You are on page 1of 21

International Shipbuilding Progress 55 (2008) 87–107 87

DOI 10.3233/ISP-2008-0040
IOS Press

Benchmark study on the use of simplified structural


codes to predict the ultimate strength of a damaged
ship hull

C. Guedes Soares a,∗ , R.M. Luís a , P. Nikolov b , J. Downes c , M. Taczala d ,


M. Modiga e , T. Quesnel f , C. Toderan g and M. Samuelides h
a Centre for Marine Technology and Engineering, Instituto Superior Técnico, Technical University
of Lisbon, Lisbon, Portugal
b Technical University of Varna, Varna, Bulgaria
c University of Newcastle, Newcastle, UK
d Technical University of Szczecin, Szczecin, Poland
e University “Dunarea de Jos” of Galati, Galati, Romania
f Principia Marine, France
g University of Liége, Liége, Belgium
h National Technical University of Athens, Athens, Greece

The objective of the paper is to evaluate the ability of simplified structural analysis methods, based on
the Smith formulation to predict the ultimate strength of a damaged ship. Such methods are now widely
accepted as a reliable and fast way to obtain the longitudinal strength of an intact ship. In order to extend
these methods to damaged ships, first a benchmark study on the intact ship was performed in order for
the differences between the methods to be evaluated. Afterwards the methods are applied to the same
ship section but with damage, which was defined by removing the structural elements from the affected
areas. These, in turn, were obtained from a previous study in which a collision was simulated using a
finite element model. Results obtained for the ultimate strength were compared against each other and
with the results of the finite element analysis. Aside from some exceptions, the results of the approximate
methods agreed well with each other for the intact and damaged conditions. The simplified methods
are more conservative than the finite element analysis in hogging while they seem to give a very good
approximation to the result for sagging with some of them overestimating this value.
Keywords: Ultimate strength, damaged hull girder, Smith method, longitudinal strength, vertical bending
moment

1. Introduction

Ships have been designed during many years on the basis of their section modulus,
which governs the initial yield stress at deck or bottom. However, the ultimate lon-
gitudinal strength of ship hulls is the governing parameter for ship structural design
and the reference value to assess its reliability.
* Corresponding author. E-mail: guedess@mar.ist.utl.pt.

0020-868X/08/$17.00 © 2008 – IOS Press and the authors. All rights reserved
88 C. Guedes Soares et al. / Benchmark study on the use of simplified structural codes

Caldwell [5] was the first to propose the assessment of the plastic collapse mo-
ment of the midship section. He idealized the ship section made of reinforcements
and associated plates as composed of plates with equivalent thickness and calcu-
lated the fully plastic moment under the assumption that the plates that would be
under compressive loads would not collapse in buckling. Caldwell’s approach was
an idealized one as in general the failure of the compressed elements will occur be-
fore full collapse of the section which in this case is not able to reach its full plastic
moment.
Smith [21] was probably the first one that dealt appropriately with this question
by considering that a midship section could be modeled as an assembly of plates and
stiffeners, which would behave independently of each other. As the section rotates
under the action of a bending moment the various elements will be subjected to dif-
ferent strains, developing stresses that will contribute to the overall bending moment
that the section is being subjected to. This method has proved a good approximation
and various authors have developed alternative approaches based on these general
principles.
There have been several studies comparing the performance of different methods
and one of the most significant one is the study made at ISSC [2]. This study has
allowed conclusions to be derived about the relative accuracy of the various methods,
applied to a representative midship section.
As the interest of considering accidental conditions at the design stage is increas-
ing, the question that has been raised is whether the present methods are also applica-
ble to damaged ship hulls and whether their performance for these sections is similar
to their performance for intact sections. The simplified methods had already been
applied to damaged sections [10] but no systematic study had been done to verify
their accuracy in those conditions.
This is the objective of the present study in which different codes have been used
to predict the ultimate strength of a damaged ship section where collapse strength
had been calculated by a non-linear finite element code.
The damaged section considered in this project is the one that resulted from colli-
sion studies performed by Voudouris et al. [22], which led to a damaged section after
simulation of a collision.
This damaged section has been adopted by various groups to calculate its ulti-
mate strength using their own code. However before this was undertaken, a round
of comparisons have been made using an intact section, to establish the basis for the
comparison on a case that was more predictable. This benchmark calculation also
allows a reference about the differences among the predictions of the codes, which
can be compared with the differences in the case of the damaged section.
Next section describes briefly the main characteristics of the codes used, Section 3
describes the structure considered and Sections 4 and 5 present the results for the
intact and damaged ship respectively.
C. Guedes Soares et al. / Benchmark study on the use of simplified structural codes 89

2. Simplified methods

As indicated before, the methods used in this work are based on the formulation
developed by Smith [21], which allows the calculation of the progressive collapse
of the hull section. The method makes a series of assumptions in order to simplify
calculations and although some of these can be questionable, it still presents the best
relationship between speed of calculation and accuracy, being an approach that can
easily be used in practice. The main difference between the methods used in this
work and Smith method is that while Smith originally used finite element analy-
sis to derive the stress–strain relationships for the elements, the simplified methods
presented here use analytical approaches to derive that relationship.
Smith’s method considers that the ship behaves under longitudinal bending like a
beam and therefore the Euler beam theory, extended to the elasto-plastic range can
be applied. The method makes the following assumptions:
• The cross-section can be divided into several elements, usually composed by a
stiffener and associated plate, which behave independently;
• The section remains planar and perpendicular to the midsurface when the cur-
vature increases;
• Overall collapse is avoided by sufficiently strong transverse frames, which im-
plies that collapse must occur between them.
Additionally, for the damaged ship, the principal axis of the cross-section is rotated
comparing to the position of the principal axis for the intact ship. Therefore, strictly
speaking, it should be necessary to consider combined bending under the action of
the vertical bending moment, contrary to the case of the intact ship where the vertical
bending moment induces vertical bending exclusively. However the angle of rotation
is small and it is assumed that the principal axes in both cases coincide, even though
the cross-section is no longer symmetrical.
The moment–curvature relationship is calculated by imposing an incremental cur-
vature around the actual neutral axis of the hull section. From the imposed curvature,
one can obtain the strain in each element of the ship, and from the stress–strain re-
lationship of each element, the stresses can be extracted. Multiplying the stresses by
the element area one gets the force at each element. By considering the resultant of
the forces (which should be zero) one can recalculate the position of the instanta-
neous neutral axis, in an iterative process that continues until the total force equals
zero. Integrating the stresses over the cross-sectional area it is possible to obtain the
moment for the given curvature. This complete process continues until one reaches
the ultimate moment (or until one stops it) following the algorithm in Fig. 1, which
has the following steps:
1. Divide the section into individual elements.
2. Derive the stress–strain relationships for each element.
3. Calculate the position of the neutral axis and impose an initial curvature.
90 C. Guedes Soares et al. / Benchmark study on the use of simplified structural codes

Fig. 1. Smith method algorithm.

4. Calculate the strain in each element and obtain the stress using the previously
defined stress–strain relationships.
5. Calculate the total force on the section.
6. If the total force is not close to zero, adjust the position of the neutral axis and
return to step 4. If the force is close enough to zero move to step 7.
7. Calculate the total moment for the given curvature.
8. If the moment obtained for this curvature is not bigger than the previous one,
increase the curvature and go back to step 4. If the moment is smaller than the
previous one go to step 9.
9. The ultimate moment is the one obtained for the previous curvature, which is
also the maximum.
C. Guedes Soares et al. / Benchmark study on the use of simplified structural codes 91

The methods used in this work adopted this general approach and are briefly de-
scribed here.
The HULLCOLL program uses the method developed by Gordo et al. [11], in
which the behavior of the elements used is the one proposed by Gordo and Guedes
Soares [9]. This approach builds upon the formulation proposed by Faulkner [7] for
plates and stiffened panels [8], which was appropriately generalised for this appli-
cation. The method considers always the lowest stress associated with one of the
following three failure modes: plate failure, flexural column failure and tripping fail-
ure of stiffeners. Because in simplified methods the elements are considered to be-
have independently, corner elements behavior can not be satisfactorily accounted
for, because it is influenced by the buckled state of the adjacent elements. The so-
lution that is generally considered less unsatisfactory, and used in the program, is
to model them as hard corners, which are elements that when compressed have a
perfect elasto-plastic behavior.
The software HULLST uses simplified analytical approaches for plate panels and
stiffeners with associated plating to derive the average stress–average strain relation-
ships. Firstly, σ–ε relationships of the plate panels between stiffeners are derived
combining elastic large deflection analysis and rigid plastic mechanism analysis [23,
24]. In the elastic analysis, the influences of initial deflection and welding resid-
ual stress are taken into account. A simplifying assumption in this approach is that
among the many deflection components only one is important for the plating col-
lapse behavior. Secondly, σ–ε relationships for components made of stiffener with
attached plating are derived considering the equilibrium of forces and moments [23].
In this derivation, σ–ε relationships of the plate panels between stiffeners are used
to represent the effectiveness of the attached plating. The flexural–torsional buckling
and yielding in stiffeners are taken into account [25].
The computer code RESULT is a module of MARS program, developed by Bu-
reau Veritas for calculation of rule scantlings of members contributing to the longi-
tudinal strength. The values of the critical buckling stresses for plate, beam-column,
flexural–torsional (tripping) and local failure modes are evaluated according to the
formulae given in the Bureau Veritas Rules, [4], in chapters “Ultimate Strength of
Ship Structures” and “Buckling Criteria”.
The program ULTIM uses analytically derived average stress–average strain rela-
tionships of elements. The post-collapse behavior of compressed elements incorpo-
rated in the current algorithm has followed the work of Adamchak [1] in principle
but with modified relationships of stress–strain from the elastic range to the post-
collapse region. The theory of plastic collapse mechanism has been implemented
relating end shortening with central deflection to define the stress–strain relation-
ship in the post collapse range, as was done by Adamchak, but the end conditions
are changed. Derivations presented in that paper are based on simply supported end
conditions of panels accounting for the effects of initial deflection and the eccentric-
ity caused by the loss of plate stiffness. These changes are much more compatible
92 C. Guedes Soares et al. / Benchmark study on the use of simplified structural codes

with Hughes [12], whose evaluation of the collapse stress of compressed panels was
followed.
In RULTIM most of the elements which compose the modeled section have the
same behavior: elastic perfectly plastic behavior in tension, and inelastic buckling
behavior in compression. Some of them use an elastic ideally plastic behavior in
both tension and compression, because of their location in reinforced zones (corner
elements) which limits buckling. Element critical loads are calculated differently,
following the geometrical and material properties of each of them. Stiffened panels
can buckle following two different ways: by panel bending (based on the Adamchak
formulation), or by a stiffener buckling. The most critical mode is used if the user
does not impose a specific one.
The LR.PASS methodology for determining the ultimate hull girder strength was
developed by Rutherford [18], for Lloyd’s Register of Shipping [13]. The software
suite contains various programs, including the two programs “20203 Hull Girder
Ultimate Strength” and “20202 Stiffened Compression Panels” which can be used
either independently or in combination with each other. The hull girder ultimate
strength program P20203 is based on Smith’s methodology [21]. This methodology
is outlined in some detail by Rutherford and Caldwell [19]. The Stiffened Compres-
sion Panels program P20202 uses a modified version of the methodology developed
by Paik and Lee [14] and Dowling [6] for determining the ultimate strength of stiff-
ened panels. This extended methodology was developed by Rutherford [18], to take
into account the analysis of plates under combined surface pressure and uniaxial in-
plane compressive loading. The program calculates two predictions of the ultimate
compressive strength of the stiffened panel, one for plate induced failure and the
other for stiffener induced failure. The lowest result is then used to select the most
appropriate load-shedding response beyond the ultimate strength level.
The PROCOL software uses a progressive collapse analysis (Smith method) to
compute the moment–curvature curve of a hull girder. The structure can be meshed
using 3 types of elements: beam-column (stiffeners or girders with attached plate),
plate or hard corner (elements supposed to collapse only by yielding – no buckling).
The material is supposed to have pure elasto-plastic behaviour. The “average
stress–average strain” curve for each element can be estimated using the following
four methods:
• The ultimate stress of the beam-column in compression is computed using an
empirical model [14,15]. The post-collapse curve is calibrated using four points;
• The ultimate stress of the beam-column in compression is computed using a
model proposed for steel structures [3,6]. This method takes into account the
initial deflection and the level of welding residual stress. The post-collapse
curve is calibrated in four points, based on the ISSC’2000 benchmark;
• The “average stress–average strain” curve is computed using the method pro-
posed in [12,17];
• The “average stress–average strain” is computed using an elasto-plastic large-
deflection analysis proposed by Yao and Nikolov [23].
C. Guedes Soares et al. / Benchmark study on the use of simplified structural codes 93

3. Cross-section and damage definition

The cross-section used comes from the ROPAX, MS DEXTRA, Fig. 2, which is a
fast ferry that was previously used on the project DEXTREMEL and therefore results
for a finite element analysis of the damaged ship are available in [22]. The study on
this ship included an initial analysis of ships collision using finite elements, where a
rigid bow was rammed into the side of a 39 m long model of the ship, as presented
in [20]. Therefore, one has the damage definition [20], and the ultimate strength of
the ship [22]. The ship main characteristics and the scantlings of the cross-section
are presented respectively in Table 1 and Fig. 3.
As explained previously, the ship section is modeled by dividing it into multiple
elements, where the total number of elements (n) depends on the methodology fol-
lowed and the software used. The modeling of the damaged cross-section was done
by identifying the damaged area of the hull girder and removing the corresponding
elements from the intact models.
For this purpose, information about the finite element model of the damaged ship
was taken from the results in [20], which are presented from Fig. 4 to Fig. 7. From
these results the damaged hull girder is chosen as the most damaged cross-section of
this model, its location is shown in Fig. 8. In this figure the height of the damaged
area is presented, and indicated as references are the positions of the main deck and
of the “L” reinforcement (which is referenced by its dimensions, “1000 × 12/300 ×
25”, in the cross-section in Fig. 3). The elements within the areas referred to as
“Intact Elements”, are still resisting the loads and therefore are kept in the model.
The elements outside these areas were considered to no longer resist to the loads and
were excluded from the models.
In Fig. 9 the models obtained for HULLCOLL, of the intact and damaged sections,
are presented side by side as an example; the models developed for other computer

Fig. 2. Profile view of the MS DEXTRA.

Table 1
Main characteristics of MS DEXTRA
Length PP 173.00 m
Breadth 26.00 m
Draft 6.50 m
Depth to upper deck 15.70 m
Speed 28 knots
94 C. Guedes Soares et al. / Benchmark study on the use of simplified structural codes

Fig. 3. MS DEXTRA cross-section.


C. Guedes Soares et al. / Benchmark study on the use of simplified structural codes 95

Fig. 4. Inside view of the damage at 0.15 s, equivalent stress.

Fig. 5. View of the damage at 0.20 s, equivalent stress.


96 C. Guedes Soares et al. / Benchmark study on the use of simplified structural codes

Fig. 6. Side view of the damage at 0.20 s, equivalent stress.

Fig. 7. Inside view of the damage at 0.20 s, equivalent stress.


C. Guedes Soares et al. / Benchmark study on the use of simplified structural codes 97

Fig. 8. Definition of the location of the damaged cross-section and of the heights of the damaged zones.

Fig. 9. Comparison between the intact section model and the damaged section model for HULLCOLL.
98 C. Guedes Soares et al. / Benchmark study on the use of simplified structural codes

codes are similar to these. One can easily notice the impact of the damage on the
cross-section. It is also possible to observe that one longitudinal bulkhead has dis-
appeared. Figure 4 shows that the damage has destroyed the connection between the
bulkhead and the main deck, soon after impact, rendering it ineffective to resist the
loading therefore one can simply remove it.

4. Results for the intact ship

The intact ship was first studied in order to evaluate how the different methods per-
formed in a more standard situation. Results were obtained with and without residual
stresses, which were taken as:

σr = 0.1σ0 , (1)

where, σ0 is the yield stress.


The shape and magnitude of the initial deflections were not specified in advance.
Each organization, based on their experience and the method used, has taken into
account their effect in different ways, which may be a source for differences.
The four different methods of PROCOL to derive the stress–strain relationship are
referenced as “Paik” [14,15], as “I.C.” [3,6], “Hughes” [12,17] and as “Yao” [23].
Two of the methods were used to calculate the linear elasto-plastic bending of the
ship, which is equal for both hogging and sagging in order to serve as reference for
analyzing the results.

4.1. Moment–curvature relationships

In Figs 10 and 11 the moment–curvature diagrams obtained for the hogging con-
ditions without and with residual stresses respectively are presented. The curves in
the linear region are very close to each other with the exception of PROCOL which
presents lower results. The areas around the ultimate bending moment and after col-
lapse are clearly different from each other due to the different stress–strain curves of
the elements obtained by the different methods.
In turn, Figs 12 and 13 present the moment–curvature relationships for the sag-
ging condition without and with residual stresses, respectively. In sagging the lin-
ear portions of all the curves seem to be in agreement between each other except
for PROCOL using the IC method. The differences between the results in the area,
around the ultimate bending moment and after collapse confirm what was observed
for hogging but the results obtained by PROCOL using the Hughes method and the
IC method seem to be much higher than other methods.
C. Guedes Soares et al. / Benchmark study on the use of simplified structural codes 99

Fig. 10. Moment–curvature relationship for the hogging condition without residual stresses.

Fig. 11. Moment–curvature relationship for the hogging condition with residual stresses.

4.2. Ultimate moment

Tables 2 and 3 present the ultimate bending moments obtained by each method
without residual stresses for the hogging and sagging conditions, respectively, where
COV is the coefficient of variation. The results confirm what was observed in the
100 C. Guedes Soares et al. / Benchmark study on the use of simplified structural codes

Fig. 12. Moment–curvature relationship for the sagging condition without residual stresses.

Fig. 13. Moment–curvature relationship for the sagging condition with residual stresses.

graphics: in the sagging condition there is a larger variability than for the results
in the hogging condition. In the hogging condition, PROCOL presents the biggest
difference in relation to the mean value but also the smallest one using the Hughes
method and the Paik method respectively. For sagging, the extreme values observed
C. Guedes Soares et al. / Benchmark study on the use of simplified structural codes 101

Table 2
Results for the collapse bending moment in the hogging condition without residual stresses
M (Nm) Differencea (%)
HULLCOLL 3.935E+09 −1.71
HULLST 4.222E+09 5.46
RESULT 3.802E+09 −5.04
RULTIM 3.613E+09 −9.77
LR.PASS 3.707E+09 −7.42
PROCOL (Paik) 3.942E+09 −1.54
PROCOL (I.C.) 4.163E+09 3.98
PROCOL (Hughes) 4.406E+09 10.05
PROCOL (Yao) 4.243E+09 5.98
Mean 4.00E+09
Variance 7.29E+16
Standard deviation 2.70E+08
COV 0.067
a Difference between the mean and each result.

Table 3
Results for the collapse bending moment in the sagging condition without residual stresses
M (Nm) Differencea (%)
HULLCOLL 2.071E+09 −23.47
HULLST 2.492E+09 −7.90
RESULT 2.183E+09 −19.33
RULTIM 2.344E+09 −13.37
LR.PASS 2.799E+09 3.42
PROCOL (Paik) 2.530E+09 −6.50
PROCOL (I.C.) 3.526E+09 30.30
PROCOL (Hughes) 3.455E+09 27.68
PROCOL (Yao) 2.954E+09 9.16
Mean 2.71E+09
Variance 2.73E+17
Standard deviation 5.23E+08
COV 0.193
a Difference between the mean and each result.

for PROCOL, with the IC method and the Hughes method are confirmed. HULLCOL
presents a very low value but it is probably due to the fact that the mean is increased
by the results of PROCOL.
Tables 4 and 5 present the results for the collapse moment with residual stresses for
the hogging and sagging conditions, respectively. The results obtained with residual
102 C. Guedes Soares et al. / Benchmark study on the use of simplified structural codes

Table 4
Results for the collapse bending moment in the hogging condition with residual stresses
M (Nm) Differencea (%)
HULLCOLL 3.799E+09 1.29
HULLST 4.045E+09 7.84
ULTIM 3.811E+09 1.60
RULTIM 3.453E+09 −7.93
LR.PASS 3.645E+09 −2.81
Mean 3.75E+09
Variance 4.80E+16
Standard deviation 2.19E+08
COV 0.058
a Difference between the mean and each result.

Table 5
Results for the collapse bending moment in the sagging condition with residual stresses
M (Nm) Differencea (%)
HULLCOLL 2.047E+09 −10.39
HULLST 2.422E+09 6.04
ULTIM 2.262E+09 −0.97
RULTIM 2.162E+09 −5.35
LR.PASS 2.528E+09 10.66
Mean 2.284E+09
Variance 3.75E+16
Standard deviation 1.94E+08
COV 0.085
a Difference between the mean and each result.

stresses have a higher variability in sagging, just as without residual stresses. In


the hogging condition the highest differences are around 8% while for the sagging
condition the highest differences are around 10%.
In hogging these are the structural elements in the lower part of the hull which are
compressed: outer and inner bottom, bottom girders and lower parts of the sides and
longitudinal bulkheads, while in sagging it is just opposite: and the decks are com-
pressed. In all procedures applied in the investigation the stress–strain relationship
follows the elastic–plastic behaviour of material for tension while the differences ap-
pear for compression. The substantial scatter of results for sagging indicates that the
differences between the applied procedures to evaluate response of stiffened plates
are more significant for thinner plating and slender stiffeners rather than for plating
of moderate thickness and stocky stiffeners typical for the bottom structure.
C. Guedes Soares et al. / Benchmark study on the use of simplified structural codes 103

5. Results for the damaged ship

After applying the damage defined in Section 3 by removing the affected elements
it is possible to calculate the ultimate bending moment for the damaged ship. The
damaged ship including the residual stresses was analysed for evaluation of stress–
strain relationships. Unfortunately, this option is not included in some of the codes;
therefore a smaller number of results is compared in this section. Nevertheless, the
results presented here give the opportunity for comparisons between some methods
and these can be extended in the future. Anyway, the number of calculations is too
small for properly analyzing the variability of the current simplified methods in pre-
dicting the ultimate moment of a damaged ship.

5.1. Moment–curvature diagrams

Figure 14 presents the moment curvature diagrams for hogging. The methods with
“RS” mean that residual stresses were used in the calculations; its value is given by
Eq. (1). The linear portion of the diagrams seems to be in agreement between all
methods. The results for the collapse bending moment seem to be close to each
other. The post collapse zone shows similar behavior in all methods, only the results
obtained with LR.PASS seem to drop below the other ones but it is still acceptable.
In Fig. 15 the moment–curvature diagrams are presented for sagging. The differ-
ences between the diagrams in the linear region are negligible. A higher variation
than in the hogging case is observed in the vicinity of the collapse bending moment.
In the post collapse zone the methods present some differences in behavior but these
are small; the main difference resides in their value.

Fig. 14. Moment–curvature diagrams for hogging.


104 C. Guedes Soares et al. / Benchmark study on the use of simplified structural codes

Fig. 15. Moment–curvature diagrams for sagging.

5.2. Ultimate moment

The results for the ultimate moment of the damaged section are to be com-
pared with the ones obtained from the finite element analysis of the damaged ship
(Voudouris et al. [22]). The finite element model is actually a piece of the cylindri-
cal body of the ship, sufficient to simulate the damage and to allow a finite element
analysis under the action of the vertical bending moment. The results obtained by
Voudouris et al. [22] are presented in Table 6.
In Table 7 the results obtained for hogging, using the simplified methods, are pre-
sented. The values obtained without residual stresses are further apart from the mean
then most of the results with residual stresses, only LR.PASS presents a higher dif-
ference. In Table 8 the results for sagging are presented. It is clear that, as for the
intact ship, the results for sagging are further away from the mean than the results
for hogging.
Table 9 presents a comparison between the results obtained by the finite element
method (FEM) and by means of the simplified methods (SM) for the cases without
and with residual stresses (RS). In hogging, the results obtained by the SM are al-
ways smaller than the ones from FEM. The introduction of residual stresses actually
increases the difference although only slightly, around 1%. In sagging the results ob-
tained by the SM are higher than the ones from FEM. In this case the introduction of
RS reduces significantly the difference, putting both results at a negligible distance
from each other.
The results were obtained with the use of only two methods; without experimental
results the conclusions reached here are obviously very limited. Looking at the re-
C. Guedes Soares et al. / Benchmark study on the use of simplified structural codes 105

Table 6
Results of the FEM analysis
Condition M (Nm)
Hogging 4.54E+09
Sagging 2.14E+09

Table 7
Results for the collapse bending moment in hogging
Hogging Without residual stresses With residual stresses

M (Nm) Differencea (%) M (Nm) Differencea (%)


HULLCOLL 3.77E+09 3.12 3.64E+09 0.58
HULLST – – 3.70E+09 2.32
ULTIM – – 3.63E+09 0.27
LR.PASS 3.55E+06 −3.12 3.51E+09 −3.17
Mean 3.66E+09 – 3.62E+09 –
a Difference between the mean and each result.

Table 8
Results for the collapse bending moment in sagging
Sagging Without residual stresses With residual stresses

M (Nm) Differencea (%) M (Nm) Differencea (%)


HULLCOLL 1.95E+09 −15.13 1.93E+09 −10.54
HULLST – – 2.21E+09 2.64
ULTIM – – 2.10E+09 −2.72
LR.PASS 2.64E+09 15.13 2.38E+09 10.62
Mean 2.29E+09 – 2.15E+09 –
a Difference between the mean and each result.

Table 9
Comparison between the mean of the results obtained using the simplified methods and the results ob-
tained by the finite element analysis
Condition FEM (Nm) SMa (Nm) SM–RSb (Nm) Diff. 1c Diff. 2d
Hogging 4.54E+09 3.66E+09 3.62E+09 24.04% 25.41%
Sagging 2.14E+09 2.29E+09 2.15E+09 −6.67% −0.60%
a Simplified method results for the damaged ship without residual stresses.
b Simplified method results for the damaged ship with residual stresses.
c Difference between results of the FEM and the simplified method.
d Difference between results of the FEM and the simplified method with residual stresses.

sults obtained it seems that the finite element method is less conservative in hogging,
than the simplified methods, while in sagging it seems to agree more with the mean
of the simplified methods that the actual methods.
106 C. Guedes Soares et al. / Benchmark study on the use of simplified structural codes

6. Conclusions

The simplified methods used in this study compare well with each other for the
calculation of the ultimate strength, giving differences that one has come to expect
from these types of calculations.
After careful analysis of the ultimate moment it seems that the results present
higher variation predicting the sagging moment rather than the hogging moment.
One should take this into consideration because the resistance to sagging is usually
the critical condition of ship hulls.
For the damaged ship, the finite element method seems to overestimate the mean
value of the simplified methods in hogging, while in sagging it gives a closer value
to the mean than the simplified methods themselves. However, two things must be
considered; first, the number of simplified methods used for the damaged ship is too
low, second, without experimental analysis it is impossible to say which method is
more precise. Therefore, this conclusion holds just for the specific case found in this
study and it is not possible to say which results are less conservative because it is
possible for it to be overestimating the actual resistance of the ship.
Considering the fact that no experimental analysis has been performed, one can
not say if the simplified methods are in fact a useful tool to calculate the resistance
of damaged ships. However, considering that simplified methods are accepted as
reliable tools for calculating the ultimate moment of intact ships even with the dif-
ferences shown in these results, and that the results obtained for the damaged ship
do not present bigger differences than the intact condition, then, it is possible to say
that simplified methods compare well among each other for the calculation of the
ultimate bending moment of damaged ships.

Acknowledgements

This paper has been prepared within the project “MARSTRUCT – Network of
Excellence on Marine Structures” (www.mar.ist.utl.pt/marstruct/), which has been
funded by the European Union through the Growth program under contract TNE3-
CT-2003-506141.

References

[1] J.C. Adamchak, Approximate method for estimating the collapse of a ship’s hull in preliminary
design, in: Proc. Ship Structure Symposium’84, SNAME, 1980, pp. 37–61.
[2] D.W. Billingsley, Hull girder response to extreme bending moments, in: Proceedings, 5th STAR
Symposium, SNAME, 1980, pp. 51–63.
[3] M.A. Bonello, M.K. Chryssanthopoulos and P.J. Dowling, Ultimate strength design of stiffened
plates under axial compression and bending, Marine Structures 6(5/6) (1993), 533–552.
[4] Bureau Veritas, Rules for the Classification of Ships.
C. Guedes Soares et al. / Benchmark study on the use of simplified structural codes 107

[5] J.B. Caldwell, Ultimate longitudinal strength, Trans. Royal Institute of Naval Architects 107 (1965),
411–430.
[6] P.J. Dowling, Design of flat stiffened plating: Phase 1 Report, CESLIC Report SP 9, Dept. of Civil
Engineering, Imperial College, London, 1991.
[7] D. Faulkner, A review of effective plating for use in the analysis of stiffened plating in bending and
compression, Journal of Ship Research 19 (1975), 1–17.
[8] D. Faulkner, J.C. Adamchack, M. Snider and M.F. Vetter, Synthesis of welded grillages to withstand
compression and normal loads, Computers and Structures 3 (1973), 221–246.
[9] J.M. Gordo and C. Guedes Soares, Approximate load shortening curves for stiffened plates under
uniaxial compression, in: Integrity of Offshore Structures 5, D. Faulkner et al., eds, EMAS, 1993,
pp. 189–211.
[10] J.M. Gordo and C. Guedes Soares, Residual strength of damaged ship hulls, in: Proc. 9th Interna-
tional Congress of International Maritime Association of the Mediterranean (IMAM2000), Ischia,
Italy, 2000.
[11] J.M. Gordo, C. Guedes Soares and D. Faulkner, Approximate assessment of the ultimate longitudi-
nal strength of the hull girder, Journal of Ship Research 4(1) (1996), 60–69.
[12] O.F. Hughes, Ship Structural Design: A Rationally-Based, Computer-Aided, Optimization Ap-
proach, Wiley, New York, 1983.
[13] Lloyd’s Register of Shipping, LR.PASS Personal Computer Programs User’s Manual: Vol. 2, Direct
Calculations, Lloyd’s Register of Shipping, London, 1997.
[14] J.K. Paik and J.M. Lee, An empirical formulation for predicting ultimate compressive strength of
plates and stiffened plates, Journal of Ship Research (1996).
[15] J.K. Paik and A.E. Mansour, A simple formulation for predicting the ultimate strength of ships,
J. Marine Science and Technology 1 (1995), 52–66.
[16] J.K. Paik, A.K. Thayamballi and S.H. Yang, Residual strength assessment of ships after collision
and Grounding, Marine Technology 35 (1998), 38–54.
[17] R.K. Rahman and M. Choudhury, Estimation of ultimate longitudinal bending moment of ships and
box girders, Journal of Ship Research 40(3) (1996), 244–257.
[18] S.E. Rutherford, Stiffened compression panels: The analytical approach, Report 82/26/R2, Hull
Structures Department, Lloyd’s Register of Shipping, London, UK, April 1984.
[19] S.E. Rutherford and J.B. Caldwell, Ultimate longitudinal strength of ships: A case study, The Trans-
actions of the Society of Naval Architects 98 (1990), 441–471.
[20] D.P. Servis and M. Samuelides, Ship collision analysis using finite elements, in: SAFER EURORO
Spring Meeting, Nantes, 28 April, 1999.
[21] C.S. Smith, Influence of local compressive failure on ultimate longitudinal strength of a ship’s
hull, in: Proceedings of the PRADS: International Symposium on Practical Design in Shipbuild-
ing, N. Zosengakkai, ed., Society of Naval Architects of Japan, Tokyo, Japan, 18–20 October, 1977,
pp. 73–79.
[22] G. Voudouris, D.P. Servis and M. Samuelides, Ultimate load calculations Finite Element Analy-
sis, DEXTREMEL Project (Contract No. BRPR-CT97-513), Document Ref. No. DTR-4.3-NTUA-
11.00, 2000.
[23] T. Yao and P.I. Nikolov, Progressive collapse analysis of a ship’s hull under longitudinal bending,
J. Soc. Naval Arch. of Japan 170 (1991), 449–461.
[24] T. Yao and P.I. Nikolov, Stiffness of plates after buckling, Trans. Kansai Soc. Naval Arch. 215
(1991), 137–146.
[25] T. Yao and P.I. Nikolov, Progressive collapse analysis of a ship’s hull under longitudinal bending
(2nd report), J. Soc. Naval Arch. of Japan 172 (1992).

You might also like