Sunteți pe pagina 1din 9

3rd International ASRANet Colloquium

10 – 12th July 2006, Glasgow, UK.

NUMERICAL MODELING OF SHIP COLLISION BASED ON FINITE


ELEMENT CODES
O. Ozguc, P.K. Das, N. Barltrop & M. Shahid, Universities of Glasgow and Strathclyde, UK
M. Samuelides, National Technical University of Athens, Greece

ABSTRACT
Non-linear finite element method (FEM) is a powerful tool for analysing ship collision problem and has
been seen more and more applications in the recent years. Expectedly, more and more FEM simulation
application will be seen in the coming years. Rapid advances in computer technology and software
capacity have made FEM simulation a viable choice [17]. Ship collision simulations still face some
difficulties in terms of providing reliable results. The accuracy of numerical modelling results
significantly depends on the proper definition of the phenomenon and careful control of the some critical
parameters such as rupture criteria, non-linearity, element type, friction coefficient, and mesh fineness.
Present paper attempts to investigate the parameters on the ship collisions and provides effective
guidelines on the implementation of the finite element codes.

NOMENCLATURE tN , tn Thickness of the N-th member of the


striking vessel or of the n-th member of the struck
le An individual element length vessel, respectively
ε P eff Effective plastic strain
1. INTRODUCTION
σo Initial yield stress
H The height of rupture aperture in the side Since early 1990’s, many predictive calculation
shell (m) and t is the side shell thickness (cm) procedures have been developed for predicting the
εe The necking strain ship responses in a collision or grounding. These
WC The total energy absorbed by damage and methods have matured to such a level that they are
now being integrated into systems for evaluation
striking and struck vessels in MJ and designs. Nevertheless, how to calculate the
εg The uniform strain ship responses in an accident continues to stand in
σy The yield function of plastic-kinematics the center of the recent research and development
material [17].
C , P Cowder – Symonds strain rate parameters
Analysis of the collision and grounding accidents
T1 max Design draft of the striking ship
can be classified into two parts, namely the
T1 min Ballast draft of the striking ship external dynamics and internal mechanics. While
T2 min Ballast draft of the struck ship the external dynamics deals with the rigid body
T2 max Design draft of the struck ship global motion of ships under action of the
collision or grounding forces and the
β Hardening parameter between 0
hydrodynamic pressures acting on the wet surface.
(kinematics) and 1 (isotropic) The internal mechanics includes evaluation of the
LN , Ln Length of damage for the N-th structural failure response of the involved ships
member of the striking vessel or for the n-th during the collision or grounding accident. Those
member of struck vessel, respectively two parts are often treated separately and in some
EP Plastic hardening modulus cases solved together [20].
PN , Pn Depth of damage for N-th member
of striking vessels or for the n-th member of struck The analysis methods of internal mechanisms can
vessel, respectively, be categorized into four groups: simple empirical
RT The damaged volume of structural steel in formulae, simplified analytical approach,
3 simplified FEM and non-linear FEM simulation.
m
t The plate thickness The internal mechanics involve yielding, tearing

1
or fracture. A non-linear FEM codes for used implicit methods. Converge of calculations is
simulating the structural behaviour needs to be much easier to realize. Explicit methodologies
fine mesh enough so as to capture such highly based computer codes include ABAQUS/Explicit,
non-linear characteristics [15]. DYTRAN, LS-DYNA, PAM-CRASH and
RADIOSS, and implicit methodologies based
Non-linear finite element (FEM) is a powerful tool codes include ABAQUS/Standard, ANSYS,
for analysing ship collision problem and has been MARC and NASTRAN.
seen more and more applications in the recent
years. Expectedly, many more FEM simulation There are mainly two important factors to consider
application will be seen in the coming years. in structural modelling of the non-linear FEM
Rapid advances in computer technology and analysis, namely element type and mesh fineness.
software capacity have made FEM simulation a A higher order element generally provides better
viable choice. Ship collision simulations still face accuracy and allows coarse mesh. But this requires
some difficulties in terms of providing reliable more computational effort because of the higher
results. The reliability of the numerical simulation order of the element being used. The importance
results largely depends on the proper modelling of of mesh fineness or element types has been
the phenomenon and realistic consideration of studied by many researchers including Kuroiwa
aspects, such as rupture criteria, non-linearity, [6], Kitamura [4], [5], Servis et al. [11], Ozguc et
element type, friction coefficient, and mesh al. [9]. It is observed that a very large number of
fineness. This paper investigates the parameters on elements are required in order to obtain accurate
the ship collisions and provides effective results for components deformed by axial crushing
guidelines on the implementation of FEM codes. It forces. Accounting for realistic size and boundary
will thus be possible to compare the conditions of FE models has been discussed by
crashworthiness of various structural arrangements Woisin [16].
under similar collision scenarios.
Application of non-linear FEM simulation has
2. NUMERICAL SIMULATION been the main theme of recent studies (Wu et al.
[17], Zhang et al. [20], Tornqvist & Simonsen [13],
Recent advances in computers and calculation Wang et al. [15], Alsos & Amdahl [1], Oh et al.
algorithms have made non-linear finite element [7], Yamada et al. [19]). Since structures behave in
analysis a viable tool for assessing the internal many complex patterns, many special modelling
mechanics of collisions. Two types of FE techniques are needed. Challenges involved in
methodologies are relevant, namely implicit and analysing such a high non-linear problem include
explicit techniques. Implicit methodologies obtain among others, structural contact, criteria for
solution by simultaneous solving of systems of material’s rupture, crack propagation (Wang et al.
equations needing frequent updating of the [15]). This trend was clearly demonstrated in the
stiffness matrix for nonlinear FE analysis. This 2nd and 3rd International Conference on Collision
places demands on the equation solver and the and Grounding of Ships.
computer capacity especially in terms of memory
resources and CPU cycles. Explicit codes obtain 3. RUPTURE CRITERIA
system solutions based on mass matrix that remain
constant not needing frequent updating as in the The modelling of rupture and tearing is the most
case of stiffness approach. Explicit method challenging task in applying the structural
however needs smaller time steps to comply with crashworthy concept. This also forms the crucial
stability requirement for equation solving. background for the important criteria of
crashworthy ships. Advanced FEM packages
To analyse a collision and grounding accident enable reliable automated simulation of the
involving high non-linearity and phenomena such structural failure process up to occurrence of
as colliding surfaces contact, friction and rupture, fracture, beyond which aids to the software, such
the explicit methodology is suitable. The required as a user-defined subroutine, are needed for
calculation efforts are fewer than the commonly tracing the initiation and propagation of cracks.

2
which the deformation is concentrated on a small
Traditionally, rupture is assumed to occur when area away from which the structure remains
the equivalent plastic strain in an analysed almost un-deformed. This theory of localized
structure reaches a critical value. This critical necking has been developed for thin metal sheets,
value, sometimes referred to as rupture strain, is in which the assumption of plane stress is valid
related to the strain-stress curves obtained from and the failure criteria can be based on the bi-axial
mechanical tests of uni-axially stretched metal principal strain state formulations. For typical ship
coupons. In the simplified analytical approaches, structures the stress state is tri-axial, which
the rupture strain varies from 1% to 20% and is complicates the development of simple rupture
normally determined based on calibration or criteria.
judgement.
In this study, the definition of failure strain based
There has been an interest in defining rupture on evaluation of the thickness measurements as
strain for FEM analyses. This critical value is recommended by Germanischer Lloyd [2] is
found to be dependent on mesh size. Ranges of employed due to its simplicity. The values of
rupture strain were studied by Simonsen & uniform strain and necking achieved from the
Tornqvist [13], Okazawa et al. [8], Yamada & thickness measurements are related to the
Endo [18], Alsos & Amdahl [1]. calculated stress and are indicated in Table 1.

Refined simulation of fracture initiation and


ε f (l e ) = ε g + ε e
t
propagation requires that mesh size are small (1)
le
enough. This in turn makes the analysis of large
ship structures very time consuming and
Table 1. Proposed failure strain versus mesh
computational demanding. It is well known that a
size, [2]
failure criterion based on the equivalent strain is Stress states 1-D 2-D
generally not valid in bi-axially loaded plates.
εg 0.079 0.056
Estimations of critical equivalent plastic strain as a
εe 0.76 0.54
function of the stress triaxiality using model tests
Element type Beam-Truss Shell-Plate
and FEM analyses were reported in Urban [14],
Tornqvist [12]. Several simple failure criteria and
4. BENCHMARKING PARTICULARS
damage models were implemented in the explicit
finite element code LS-DYNA. Tornqvist &
This section describes benchmark studies
Simonsen [13] have shown that the so-called
comparing numerical model results with the ISSC
RTCL (combined Rice-Tracey and Cockroft-
model tests including details of the finite element
Latham) criteria that account for the tri-axial
tools and general parameters used in the numerical
nature of the fracture gives good comparison to
analysis.
the test results for different materials and various
stress/strain states for validating these fracture
4.1 ISSC TEST
criteria and damage models.
The Association of Structural Improvement of
The rupture failure may be explained using metal
Shipbuilding Industry of Japan [3] conducted
forming theory. The maximum strain that the
extensive collision and grounding tests. One of the
material can sustain is limited by the local plastic
collision tests is a double side structure model
instability. At the failure, plastic deformations
made of mild steel struck by a model bow. In the
concentrate on local areas. These areas have
dynamics test the bow model fell freely from a
typical dimension of plate thickness. The failure
height of 4.8 m above the initial position of the
process can be divided into diffuse necking and
outer hull, which gave an impact velocity of about
local necking phases. Diffuse necking develops
9.7 m/sec.
slowly as a result of strain rate hardening, and
occurs when the load reaches the maximum value.
The final failure occurs by local necking during

3
4.2 NUMERICAL MODEL

The typical shell element size for the whole Figure 1. Collision force versus penetration (top) and
simulated structure was about 15x15 mm, which absorbed energy versus penetration (down)
gave a total number of 56908 shell elements.
Contact was modelled between the indenter and all
structural members, including self-contact, where
the friction coefficient was assumed to be 0.3.
Strain rate effects were included in the material
model, namely Cowder-Symonds parameters
Ccs=3600 (1/s) and q=5.5. In order to avoid
hourglassing, the stiffeners and the flanges were
modelled with fully integrated shell elements with
five-integration points through the thickness. The
remaining bottom structure was modelled with the
under integrated Belytschko-Tsay elements with
one integration point in plane and five-integration
points through the thickness.

4.3 RESULTS

The force-penetration curve determined from the


finite element simulation of the dynamics test is
shown in Figure 1. Good correlation exist between
the experimental and FEA result. However the
contact force from the finite element simulation is
slightly smaller for penetration higher than 0.4m,
approximately. While absorbed energy –
penetration curve has excellent correlation
between experiment and numerical approach Figure 2. Finite element model with very fine mesh (top) and
(which is also shown in the Figure 1). The Figure penetration of 500 mm (down)
2 present the deformation pattern obtained from
5. MODELLING OF FULL SCALE
ANSYS LS-DYNA, where penetration is 500 mm.
COLLISION SIMULATION
3.5
This section presents a numerical study on the
3.0
Experiment double side skin (DSS) bulk carrier design, which
ANSYS - LS DYNA
2.5 is carried out for the evaluation of energy
Force (MN)

2.0 absorption and penetration depth for various


1.5 collision scenarios. The dynamic analysis
1.0
calculations are conducted by means of a non-
linear explicit finite element code, ANSYS LS-
0.5
DYNA [10].
0.0
0.0 0.1 0.2 0.3 0.4 0.5 0.6
Penetration (m) The struck vessel of the Capsizes type DSS design
are assumed to be entirely standstill and the
1.2
striking vessel of an Aframax oil tanker with
1.0 Experiment bulbous bow shape are modelled as rigid body.
ANSYS - LS DYNA
0.8
Energy (MJ)

The principal dimensions of DSS bulk carrier


0.6 vessel are listed in Table 2 and one-cargo tank
0.4

0.2
4
0.0
0.0 0.1 0.2 0.3 0.4 0.5 0.6
Penetration (m)
section is shown in Figure 3 while Figure 4 shows for different striking vessels, various collision
the bow shape of striking vessel. speed and angles, different structural arrangement
and different failure criteria. These requirements
effectively drove the consequences analysis
Table 2. Principal dimensions of DSS vessel towards applying the detailed finite element
Items analysis (FEA) for calculating structural damage.
Length overall (m) 279.29
Length between perpendiculars (m) 279 The principal dimensions of DSS bulk carrier
Breadth (m) 45
Depth (m) 24.1
vessel are listed in Table 2 and one-cargo tank
Draft (m) 16.5 section is shown in Figure 3 while Figure 4 shows
Block coefficient 0.860 the bow shape of striking vessel.
Transverse frame spacing (m)
Bottom 2.58 The results of the simulation further depend on
Side 0.86 two main factors, namely vertical position, i.e.
Deck 5.16
different loading conditions of one ship with
respect to the other, longitudinal position of
impact – the impact may take place directly on a
transverse web frame. These factors together with
the striking bow and the assumption of a right
collision angle and central collision identify the
collision scenario to be investigated.

The vertical relevant position of the two ships is


defined by their drafts, which are given by the
following formulae [2] and Figure 5 shows the
draft differences between striking and struck ships.

3.T1 min + T1maz


∆T1 = T2 max −
Figure 3. One-cargo ANSYS tank geometry model for DSS 4
bulk carrier
3.T1 max + T1 min
∆T2 = T2 max − (2)
4
T2 min + 3.T2 max
∆T3 = − T1 max
4
T + 3.T2 min
∆T4 = 2 max − T1 max
4

Figure 4. Bow shape for striking vessel

5.1 ASSUMED COLLISION SCENARIOS


Figure 5. Draft differences between striking and struck ships
During the risk analysis, collision frequencies are
determined and limitations are recognized in the Four different collision cases are accounted into
damage estimates using traditional approaches. non-linear finite element analyses by means of
The interested parties want to have a better ANSYS LS-DYNA. The impact speed of striking
understanding of collision resistance accounting

5
ships is 10.5 m/sec and the total simulated equation (5) with the expression 0 .49 ∑ Ht 2 ,
collision scenarios time is 0.3 sec. The density of where H is the height of rupture aperture in the
element mesh size is 125 mm in contact areas and side shell (m) and t is the side shell thickness
375 mm for remaining areas. In order to reduce (cm).
CPU time, only one half-cargo tank is modelled
and two ends are clamped while symmetric Table 4. All investigated ship-ship collision
boundary condition is introduced to half scenarios in this study
longitudinal section. Plastic-kinematics material is Case Striking ship & DSS
considered into analyses and is summarized in
Table 3. The yield function of plastic-kinematics
material is given as equation (3). All investigated Case
1
ship-ship collision scenarios are shown in Table 4.
∆T1 = 5.24 m
⎡ ⎛ ⎞ . 1/ P

ε
σ y = ⎢⎢1 + ⎜⎜ ⎟⎟
⎜C ⎟
(
⎥ σ + βE ε eff
⎥ o P P ) Case
⎢⎣ ⎝ ⎠ ⎥⎦ 2

(3) ∆T2 = 2.8 m


E tan E
EP =
E − E tan Case
3
Table 3. Material properties of plastic – ∆T3 = 0.1 m
kinematics used in this study
Density (kg/mm3) 7.85e-6
Young’s modulus (MPa) 210000 Case
Poisson ratio 0.30 4
Yield strength (MPa) 315
∆T4 = -2.7 m
Tangent modulus (MPa) 625
Hardening parameter 0
Strain rate (C, 1/sec) 3200 5.3 CALCULATION OF INTERNAL
Strain rate (P) 5 ENERGY
Failure Strain 0.20
The calculated plastic deformation energy
5.2 THE MODIFIED MINORSKY METHOD (internal energy) versus penetration for DSS
vessel subject to four different collision cases is
In order to compare the numerical results the summarized with considering modified Minorsky
modified Minorsky approach is employed. Based correlation at t = 0.3 sec in Table 5 while Figure 6
on the statistics of certain collision accidents, through Figure 9 illustrate the deformations at t =
Minorsky suggested an empirical linear correlation
between the structural resistance parameter and Internal Energy
absorbed energy as follows: Collision Penetration (MJ)
Vessel
cases (m) LS
Item Minorsky
DYNA
WC = 47.2 RT + 32.7 (4) Case 1 2.67 36.7 30.8
Case 2 2.13 62.1 54.3
DSS
RT = ∑ PN LN t N + ∑ Pn Ln t n
Case 3 1.56 61.2 43.7
(5) Case 4 2.15 48.1 29.1
N =1 n =1
0.3 seconds for DSS bulk carrier vessel.
Woison [16] derived a formula, relating the energy
absorbed by damage of both striking and struck Table 5. Internal energy for four collision cases
vessel structures to the damaged volume of
structural steel, which replaces the constant 32.7 in

6
Table 6 presents a summary of the results. Column
2 indicates the energy absorbed when rupture
takes place of DSS bulk carrier vessel while
column 3 shows the critical energy absorbed when
the inner skins of the cargo hold is ruptured.

Table 6. The summary of the rupture energies


of DSS bulk carrier vessel
Case
DSS - Outer DSS - Inner shell
shell (MJ) (MJ)
1 9.81 36.62
Figure 6. Collapse pattern for Case 1 2 12.43 40.11
3 36.94 No rupture
4 6.42 No rupture

The comparing of the calculated deformation


energy dependent on the penetration for each
collision case of DSS bulk carrier design with the
modified Minorsky correlation is illustrated in
Figure 10.

Case 1
40

Figure 7. Collapse pattern for Case 2 Ls Dyna - DSS


35

Minorsky - DSS
30

Rupture Shell - DSS


Internal Energy (MJ)

25

Rupture Inner Shell - DSS


20

15

10

0
0 0.5 1 1.5 2 2.5 3
Penetration (m)

Figure 8. Collapse pattern for Case 3 Case 2


70

60

50
Internal Energy (MJ)

40
Ls Dyna - DSS

30
Minorsky - DSS

20
Rupture Shell - DSS

10
Rupture Inner Shell - DSS

0
0 0.5 1 1.5 2 2.5
Penetration (m)

Figure 9. Collapse pattern for Case4

7
Case 3
70

60

50
Internal Energy (MJ)

40

30 Ls Dyna - DSS

Minorsky - DSS
20
Rupture Shell - DSS
10
No Rupture Inner Shell

0
0 0.5 1 1.5 2
Penetration (m)

Figure 11. Various impact speed effect on the internal


energy

Case 4
60

Ls Dyna - DSS
50
Internal Energy (MJ)

Minorsky - DSS
40

30 Rupture Shell - DSS

20 No Rupture Inner Shell

10

0
0 0.5 1 1.5 2 2.5
Penetration (m)

Figure 10. Internal energy versus penetration curves for all


investigated collision case Figure 12. Different friction coefficient effect on the internal
energy
6. PARAMETRIC STUDIES 7. CONCLUSIONS
In design, the collision resistance of vessel The paper presented a study that aims to
structures should be improved to the extent investigate the parameters on the ship collisions
possible. The present procedure is now applied to and to provide effective guidelines on the
investigate the possible influences of impact speed implementation of finite element codes. Through
and friction coefficient. For this purposes, only this investigation, the following conclusions can
case 2 is analysed to examine their effect on the be drawn:
absorbed energy. Figure 11 shows the impact
speed effect versus penetration while Figure 12 ™ From the comparison of experimental and
indicates friction ratio effect. numerical results using ISSC collision test
model, it seems that a non-linear explicit
As a part of the exercise, different friction finite element code, ANSYS LS-DYNA is
coefficients are evaluated. It is concluded that quite adequate to perform ship collision
friction coefficients have become quite significant analysis.
so that the contacts take place in a wide area ™ The failure criterion was based on GL
and/or last long. approach. The value of the maximum
plastic strain was selected to be 20 %.
However, more effort is needed in order to
refine rupture criterion.

8
™ The ship structural design has very collision and grounding damage, Proceedings of the 7th
significant influence on the collision PRADS, 173-179.
[6] Kuroiwa, T. (1996). Numerical simulation of the actual
resistance. The collision energy absorption collision and grounding experiments, International
capability depends on the thickness of Conference on Design and Methodologies for Collision and
outer shell, inner shell, side stringers, Grounding Protection of Ships, San Francisco, USA.
transverse webs, width of the side ballast [7] Oh, M., Kim, J.H., Jang, Y.S. and Bird, E. (2005).
tank and width of lower and upper wing Impact analysis of Greater Plutonio FPSO considering ship
collision, International Offshore and Polar Engineering
tanks. Conference (ISOPE), Seoul, Korea, 19-24 June 2005.
™ The results obtained from the finite [8] Okazawa, S., Fujikubo, M. and Hiroi, S. (2004). Static
element simulation may be used a) for the and dynamic necking analysis of steel plates in tension,
assessment of the collision behaviour of a International Conference on Collision and Grounding of
ship under defined collision scenario, b) Ships (ICCGS), Izu, Japan, 25-27 October 2004.
[9] Ozguc, O., Das, P.K. and Barltrop, N.D.P. (2005). A
for the relative comparison of structural comparative study on the structural integrity of single and
arrangements and c) for the validation of double side skin bulk carriers under collision damage,
analytical techniques for ship collision Marine Structures, 18: 511-547.
analysis. [10] Ozguc, O., Samuelides, M. and Das, P.K. (2005). A
™ The rapid advances in computer comparative study on the collision resistance of single and
double side skin bulk carriers, International Congress of
technology make numerical simulation, a International Maritime Association of the Mediterranean,
formidable task only a couple years ago, a Lisbon, Portugal, 26-30 September.
viable choice now. Finite element [11] Servis, D.P., Samuelides, M., Louka, T. and Voudouris,
technology offers reliable codes and G. (2002). The implementation of finite element codes for
elements that are suitable for solving very the simulation of ship-ship collisions, Journal of Ship
Research, 46(4): 239-247.
complex engineering problems including [12] Tornqvist, R. (2003). Design of Crashworthy Ship
large displacements, large strains, and a Structures, PhD thesis, Technical University of Denmark,
contact provided that the proper definition Lyngby, Denmark.
of the problem is established. [13] Tornqvist, R. and Simonsen, B.C. (2004). Safety and
structural crashworthiness of ship structures: Modelling tools
and application in design, International Conference on
ACKNOWLEDGEMENTS Collision and Grounding of Ships (ICCGS), Izu, Japan, 25-
27 October 2004.
The work has been performed in the scope of the project [14] Urban, J. (2003). Crushing and Fracture of Lightweight
MARSTRUCT, Network of Excellence on Marine Structures, PhD thesis, Technical University of Denmark,
Structures, (www.mar.ist.utl/marstruct/), which has been Lyngby, Denmark.
financed by the EU through the GROWTH Programme [15] Wang, G., Jiang, D.J. and Shin, Y. (2003).
under contract TNE3-CT-2003-506141. Consideration of collision and contact damage risks in FPSO
structural design, OTC-15316, Offshore Technology
REFERENCES Conference, Houston, TX, 5-8 May 2003.
[16] Woisin, G. (1979). Design against collision. Proc of the
[1] Alsos, H.S. and Amdahl, J. (2005). International First Int. Symposium on Advances in Marine Technology,
grounding of disabled ships, Marine 2005 – Computational Norway.
Methods in Marine Engineering, Oslo, Norway, 27-29 June [17] Wu, F., Spong, R. and Wang, G. (2004). Using
2005. numerical simulation to analyse ship collision, International
[2] GL (1992). Rules for Classification and Construction Conference on Collision and Grounding of Ships (ICCGS),
Ship Technology Part –1 Seagoing Ships Chapter 1 Hull Izu, Japan, 25-27 October 2004.
Structures Section 33 Strengthening against Collisions, [18] Yamada, Y and Endo, H. (2004). Collapse strength of
Germanischer Lloyd, Hamburg, Germany. the buffer bow structure in oblique collision, International
[3] ISSC. (2003). Committee V.3 Collision and Grounding. Conference on Collision and Grounding of Ships (ICCGS),
15th International Ship and Offshore Structures Congress Izu, Japan, 25-27 October 2004.
(ISSC), San Diego, USA. [19] Yamada, Y., Endo, H. and Pedersen, P.T. (2005).
[4] Kitamura, O. (2001). FEM approach to simulate of Numerical study on the effect of buffer bow structure in
collision and grounding damage, The Second International ship-to-ship collisions, International Offshore and Polar
Conference on Collision and Grounding of the Ships, Engineering Conference (ISOPE), Seoul, Korea, 19-24 June
2005.
Copenhagen, Denmark.
[20] Zhang, S., Ocakli, H. and Pedersen, P.T. (2004).
[5] Kitamura, O., Kuroiwa, T., Kawamoto, Y. and Kaneko, Crushing of ship bows in head-on collision, International
E. (1998). A study on the improved tanker structure against Journal of Maritime Engineering, Transactions of the Royal
Institution of Naval Architects, 146:A2, 39-76.

S-ar putea să vă placă și