure that when the curvature is small, The hull
bends within the elastic range and the location of the instantaneous neutral axis is
basically unchanged. The small di!erence of the location under hogging or sagging
from that of the elastic neutral axis at zero curvature is due to the e!ects of residual
stresses and initial de#ections that are taken into account in the calculation. No
320 Y. Hu et al. / Marine Structures 14 (2001) 311}330
Fig. 3. Moment-curvature curve of 34,000 ton bulk carrier.
residual stress and initial de#ection is considered in calculating the elastic neutral axis
at zero curvature. The distance from the elastic neutral axis to the base line is
z
C
"6.17 m. When the curvature becomes large enough to cause buckling of the
compressed structural members, the location of the instantaneous neutral axis cha-
nges. It moves towards the deck in hogging condition, while towards the bottom in
sagging condition. The distances from the instantaneous neutral axes at the ultimate
state to the base line is z
S
"8.397 m in the hogging condition and z
S
"4.454 m in the
sagging condition.
Y. Hu et al. / Marine Structures 14 (2001) 311}330 321
Table 3
Results of 34,000 ton bulk carrier
With e!ects of residual stresses, etc. Without e!ects of residual stresses, etc.
Sagging Hogging Sagging Hogging
M
W
(;10" kNm) 2.519 2.519
M
N
(;10" kNm) 3.950 3.950
M
S
(;10" kNm) 2.780 2.778 3.169 3.361

S
(;10`) 2.83 2.95 2.28 2.34
M
S
/M
W
1.104 1.103 1.258 1.334
M
S
/M
N
0.704 0.703 0.802 0.851
Fig. 4. Location of neural axis of hull cross-section.
The stress distributions over the hull cross-section at the ultimate states in both
hogging and sagging conditions are plotted in Fig. 5. The stress unit in the "gure is
MPa. It should be noted that the stress in the "gure is the average stress of the
elements calculated by the simpli"ed method, which does not re#ect the di!erence
between the sti!ener and the plates that compose the element. It can be found in the
"gure that at the ultimate state in hogging, panels at the bottom of the hull and in the
lower part of the side shell have buckled. Stresses on these locations are approxim-
ately equal to the critical stress of the panels. While structural members on the deck, in
the topside tank and in the upper part of the side shell are still in the elastic range with
a linear stress distribution. At the ultimate state in sagging, panels on the deck, in the
topside tank and in the upper part of the side shell have buckled. Stresses on these
locations are approximately equal to the critical stress of the panels. On the other
hand, structural members at the bottom of the hull and in the lower part of the side
shell are still in the elastic range with a linear stress distribution.
322 Y. Hu et al. / Marine Structures 14 (2001) 311}330
Fig. 5. Stress distribution under ultimate state.
Fig. 6. Load-end shortening curves of panel elements in #exural buckling mode.
The results show that most of the panels that fail in compression are in the #exural
buckling mode. Typical load-end shortening curves of the panel elements in the
#exural buckling mode are shown in Fig. 6. A few panels fail in the tripping mode. For
example, the panel elements on the side girders at the bottom trip at the ultimate state
Y. Hu et al. / Marine Structures 14 (2001) 311}330 323
Fig. 7. Load-end shortening curve of panel element in tripping mode.
in hogging (see Fig. 5). Typical load-end shortening curve of the panel element in the
tripping mode is shown in Fig. 7.
4. Ultimate strength of the bulk carrier under combined vertical and horizontal bending
moments
Further consider the ultimate strength of the 34000ton bulk carrier under com-
bined vertical and horizontal bending moments. An interaction curve is obtained from
a series of calculation for the hull bending with di!erent angles of the curvature vector
from hogging of vertical bending (the angle of the curvature vector 0"03) through
horizontal bending (0"903) to sagging of vertical bending (0"1803). The interaction
curve is plotted in Fig. 8. It can be seen that the interaction curve in the "gure is
asymmetrical with the following characteristics.
(1) When the hull is under horizontal bending, that is, the angle of the curvature
vector is 0"903, the angle of the resultant bending moment vector, , is not equal
to 903 due to the e!ect of non-linearity. When 0"903, there exist both horizontal
(about the Z-axis) and vertical (about the Y-axis) bending moments on the hull
cross-section. For the present bulk carrier, when 0"903, the horizontal and
vertical bending moments are M
8F""
"3.895;10" kNm and M
7F""
"
0.195;10" kNm, respectively. The angle of the resultant bending moment is
"86.743.
(2) If the hull cross-section is subjected only to horizontal bending moment, that is,
the angle of the resultant bending moment vector is "903, then the angle of the
curvature vector is not necessarily equal to 903. The angle of the curvature vector
is 0"973 when "903 for the present ship. Thus, both horizontal and vertical
bending will occur under the action of a pure horizontal bending moment.
324 Y. Hu et al. / Marine Structures 14 (2001) 311}330
Fig. 8. Interaction curve for ultimate strength of 34,000ton bulk carrier under combined vertical and
horizontal bending moments.
(3) The maximum value of the horizontal bending moment occurs neither at 0"903
nor at "903. For the bulk carrier analyzed in this paper, the maximum
horizontal bending moment occurs at 0"723 and "81.443. The maximum
value of the horizontal bending moment is M
8S
"3.972;10" kNm. At the
moment the horizontal bending moment reaches its maximum, there exists
a vertical bending moment on the hull cross-section, which is
MH
7
"0.598;10" kNm for the present ship. If "903, when the vertical bending
moment is equal to zero obviously, the horizontal bending moment for the
present bulk carrier is M
8P""
"3.820;10" kNm, which is not the maximum.
(4) The interaction curve can be divided into two parts. The part on the left side of the
point MH
7
and that on the right side have di!erent shapes.
The reason for the asymmetry of the interaction curve is that the hull cross-section
of the bulk carrier is not symmetrical about the horizontal axis and the behavior of the
structural members under compression is di!erent from that under tension due to the
non-linearity caused by buckling. Therefore, the shape of the interaction curve of bulk
carriers is di!erent from that of single hull oil tankers with hull cross-sections basically
A further discussion can be made on the relationship between the angle of the
curvature vector and the angle of the resultant bending moment vector. If the location
Y. Hu et al. / Marine Structures 14 (2001) 311}330 325
Fig. 9. Relationship between angle of bending moment and angle of curvature.
of the instantaneous neutral axis is given, then the horizontal and vertical bending
moments on the hull cross-section can be written as.
M
7
"
L

G
z
G
(!o
G
(c
G
))A
EG
,
M
8
"!
L

G
y
G
(!o
G
(c
G
))A
EG
.
(10)
The magnitude and the angle of the resultant bending moment vector are
M"(M`
7
#M`
8
, "arctan
M
8
M
7
, (11)
respectively. In the above equations, the stress o
G
is determined by the strain c
G
and the
strain is further determined by the magnitude and the angle 0 of the curvature
vector. When one or more structural members buckle under compression, the
stress}strain relationship becomes non-linear, therefore, the angle of the curvature
vector and the angle of the resultant bending moment vector are not the same in
general cases. The relationship between the angle of the curvature vector and the angle
of the resultant bending moment vector for the present bulk carrier is plotted in Fig. 9.
The ultimate strength under combined vertical and horizontal bending moments
for another 74,000 ton bulk carrier is also investigated. The resulted interaction curve
is plotted in Fig. 10, which has an asymmetrical shape similar to that of the 34,000 ton
bulk carrier. The ultimate vertical bending moments in hogging and sagging condi-
tions are M
7SFME
"5.536;10" kNm and M
7SQ?E
"3.957;10" kNm, respectively.
The horizontal and vertical bending moments at 0"903 are M
8F""
"6.561;
10" kNm and M
7F""
"0.628;10" kNm, respectively, with the angle of the result-
ant bending moment being "84.543.When the angle of the resultant bending
moment vector is "903, the angle of the curvature vector is 0"1043. The
326 Y. Hu et al. / Marine Structures 14 (2001) 311}330
Fig. 10. Interaction curve for ultimate strength of 74,000 ton bulk carrier under combined vertical and
horizontal bending moments.
maximum horizontal bending moment occurs at 0"723 and "79.003, with a
maximum value of M
8S
"6.742;10" kNm and a corresponding vertical bending
moment M
H
7
"1.311;10" kNm. The horizontal bending moment at "903 is
M
8F""
"6.276;10" kNm.
Mansour et al. have proposed an empirical interaction equation based on the
calculated results for one container ship, one tanker and one cruiser, which is [30]

M
7
M
7S

#0.8

M
8
M
8S

`
"1.0,

M
7
M
7S

'

M
8
M
8S

M
8
M
8S

#0.8

M
7
M
7S

`
"1.0,

M
7
M
7S

M
8
M
8S

. (12)
Another interaction equation was proposed by Gordo and Guedes Soares based on
the results for four tankers, which is [31,32]

M
7
M
7S

?
#

M
8
M
8S

?
"1.0, 1.50(:(1.66. (13)
These two equations can "t the calculated data well for oil tankers, but since they are
in symmetrical forms, they cannot re#ect the asymmetry of the interaction relation-
ship for bulk carriers. According to the results shown in Figs. 8 and 10 for the 34,000
Y. Hu et al. / Marine Structures 14 (2001) 311}330 327
and 74,000 ton bulk carriers considered in this paper, the following equation is
now proposed.

M
7
!M
H
7
M
7SFME
!M
H
7

M
8
M
8S

"1.0, M
7
*M
H
7
,

M
7
!M
H
7
M
7SQ?E
!M
H
7

?
`
#

M
8
M
8S

?
`
"1.0, M
7
(M
H
7
. (14)
where M
7SFME
and M
7SQ?E
are the ultimate bending moments in hogging and sagging
conditions of the vertical bending. The vertical bending moment is de"ned as positive
in hogging and negative in sagging in the above equation.
For the 34,000 ton bulk carrier, when the parameters are taken as :

"1.5 and
:
`
"1.6, the curve generated from Eq. (14) is in fairly good agreement with the data
calculated from the simpli"ed method as shown in Fig. 8. For the 74,000 ton bulk
carrier, the curve with :

"1.4 and :
`
"1.8 "ts the calculated data well as shown in
Fig. 10. More reasonable values of the parameters :

and :
`
can be obtained by
statistics of the results from a series of calculation for more bulk carriers. For the
purpose of comparison, curves from Eqs. (12) and (13) are also plotted in Figs. 8
and 10.
By employing Eq. (14), an interaction equation can be established for any bulk
carrier by calculations for hogging and sagging of vertical bending and searching for
the angle corresponding to the maximum horizontal bending moment, which is
0"723 for the bulk carriers considered in the present paper.
5. Conclusions
The simpli"ed method based on the discrete hull cross-section model is a simple
and e!ective method to estimate the ultimate longitudinal strength of ship hulls.
A 34,000 ton bulk carrier is analyzed by using a simpli"ed method in this paper. The
moment}curvature curves, the values of the ultimate longitudinal bending moments,
the locations of the instantaneous neutral axes at the ultimate states and the stress
distribution over the hull cross-section at the ultimate states are obtained for both
hogging and sagging conditions of vertical bending. The ultimate strength of the ship
under combined vertical and horizontal bending moments is also investigated. An
interaction curve is obtained according to the results of a series of calculation for the
hull subjected to bending conditions with di!erent curvature angles. It is found that
the interaction curve is asymmetrical because the hull cross-section is not symmetrical
about the horizontal axis and the behavior of the structural members under compres-
sion is di!erent from that under tension due to the non-linearity caused by buckling.
The angle of the resultant bending moment vector and that of the curvature vector are
di!erent in general cases. An interaction equation suitable for bulk carriers is pro-
posed based on the results of the analyzed 34,000 ton bulk carrier and another
74,000ton bulk carrier.
328 Y. Hu et al. / Marine Structures 14 (2001) 311}330
Acknowledgements
The authors wish to express their appreciation to the China Classi"cation Society
for supporting this research. However, any views in this paper are those of the authors
and do not necessarily re#ect the o\$cial views of the CCS.
References
[1] International Ship and O!shore Structure Congress. Report of Committee III.1. Proceedings of 11th
ISSC. Wuxi, China, 1991.
[2] International Ship and O!shore Structure Congress. Report of Committee III.1 Ductile Collapse.
Proceedings of 12th ISSC. St. John's, Canada, 1994.
[3] International Ship and O!shore Structure Congress. Reports of Committee II.1 Quasi-Static Re-
sponse and Committee III.1 (Ultimate Strength). Proceedings of 13th ISSC. Trondheim, Norway,
1997.
[4] Hughes OF. Ship structural design: a rationally-based, computer-aided optimization approach. New
York: Wiley, 1983.
[5] Chen YK, Kutt LM, Piaszczyk CM, Bieniek BP. Ultimate strength of ship structures. Trans SNAME
1983;91:149}68.
[6] Kutt LM, Piaszczyk CM, Chen YK, Liu D. Evaluation of the longitudinal ultimate strength of
various ship hull con"gurations. Trans SNAME 1985;93:33}53.
[7] Ueda Y, Rashed SMH. The idealized structural unit method and its application to deep girder
structures. Comput Struct 1984;18(2:):227}93.
[8] Paik JK. Ultimate hull girder strength analysis using idealized structural unit method. Proceedings of
p. 778}791.
[9] Smith CS. In#uence of local compressive failure on ultimate longitudinal strength of ship's hull.
Proceedings of the International Symposium on Practical Design in Shipbuilding. Tokyo, 1977.
p. 73}79.
[10] Frieze PA, Lin YT. Ship longitudinal strength modelling for reliability analysis. Proceedings of the
Marine Structural Inspection, Maintenance and Monitoring Symposium, SSC/SNAME, Arlington,
VA, 1991: III.C. p. 1}20.
[11] Mansour A, Wirsching P, Luckett M, Plumpton A et al. Assessment of reliability of ship structures.
SSC-398, Ship Structure Committee, 1996.
[12] Billingsley D W. Hull girder response to extreme bending moments, Proceeding of the "fth STAR
Symposium, SNAME, Coronado, California, 1980. p. 51}63.
[13] Dow RS, Hugill RC, Clarke JD, et al. Evaluation of ultimate ship hull strength. Proceedings on
Extreme Loads Response Symposium, SNAME, 1981. p. 133}47.
[14] Adamchak JC. An approximate method for estimating the collapse of a ship's hull in preliminary
design. Proceedings of the Ship Structure Symposium, SNAME, Arlington, VA, 1984. p. 37}61.
[15] Rutherford SE, Caldwell JB. Ultimate longitudinal strength of ships: a case study. Trans SNAME,
1990;98:441}71.
[16] Frieze P A, Lin Y T. Ship longitudinal strength modelling for reliability analysis. Proceedings of the
Marine Structural Inspection, Maintenance and Monitoring Symposium, SSC/SNAME, Arlington,
VA, vol. III.C. 1991. p. 1}20.
[17] Yao T, Nikolov PI. Progressive collapse analysis of a ship's hull under longitudinal bending. J Soc
Naval Arch Japan, 1991;170:449}61.
[18] Beghin D, Jastrzebski T, Taczala M. RESULT-A computer code for evaluation of the ultimate
longitudinal strength of hull girder. Proceedings of the sixth International Symposium on Practical
Design of Ships and Mobile Units, 1995. p. II832}43.
[19] Hansen AM. Strength of midship sections. Marine Struct 1996;9:471}94.
Y. Hu et al. / Marine Structures 14 (2001) 311}330 329
[20] Gordo JM, Guedes Soares C, Faulkner D. Approximate assessment of the ultimate longitudinal
strength of the hull girder. J Ship Res 1996;40:60}9.
[21] Sun J, Hu Y. A simpli"ed method to estimate the ultimate longitudinal bending moment of ship hulls
and development of the computer code. Shipbuilding of China, 2001, to be published.
[22] Hu Y, Sun J. An approximate method to generate average stress-strain curve with the e!ect of residual
stresses for rectangular plates under uniaxial compression in ship structures. Marine Struct
1999;12(9}10:):585}603.
[23] Faulkner D. A review of e!ective plating for use in the analysis of sti!ened plating in bending and
compression. J Ship Res 1975;19:1}17.
[24] Valsgaard S. Numerical design prediction of the capacity of plates in biaxial inplane compression.
Comput Struct 1980;12:729}39.
[25] Gordo J M, Guedes Soares C. Approximate load shortening curves for sti!ened plates under uniaxial
compression. In: Faulkner, D., Cowling, M.J., Incecik, A., Das, P.K., editors. Integrity of o!shore
structures-5. EMAS, Arly, 1993. p. 189}211.
[26] Hu Y, Sun J. Simpli"ed method to estimate the ultimate longitudinal bending moment of ship hulls
and the computer program based on the simpli"ed method. Research Report, Shanghai Rules and
Research Institute, China Classi"cation Society and School of Naval Architecture and Ocean
Engineering, Shanghai Jiaotong University, China, 1999 [in Chinese].
[27] Hu Y, Sun J. An approximate method to generate the load-end shortening curve for longitudinally
sti!ened panels in the beam-column type #exural buckling mode. First Chinese Conference on Ship
and Ocean Engineering, Shanghai, China, August 2000 [in Chinese].
[28] Hu Y, Chen B, Sun J. Tripping of thin-walled sti!eners in the axially compressed sti!ened panel with
lateral pressure. Thin-Walled Struct 2000;37(1):1}26.
[29] Faulkner D. Compression strength of welded grillages. In: Evans, J.H., editor. Ship structural design
concepts, Centreville, Maryland, USA: Cornell Maritime Press, 1975.
[30] Mansour A E, Lin Y H, Paik J K. Ultimate strength of ships under combined vertical and horizontal
moments. Proceedings of the Sixth International Symposium on Practical Design of Ships and
Mobile Units, (PRADS'95), vol. II, 1995. p. 844}56.
[31] Gordo JM, Guedes Soares C. Collapse of ship hulls under combined vertical and horizontal bending
moments. Proceedings of the Sixth International Symposium on Practical Design of Ships and
Mobile Units, (PRADS'95), vol. II, 1995. p. 808}19.
[32] Gordo JM, Guedes Soares C. Interaction equation for the collapse of tankers and containerships
under combined bending moments. J Ship Res 1997;41(3):230}40.
330 Y. Hu et al. / Marine Structures 14 (2001) 311}330