You are on page 1of 6

Proceedings of the ICCGS 2016

15-18 June, 2016


University of Ulsan, Ulsan, Korea

Nonlinear finite element analysis for ultimate hull girder strength of


container ship
Cheol-wung Yeom1), In-sik Nho2), Jung-woo Moon1)
1)
Korean Register, Busan, Korea
2)
Professor, Dept. of Naval Architecture & Ocean Engineering, Chungnam Univ. Daejeon, Korea

The present study had focused two points. One is


Abstract confirming of availability on the simplified method in CSR-
H for the container ship. Another is evaluating the partial
The ultimate hull girder capacity of container ship has been safety factor reflected a local bending effect of double
focused due to the recent accidents. Generally, Smiths bottom under hogging condition. For these, the nonlinear
method is well known as the simplified method to access finite analyses were carried out for the two fine meshed hold
rapidly for ultimate hull girder capacity except nonlinear models with regard to 8,000 and 13,000 TEU sized ships.
finite element analysis, hereinafter NL-FEA. This simplified Also, the actual loading conditions including the lateral
method, however, has been adopted recently to apply to loads like sea-pressure and cargo are applied. Firstly, the
container ship in accordance with UR S11A originated from ultimate hull girder strength, Mu_csr, was calculated by
IACS Common Structural Rules, hereinafter CSR-H. The using the simplified method as shown in Table 1.
targets of this study based on NL-FEA are to verify the Ms+1.2*Mw in the table is from CSR-H for the
effectiveness of this simplified method for container ships convenience of comparing (Ms in this Table is the static
ultimate hull girder strength and to propose the safety factor bending moment under homogenous loading condition, Mw
considering the local bending in double bottom structures means wave induced bending moment and the coefficient,
due to out of plane loads. Two different sized ships and 1.2, means the partial safety factor adopted for assuming the
three loading conditions which are pure bending, homo- extreme condition). Based on the results, both ships have
loading and alternate (one-bay empty) condition were over 150% of enough redundancy comparing with the
considered for this study. The results of present study required longitudinal strength in CSR-H.
showed that CSR-Hs simplified method can be available
for the ultimate hull girder strength of container ship and Table 1: Ultimate hull girder strength (hogging) for container
over 1.2 of safety factor should be applied in order to ship
consider the local bending effect in double bottom structures Mu_CSR / (Ms + 1.2*Mw)
due to out of plane loads such as sea pressure and cargo.
Ship A (8,800 TEU) 1.52
Introduction Ship B (13,000 TEU) 1.53

The verification of the longitudinal hull girder ultimate Generally, the simplified method called as Smiths Method
strength capacity of the container ship is getting more can calculate rapidly the ultimate hull girder strength
focused due to recent accidents, MSC Napoli from MAIB regardless ship-type because this method uses the pre-
report and MOL Comport from NK report. Furthermore, the defined stress-strain relationship of one-bay stiffened panels
effects of whipping and springing should be considered which form a mid-ship section. However, the accuracy of
according to the tenfigdency that hull size is going to be the results from this method depends on the how to define
bigger increasingly. IACS released the new regulation exactly the stress-strain relationship. Furthermore, this
known as UR S11A for the ultra large container ship in method cannot reflect the out of plane load like self-weight,
2015. It has included the provision regarding the simplified cargo, hydrostatic pressure and wave induced pressure
method for ultimate hull girder strength which is the same as except compression and tension due to hull girder bending.
in Harmonized Common Structural Rules for Bulker and In other words, it cannot consider the local bending behavior
Tanker, hereinafter CSR-H, to calculate the hull girder of double bottom due to lateral loads causing a decrease of
capacity of container ship. ultimate strength. Therefore, CSR-H give 1.1 ~ 1.25 of the

319
partial safety factors considering local bending effect. hogging moment.
Except the simplified method mentioned above, only
nonlinear finite element analysis, hereinafter NL-FEA, is
available to calculate the ultimate capacity. Generally, NL-
FEA can be applied the simulation of damage caused by
buckling and plastic deformation because of including
geometric and material nonlinearities of structural members.
Accordingly, NL-FEA would be appropriate approach in
Ship A
order to simulate the hull girder collapse behavior under
extreme bending moment. Also, this approach spends the
expensive cost due to making a very fine-meshed model and
huge iterative calculation in order to explain structural
nonlinear behavior.
NL-FEA can be divided into two methods, the implicit
method like as NASTRAN Sol 600 and the explicit method
as LS-DYNA-explicit version. The implicit method needs Ship B
the inverse matrix of stiffness at every incremental iteration Fig. 1: Finite element model (1/2 + 1 + 1/2 hold)
step and is used to apply with the static and quasi-static
problem. The explicit method needs only the inverse matrix Also, the fine meshed region in way of longitudinal
of mass at initial step during incremental procedure and is direction was three transverse web frame space from the
used to apply with the dynamic problem likewise impact or support bulkhead to the water-tight bulkhead in mid-hold
collision. The buckling and post-buckling problem due to because buckling due to local bending of double bottom
hull girder bending is apparently in the category of static might be occurred easily in this region. The minimum edge
and quasi-static problem. Accordingly, the implicit method length of element in fine meshed area was about from 50 to
should be applied for such problems, refer to PRADS 2013. 100 mm. The most cut-outs, even slot-hole, were included in
By the way, it has been known as diverging easily in case of the fine meshed area. The size of element for other region of
the very large finite element model, more than 1,000,000 finite element model was the spacing between stiffeners.
degree of freedoms. Actually, two commercial implicit The more 5 elements with expanding of their edge length
codes had applied in this study but the both could have not were arranged in the transition area from fine meshed area
solved continuously and stopped finally before the expected to coarse meshed area in order to avoid unexpected large
loading step despite the several attempts including mesh size deformation and drastic change of stress. Total degree of
control, geometrical changes and managing loading step size, freedom of each finite element model in this study was more
etc. The reasons may be from an unstable matrix solver or than 1,800,000.
computing machine. Therefore, the explicit method, known Only two type elements, 4-node shell element and 2-node
as with good convergence but lower accuracy due to beam element, were used for the modeling. 4-noded fully
application of reduced integration scheme and others, was integrated shell element recommended from LS-DYNA was
used to calculate the ultimate hull girder strength instead of used for all members including plate, stiffener web and
the implicit method in this study. If with appropriate total flange in the finite element model. This element can provide
analysis time without drastic loading change, the explicit the piecewise linear hardening for elasto-plastic behavior. 2-
method may be a good alternative for the static or quasi- noded beam element, known as Belytschko-Schwer
static problems. The one of recent studies has actually used resultant beam element, is limited in the elasto-perfect
the explicit code, refer to NK report, 2014. The present plastic behavior unfortunately, so that this element was used
study went after based on this report. for stiffener only in coarse meshed area.

Finite element model Initial imperfection and material nonlinearity


As shown in Fig. 1, the longitudinal range of finite element Most ships are not perfect geometrically due to weldments
model was 4 bays located in amid-ship in order to apply the depending on the building skill. In order to reflect the actual
out of plane loads under the homogeneous loading or status of structures, the initial imperfection should be
alternate (one-bay empty) loading condition. The transverse included into the fine meshed finite element model when
range of model was portside half-model considering the NL-FEA is carried out. Actually, the geometrical initial
symmetry in way of breadth to reduce calculating time. The displacement could cause easily the buckling phenomena
fine meshed region in way of vertical direction was limited which cannot be occurred in the perfect model under simple
to the below neutral axis because the container ship is compression.
operated typically in hogging condition and the compression
in the region of the below neutral axis is occurred due to

320
Such imperfection can be divided into two types, local and For material nonlinearities, the Rule minimum yield stress
global in the stiffened panel. The local imperfection might and true-stress and true strain relationship was idealized as
be introduced by lowest elastic bucking mode, hungry horse piece-wise linear in accordance with European Standards
shape or sinusoidal half-wave shape for local plate between EN-10025 as shown in Fig. 3. The failure strain for rupture
longitudinal stiffeners. In this paper, as shown in left upper was not applied because it has not effect on the ultimate
part of Fig. 2, four sinusoidal half-waves between transverse strength regardless with the applying of failure strain
webs were used because the hungry horse shape can resist
more than sinusoidal shape under the compression and the
lowest elastic buckling mode shape for the initial
imperfection is very difficult to make in case of a quite
complex large finite element model. s/200 in size as
maximum initial deflection (s means a space between
stiffeners) for local imperfection was used in accordance
with IACS Recommendation 47. In case of global
imperfection, one sinusoidal half-wave for the entire
stiffened panel was introduced as shown in right upper part
of Fig. 2. Tripping of web and flange according to global
imperfection of stiffened panel was also initialized with
sinusoidal half-wave shape for lateral bucking mode. L/1000
in size as maximum initial deflection was used for global Fig. 3: Linear piece-wise curve for true stress-true strain
relationship
imperfection. L means a stiffeners span. The simple
equations described in IACS Procedure, 2009 were used for
establishing the sinusoidal half-wave shape. As shown in Boundary and loading conditions
lower part of Fig. 2, total initial imperfection combined with
local and global was applied to the bottom shell, inner For hull girder being, the boundary condition can be a
bottom shell, girders, longitudinal bulkhead and side shell. simple support based on the both ends linked with rigid
body element or cantilever system based on that the aft end
is clamped while the forward end is free linked with rigid
body element. As shown in Fig. 4, the cantilever system
was applied with the symmetry condition in way of
transverse direction at the center line considering the half
model and the establishment of constraints in way of
longitudinal direction at the longitudinal strength member
Local initial imperfection Global initial imperfection in this study.
While ship is operating, the all sectional loads which are
shear forces and moments are different at the each section
but all forces throughout the whole ship length are to be in
the equilibrium. In case of the limited model with 2 or 3
hold length, the unnecessary additional moment due to
imbalance reduced from different sectional load in way of
Section view longitudinal direction would be happened, so that the
Fig. 2: Section view for initial imperfection cancellation of this addition moment should be needed. If
the additional moment due to imbalance is hogging moment
In general, the initial stress is originated from welding under the hogging condition and the structural response
residual stresses and have on effect of decreasing the instead of the applied load is utilized to get the ultimate
ultimate strength. But, the decreasing amount has not been capacity, the cancellation of the additional moment would
known exactly. 2 ~ 15% of decreasing was reported by not be necessary. The structural response means the ultimate
Gannon et al, 2012. Though the initial stress should be capacity obtained from the integration of longitudinal
included in finite element model as an input, it was not stresses of all longitudinal elements at the collapsed section
possible to input manually all initial stresses into the as shown in Fig. 5. Also, the integration means the
complex large finite element model, so that the present summation of longitudinal stresses multiplied by the
study didnt include the initial stresses. The application of rotational arm between the vertical location of each element
initial stress could be going to try through further study. In and neutral axis of the section.
case of CSR-H, 1.1 of partial safety factor considering
welding residual stress or uncertainties is applied to the
ultimate strength criteria.

321
is very short, the structure might be behaved dynamic
drastically not static or semi-static like hull girder bending.
Also, if loading time is long, the analysis time could be
wasted because the time increment depends on the minimum
edge size of element. Therefore, the convergence test should
be needed in order to establish the appropriate total loading
time. As shown in Fig. 7, the several loading times from 0.5
second to 3 second for hull girder bending only were tested
Fig. 4: Boundary condition for the model. In the Fig. 7, the curvature was from the
rotation of rigid plane at the free end of the model and the
This approach is more accurate than the approach based on moment was from the incremental applied load. Based on
the maximum applied load. Therefore, the cancellation was the result, the good convergence shows when the required
skipped and the pure-bending moment was applied in the time for bending moment is near 3 second. Accordingly, the
present study. Actually, the additional moment due to loading time for the static and dynamic lateral load was
imbalance was smaller than 5% of Rule design hull girder allocated with 1 second each and the loading time for static
capacity. moment, wave induce moment and maximum target
moment was allocated with 1 second each.
1.4

Mu / Mu (2 sec. loading time)


1.2

1.0

0.8

0.6 Total time = 0.5 sec


Total time = 1.0 sec
Fig. 5: Longitudinal location for structural response calcula- 0.4
tion Total time = 2.0 sec
0.2
Total time = 3.0 sec
The loading conditions for NL-FEA were pure-bending, 0.0
homogeneous and alternate conditions as shown in Fig. 6. 0.0E+00 1.0E-04 2.0E-04 3.0E-04 4.0E-04 5.0E-04
The pure-bending condition was needed to compare with the Curvature (1/m)
result from the simplified method in CSR-H. The other two Fig. 7: Convergence test for total applying time based on
condition were considered to verify the partial safety factor applied moment
considering the local bending effect in double bottom and
originated from UR S11A. Sea pressure, one of the lateral Results
loads, was applied with static and dynamic term according
to Rule regulations. Cargo load was actual containers As shown in Fig. 7, the ultimate point cannot be found out
weight including the stacked containers on the hatch cover. in the moment-curvature curves because that the moment
To magnify the local bending effect, the ballast water or were based on the incremental applied load. In contrast to,
consumable weights in double bottom was removed. The the ultimate point can be apparently showed in Fig. 8 when
loading was applied with static lateral load, static moment, the structural response was utilized.
wave induced pressure, wave induced moment and the
incremental moment in order. 1.2

1.0
Mu (FEA) / Mu (CSR-H)

0.8

0.6

0.4
Ship A
0.2
Ship B
0.0
Fig. 6: Homogeneous loading and one-bay empty condition 0.0 0.5 1.0 1.5 2.0 2.5 3.0
Time (Sec)
As mentioned earlier, NL-FEA was carried out by the Fig. 8: Moment-time curve based on structural response under
explicit approach. If total loading time in explicit approach pure-bending condition

322
The pure-bending analysis was carried out for 3 second of 1.2
the applied loading time with 120% of ultimate hull girder
moment based on CSR-H as the maximum target moment. 1.0

Mu (FEA) / Mu (CSR-H)
Fig. 8 shows that the ultimate capacities by NL-FEA for the
hold model with initial imperfection are nearly the same as 0.8
those of CSR-H, so that the effectiveness of simplified
method in CSR-H was proved under pure-bending condition. 0.6
The numerical values of ultimate capacity are showed in Pure-bending
Table 2. 0.4
Fig. 9 shows the von-Mises stress contour and deformed Homo-loading
shape of the target ships when the ultimate hull girder 0.2
Alternate-Loading
strength was reached under the homogeneous loading.
0.0
0.0E+00 1.0E-04 2.0E-04 3.0E-04 4.0E-04 5.0E-04 6.0E-04 7.0E-04
Curvature(1/m)
Fig. 10: Moment-curvature curve for Ship A
1.2

1.0

Mu (FEA) / Mu (CSR-H)
0.8

0.6

0.4
Pure Bending
(Ship A)
0.2

0.0
0.0E+00 1.0E-04 2.0E-04 3.0E-04 4.0E-04 5.0E-04 6.0E-04 7.0E-04
Curvature (1/m)
Fig. 11: Moment-curvature curve for Ship B

In case of Ship A, the ultimate capacity was decreased about


13 ~ 22% in comparison with the ultimate capacity of pure-
bending condition. In case of Ship B, the decrement of the
ultimate capacity under the actual loads was about 20 ~ 26%
in comparison with the capacity of pure-bending condition.
As shown in Figures, the slop of curves before the ultimate
(Ship B) point was oscillated not straight. The reason may be from
Fig. 9: von-Mises stress contour & deformed shape at the that the explicit method is basically dynamic analysis. The
ultimate hull girder strength for homo-loading condition NL-FEA was spent more 90 hours for each loading
condition at the Intel i7 4-cores parallel processing
The plastic deformation had occurred in the fine-meshed equipment. The detail values of ultimate capacity are
area with initial imperfection between water-tight bulkhead showed in Table 2.
and support bulkhead. It had started from the bottom shell to
the side shell via the bilge plate and the progressive flow of Table 2: Summary of NL-FEA results
plastic deformation had occurred similar in both ships. Ship A Ship B
Before the target moment, these analyses had stopped and (8,800TEU) (13,000TEU)
diverged due to the excessive plastic deformation and
Pure
increasing of curvature even if the explicit method known as 0.987 0.988
Mu_FEA Bending
a very stable scheme.
Fig. 10 and 11 show the structural response based moment / Homo-
0.862 0.799
and curvature curves for all loading conditions. Mu_CSR- Loading
H Alternate-
0.783 0.744
Loading

323
Conclusions References

In the present study, the explicit NL.FEA was used to Marine Accident Investigation Branch (MAIB). 2008.
calculate the ultimate hull girder capacity of container ship. Report No 9, 2008 United Kingdom, Report on the
In case of pure bending condition, the ultimate capacity investigation of the structural failure of MSC Napoli.
from NL.FEA and CSR-H was almost same. Therefore, the United Kingdom: MAIB.
CSR-Hs simplified method can be proved to apply the Nippon Kaiji Kyokai (NK). 2014. Investigation Report on
container ship. Also, in order to get more exact ultimate hull Structural Safety of Large Container Ship. Japan: NK.
girder capacity, the structural response based capacity was Darie, I., Roerup, J. and Wolf, V. 2013. Ultimate strength of
relevant instead of the incremental applied load for the a cape size bulk carrier under combined global and local
calculation of ultimate capacity while explicit analysis is loads. Proceedings of the PRADS pp.1173-1180, 20-25
utilized. The reduction of the capacity under homogenous Oct, CECO, Changwon. 2013.
loading condition was from 13% to 20% in comparisons LSTC, 2006. LS-DYNA Userss Manual.
with the capacity form the simplified method, so that the Gannon, L., Liu, Y. Pegg, N. and Smith, M.J. 2012.
about 1.2 of safety factor in case of the homogeneous Effect of Welding-Induced Residual Stress and
loading should be necessary to introduce. For case of the Distortion on Ship Hull Girder Ultimate Strength.
alternate (one-bay empty) condition, the reduction was from Marine Structures, 28(1), pp.2549.
22% to 26% and over 1.25 of safety factor should be applied. European Standards, EN-10025. 2004. Structural Steels.
The present study had carried out for only two different Part 2, 3, 4 and 6. European Standards: Czech Republic.
sized ships is not enough to define the safety factor. Further IACS, Harmonization Project Team 2. 2009. Non-linear
studies for more than 10 container ships sized differently Finite Element Collapse Analyses of Stiffened Panels
should be done in order to define the more exact factors. Procedure Description, www.iacs.org.uk
IACS. 2013. Recommendation No. 47, Shipbuilding and
Repair Quality Standard Rev. 7.
IACS. 2014. Common Structural Rules for Bulk Carriers
and Oil Tankers Rev. 1.
IACS. 2015. Unified Requirements, Strength of ships, S11A
(UR S11A)

324

You might also like