Sei sulla pagina 1di 8

Proceedings of The Thirteenth (2003) International Offshore and Polar Engineering Conference Honolulu, Hawaii, USA, May 25 30,

2003 Copyright 2003 by The International Society of Offshore and Polar Engineers ISBN 1 880653 -60 5 (Set); ISSN 1098 6189 (Set)

Ship-Structure Collisions: Development of a Numerical Model for Direct Impact Simulations


Ahmed Derradji-Aouat and Gavin J. Earle
National Research Council of Canada, Institute for Marine Dynamics, St. Johns, Newfoundland, Canada

ABSTRACT
Numerical experiments were conducted to estimate the loads generated during head-on collisions of moving ships with fixed marine structures. A finite element model for the CPF (Canadian Patrol Frigate) was developed. Then, the ship model was tested in a numerical towing tank, where simulations of CPF impacts with a rigid structure were performed. These numerical experiments are, entirely, based on the use of explicit finite element method, employing several commercial and in-house software packages. The research shows a step-by-step procedure for numerical modeling of ship-structure impacts. The contact theory (contact between the ship and the structure during impact) and the coupling theory (fluid-structure interactions) are considered in the model. Typical results of impact loads (both structural and hydrodynamic impact loads) are presented and discussed. Conclusions and recommendations are presented.

INTRODUCTION
With increased oil and gas activities off the coast of Newfoundland and Labrador, Canada, the risk for collisions of ships with marine installations (such as an accidental collision of an oil tanker with an offshore platform for oil/gas E/P) is real. The probability for a severe accident (with significant environmental, human, and capital liabilities) is increased by the fact that the environment off the Canadian east coast is harsh and hazardous. It is characterized by severe sea states (very high waves), dangerous maneuvering and operating conditions (due to the presence of sea ice and icebergs), and persistent poor weather and visibility conditions (due to fog). Therefore, the potential for loss of life, environmental pollution and capital liability are serious operating and design conditions. They are critical elements to be considered in regulating navigation of vessels traveling (or operating) off the Canadian east coast. In most case scenarios, potential accidents can be forecasted and, consequently, avoided. However, for uncontrolled operational circumstances and environmental conditions, accidents may well take place (this is the worst case scenario). Thus, the design criteria for vessels and offshore installations must consider accident scenarios (for mitigation and consequence analysis), in addition to the requirements imposed by the standards for design of ships in normal operating conditions. A literature survey shows that much of the recent research efforts have been devoted to the collisions and accidents of ships operating on inland waterways in The Netherlands (Wevers and Vredeveldt, 1999). Other research work dealt the crashworthiness1 of ships in Japan (Kuroiwa, 1995, Kawamoto, 1995 and Kitamura, 1997 and 2000). In the Netherlands, several full-scale field experiments were performed (actual ships were used. In one test case, one ship with solid-reinforced bow impacted with the mid-side of another one, Wevers and Vredeveldt, 1999). These types of experiments are very much needed to shed some scientific light and understanding on the mechanics and the dynamics of collisions and crashworthiness of marine vehicles. Also,
1

NOMENCLATURE
CPF = Canadian Patrol Frigate. CFD = Computational Fluid Dynamics CSD = Computational Structural Dynamics (E) FEA (M) = (Explicit) Finite Element Analysis (Method). BEM = Boundary Element Method. FSI = Fluid-Structure Interactions. E/P = Exploration and Production. HPC = High Power Computers. ICE_FOR = FE program for the behaviour and failure of ice. ANSYS = FE software package for structures LS-DYNA = FE software package for FSI GEDAP = Generalized Experiment Control and Data Acquisition Package CWT/OEB = Clear Water Tank/Offshore Engineering Basin. ALE = Arbitrary Lagrangean-Eulerian numerical solution E, , , B and = Modulus of elasticity, Poissons ratio, density, bulk modulus, and viscosity, respectively. T, t, t, v, I = Total time for simulation, time, time increment, respectively. = Velocity is fluid EOS = Equation of state for pressure calculations. All units are SI units

Structural failure and calculations of energy absorption indices

520

the tests provide valuable data needed for the development and validation of analytical and numerical predictive design models. However, full scale ship testing is a large and complex undertaking. Also, it is costly and time consuming. Numerical methods may be best suited to estimate crashworthiness of ships and provide a rough idea of impact loads and the consequence resulting in a given accident scenario.

years, many mathematical models single equations and multi equations models, were developed to predict the loads generated during collisions (or grounding) of ships. The vast majority of these models were based on energy conservation principles (Simonsen, 1998a,b and Terndrup and Zhang, 1998). From the literature, it seems that a global analysis approach is preferred (where the hydrodynamics, structural dynamics/deformation, and the fluid-structure interactions are considered in the solution). The use of single analytical equations may be not adequate to compute impact loads and structural failure is ship collisions (the problem dynamic and highly non-linear). In this paper, the Marine Dynamics Virtual Laboratory (MDVL) is used simulate collisions of a ship with an offshore structure. The MDVL is an expandable bundle of commercial and in-house developed software packages (Derradji-Aouat, 2001). Among these packages are: ANSYS, LS-DYNA, GEDAP, and ICE_FOR. Both ANSYS (www.ansys.com) and LS-DYNA (www.lstc.com) are commercially available codes, GEDAP is a package developed by NRC (www.chc.nrc.ca/) to generate drive signals to activate the wave makers in the CWT/OEB, and ICE_FOR is a material constitutive model for the mechanical behaviour of ice, including its damage (fracture) and failure (Derradji-Aouat, 1994, 2000, 2002 and Derradji- Aouat et al., 2000).

Rigid Wall Model

Di Center-line

CWT
X Y Figure 1: Schematic for a ship structure collision model.
Over the years, many numerical studies (FEM, BEM) on structural damage and magnitude of collision loads on ships were carried out. For example, Kitamura (1997, 2000) presented the results of LS-DYNA runs for the design of a bow for a container ship colliding with a vertical structure. Tangborn et al. (1998) used, also, LS-DYNA to back calculate the size and mass of an iceberg that struck the oil tanker Overseas Ohio. Glykas and Das (2001) used ABAQUS to calculate the elastic and plastic deformation of a ship bow striking a rigid body (head on collision). It is interesting to note that, in the latter study, the model for the ship bow model included the outer shell as well as decks and bulkheads. The results indicated that a large amount of the kinetic energy is transferred into plastic (permanent) deformation. In turn, that plastic stored energy is dissipated through structural collapse. Alternative to the numerical approach is the analytical one. Over the

Ship Model

Figure 2a: Pictue for the CPF model.


It is hoped that the ultimate purpose of the MDVL is to provide a virtual replica of the CWT test facility at the Institute for Marine Dynamics of the National Research Council of Canada (http://www.nrc.ca/). IMDs existing experimental data, obtained over the last 15 years, will serve as a ground to develop and validate the MDVL. Numerically, the overall vision of the MDVL is based on the hypothesis that the combination of a virtual numerical wave maker and structural models for ships a single numerical fluid tank fluid enables investigators to perform fully coupled CSD and CFD computer experiments of collisions and accidents of ships and offshore structures in various sea states and conditions. All software components in the MDVL are based on FEM. It should be pointed out that, because of the generality of the FEM, the complexity or simplicity of the models used in the numerical simulations depends on the users needs and objectives, it is limited only by the available computer hardware capacity.

521

elements (Derradji-Aouat et al., 2001) In this work, numerical simulations of collisions of a ship with an offshore structure are performed. A scaled down (1:20) model for an actual ship (CPF) is used in a numerical CWT. For simplicity and clarity, in this paper, only simulations for head-on collisions (Fig.1) are considered, although oblique and/or lateral collision scenarios can be simulated. All simulations were performed in clam water conditions (no waves). In these simulations, the FE model for the CPF was used because it was readily available. The model was developed for a prior research works (Derradji-Aouat et al., 2001). A picture for the scaled down physical model in given in Fig. 2a. The structure used in the simulations is a rigid vertical wall. A shipwall model - geometry configuration - is given in Fig. 2b. Notice that the rigid wall in fixed at the bottom nodes. The vertical rigid was created from a rectangular block (2.0 m wide, 0.5 m thick, and 4.15 m deep) and meshed using 8-node brick elements. All nodes at the bottom layer of the wall model were fixed in all directions. The numerical tank CWT was modeled using a rectangular basin of 16.0 m long, 6.0 m wide and 5.0 m deep. These dimensions are much smaller than the dimensions of the actual CWT at the IMD, but they were chosen to reduce the total number of elements as well as CPU time. Figure 3 shows the geometry of the tank model, it is made up of two spaces solid boxes. The lower box (16.0 m X 6.0 m X 4.0 m) is used to model the water, while the upper box (16.0 m X 6.0 m X 1.0 m) is used to model the air space. From the FEM viewpoint, the two boxes represent an Eulerian space, the two parts of the Eulerian are needed so that free surface effects (water-air interactions) can be modeled. The two parts of the Eulerian space are meshed using 8-node brick elements (Fig. 3a). Notice that the mesh is refined along the zone near free surface. Smaller element sizes at the waterline are needed for more accurate predictions of bow waves and wake profiles. All three FE models (ship, wall, numerical tank) were combined together (Fig. 3b). Both ship and wall are placed on the centerline of the tank model. The mechanical material properties for each model are given in Table 2. The fluid (water and air) boundary conditions are assumed as follows: All nodes along the sides of the tank (Y = 3.0 m) were fixed in the ydirection. The nodes at the top and the bottom of the tank (Z = - 4.0 m and m, Z = + 1.0 m) were fixed in z-direction. Nodes at both ends of the tank (X = 0.0 m and X = 16.0 m) are fixed in the X direction. A schematic for the boundary conditions is given in Fig. 4.

Ship-CPF Model

Rigid Wall

Water Tank

Table 1: Geometric and FE properties Figure 2b: Model for ship hull and wall; initial geometry set up
In this paper, it is aimed to achieve the following four objectives: a) Develop and present FE models for the ship (CPF), structure (rigid vertical wall) and CWT. b) Conduct numerical simulations for head on collisions scenarios. c) Analyze and discuss the numerical results and provide qualitative conclusions for the use of numerical methods to simulate shipstructure collisions and accidents d) More importantly, present a combined CFD/CSD numerical solution for marine and offshore structures. FE MODELS AND INPUT PARAMETERS FE models for the CPF, water towing tank, and a rigid wall are developed. The geometry particulars for each model are given in Tables 1a, b. All three FE models are constructed in ANSYS pre-processor, their statistics are given in Table 1c. The FE model for the ship was built from CAD drawings of the CPF physical model design. IGES surfaces were generated in CADKEY, and then, the surfaces were imported into ANSYS using IGES transfer routines. Those surfaces were meshed in ANSYS using 4-node shell Table 1a. Geometry and particulars of the CPF model scale, Fig. 2a Length (at waterline), m 5.950 Beam (width, at water line), m 0.750 Volume, m3 1.910 Displacement, kg 523.900 Draft, m 0.211 Table 1b. Geometry of CWT tank (water, air) and wall FE models. Geometric Property Water Space Air Space Wall Length, m 16.0 16.0 2.00 Width, m 6.0 6.0 0.50 Depth (water), 4.0 1.0 4.15 Height (air), m Table 1c. FE model Statistics FE Property Ship Model Elements 11,242 Nodes 1,729

Tank model 78,000 84,537

The ship is subjected to a velocity of 3.4 m/s. This value for ship velocity was chosen to simulate impacts at very high speeds). In full scale, this corresponds to velocity of about 30 knots (considering model scale of ~ 1:20).

522

Table 2: Material Properties


Property E, N/m2 B, N/m2 Ship 1 X 108 0.3 1000 N/A N/A Water N/A N/a 1000 2 X 107 0.0 Air N/A N/A 1000 2 X 107 0.0 Wall N/A N/A N/A N/A N/A

As shown in Fig. 5, at impact, the ship is stopped, and then, pushed backwards (reaction from the rigid wall), showing a negative velocity (down to 0.25 m/s. Once the ship start to move backwards, it, quickly, stabilize and the complete stop is achieved. It is important to note that the reaction from the rigid wall and the subsequent backwards movement of the ship (up to the complete stop) are all computed numerically.

In the table; N/A = Note Applicable


Ship

For the numerical simulations, at time t = 0 sec., the ship is stationary, and the structure is at an initial separation distance, Di.. At time t = 0+ sec., the ship starts to move forwards at a velocity, v, along the longitudinal axis of the tank (in the positive X direction, Fig. 1). For clarity, the assumptions used in this study are: All simulations are performed in calm water (no waves). All simulations deal only with head-on collision scenarios (no side and/or oblique collisions). Viscosity effect was not considered Viscosity, = 0. Material behaviour of the ship-structure is linear elastic. The wall-structure is rigid. Basically, all of the assumptions are made to cut down the computer processing time.
Tank

Wall

Figure 3b: CPF, tank and rigid wall FE meshes


There are 2 fundamental reasons for why the input velocity scenario (in Fig. 5) was chosen.

Air space

W ater space

First: If high ship velocity (3.4 m/s) is applied at t = 0+, sudden and significant disturbance in the water will take place, and consequently, this may leads to numerical instability (and/or a loss) of coupling. Example for a numerical instability, the Lagrangean nodes (ship) will lead to excessive displacement in the Eulerian nodes (excessive deformation in the fluid domain). The loss of coupling means that the FSI equations are not activated, during the numerical calculations (this means that no interface between ship and water takes place). Second: If the forced velocity term in not released (just before actual impact), the ship would have been forced to go through the structure (which is unrealistic) because this ship is forced to move forwards and the structure is fixed in its place. The contact algorithm would have been overridden (contact equations are not activated).
One advantage of EFEM is that a convergence towards a solution is assured (guaranteed). However, its main challenge is to set up the problem such that both coupling and contact equations are activated and work properly. Therefore, in ship collision simulations, it is recommended the use, for input, of a progressively increasing ship speed. This is needed to avoid numerical instabilities, loss of coupling, and to develop and propagate the appropriate bow wave when full speed is reached. Also, considerable attention should be paid to the magnitude of time step, size of elements and fluid properties (as will be explained below).

Figure 3a: FE model for the towing tank.


To avoid numerical instabilities, at the start of the simulation, the value of velocity was input as an increasing value, from 0.0 m/s to 3.4 m/s, (rather than applying 3.4 m/s at the start of the simulations). As shown in Fig. 5, once at 3.4 m/s, the velocity was kept constant throughout the rest of the simulations. However, just before the impact (few milliseconds before actual contact), the forced velocity term was released and the ship was allowed to continue its forward movement (propelled by its own momentum at the time of velocity release). Therefore, the ship impacts the wall not at a forced velocity of 3.4 m/s, but rather due to its momentum at the time of velocity release.

523

function of the ratio of the size of the smallest element in the mesh, Lc, to the numerical value of the speed of sound, Vs, in the materials considered in the analysis (ship structural materials, materials of the offshore structure itself, water and air).
Free Surface Air Space

t = f(Lc/Vs).

2.a

Water Space Fixed in Y direction

The velocity of sound in a material is function of the elastic properties of that material (in this case elastic properties for the ship material, materials for the offshore structure, and properties for water and air). For example, in elastic materials, Vs is:

Tank wall
Fixed in Z direction
Tank Floor

Vs =

E(1- ) (1+ )(1- 2)

2.b

Typical values for Vs are: 331 m/s for air, 1478 m/s for water, 5240 m/s for steel, etc.

Figure 4: Boundary Conditions for the towing tank model.

Figure 6a. Bow wave of CPF, fluid movement in the tank

Figure 5: Input-Output velocities (for representative nodes)


(Time, s, and Velocity, m/s) ALE SOLUTION AND TYPICAL RESULTS The ALE governing equations for conservations of mass, momentum and energy in the fluid are (Souli, 1999):

i - i = - xi t xi
i = ij, j b i - i i t xj

for mass

1.a

for momentum

1.b

Figure 6b. Bow wave of CPF, at the time of impact


Eq. 2b is needed ensure that, during the each time step of the simulations, the stress wave propagates through a distance of less oneelement size. Consequently, larger values for the material elastic properties result in smaller time steps. Also, smaller element sizes result in smaller time steps. The numerical solution technique used to combine a Lagrangean mesh (ship) with an Eulerian one (water and air) uses a split operator to solve

E E = (ij i, j + b, i ) - j t xj

for energy

1.c

For the numerical computations, the overall time for the simulations is divided into time steps. The size of the time step is calculated as

524

the ALE governing equations. The concept of a split operator is based on the principle that, during each time step, the complete solution is obtained in 2 steps. As seen in the governing equations, the right hand side of each one is, essentially, made up of two different parts. The 1st represents the behaviour of the fluid in the Lagrangean phase while the 2nd part represents its behaviour in the Eulerian one.

and P is the hydrostatic pressure. P is calculated using a polynomial Equation of State (EOS). In its simplest form, the EOS for P is:

PEOS = B (

- 1) o

4b

where and o are the current and initial fluid densities in a given Eulerian element.
In the 2nd step, advection calculations are performed. Advection is transport of mass, momentum and energy across the boundaries of the Eulerian elements. In CFD, the general advection equation:

+ div ( , v, X i , X j ) = 0 t

5a

where is the parameter to be conserved (mass, momentum or energy) and is the flux function. In a numerical integration form, the general advection-flux rule is:

S e Ve = S l Vl +
Figure 6c. von Mises impact stresses on the bow of the ship

j Sl faces

* Flux j

5b

where S indicates the parameters to be conserved (mass, momentum, energy). V is the volume of the element. Sj refers to the parameters to be conserved in the adjacent Lagrangean element j. Fluxj is the volume flux through adjacent Eulerian element j. The subscripts e and l indicate Eulerian and Lagrangean, respectively. The parameters to be conserved are: S = E/V for energy advection, S = for mass advection, and S = .i (i=1,2,3) for momentum advection. Once the 2nd step is completed. The calculations move on the next time step. This process is repeated until the calculations in all time steps are completed. It is important to note that, in the first step of the solution, the nodes in the Eulerian mesh are allowed to be deformed (displaced) by the moving Lagrangean mesh (ship). However, in the 2nd step, a penalty (restoring) function is applied on the Eulerian nodes to restore them to their original locations (geometric state). Restoring calculations are required because, in an Eulerian mesh, the nodes and elements are fixed, in space, while materials (structure fluid) flow in and out of the elements via advection and flux calculations. The restoring loads are applied on the corresponding Lagrangean nodes so that stresses and strains within all elements of the ship are calculated using the specified material constitutive model (elastic, plastic, etc.). The free surface is included in the calculations using the advection equations. Both water and air are subjected to the same equations. Water and air are fluids; they have 2 different values for densities, bulk moduli, and EQS. The ability of an Eulerian element to accommodate more than one material (water, air, structure) is called multi-material modeling, it allows the movement of the ship within the Eulerian domain and the generation of free surface. In ALE terminology, the expression coupling model is used to indicate the interface of water with the structure (either the structure of the ship or the offshore structure). Coupling is the passing of information between Lagrangean nodes and Eulerian nodes (it is the

Figure 6d. von Mises stresses ( selected nodes are represented).


In the 1st step is a Lagrangean one, where the ship mesh moves forwards (according to the prescribed input velocity of the ship). In the fluid, both velocity and internal energy are calculated as:

i = ij, j b i t

and

E = ij i, j + b, i t

Where the stresses, ij, are calculated using:

ij =

( )

ij d

+ P[I] = 2 ij + PEOS ( K , )

4a

where [I] is the unity matrix. Both (ij) and (ij)d are the normal stress and deviatoric stress tensors, respectively. ij is the strain rate tensor,

525

interface behaviour at the boundary layer, FSI equations). Usually, the interfacing nodes (fluid nodes that are in contact with the Lagrangean nodes) are coupled together (numerically). The interfacing nodes are forced to have the same velocity, same acceleration and/or the same pressure values. This coupling constraint may be set along a given direction (such the normal or lateral directions) or it can be set in all three directions. At the time of impact between the ship and an offshore structure, there exists a contact between the two structures. The contact theory allows the transfer of action-reaction forces between the two. Numerically, in contact problems, there is always a master structure and a target structure. In the present analysis, the ship is a master structure while the wall is the target one. During the simulations, a considerable computer processing time is spent over detecting the target. Contact is established (or target is detected) once the nodes (or surfaces) of the master and those of the target occupy the same space (within the accepted numerical tolerance). Contact forces are calculated from the structural stiffness of the ship material and displacement of its impacted nodes. Stresses and strains in the structural material of the ship are calculated from its constitutive model (in the present study, linear elastic structural model is used). Potential and kinetic energies are calculated from the impacted masses and the displacement of the master structure.

Calculated deviatoric (shear) stresses in selected elements are given in Fig. 7c. Pressure levels at selected elements in the bow of the ship are given in Fig. 7d. From the structural analysis point of view, the magnitude of the deviatoric stresses, the distribution of von Mises stresses and the magnitude of pressure loads produced by model are reasonable. However, the solution does not explain geometric (hull girders and stiffeners) or material non-linearity, which are not modeled.

Figure 7b. von Mises stresses, selected elements in the bow. (Time, s, and Stress =MPa)

Figure 7a. Accelerations in X direction at selected nodes (Time, s, and Acceleration, m/s^2)
In marine collision problems, the contact formulation is of the up-most importance. If contact between the ship and the structure is not detected (solution did not converge), the master structure may go through the wall without detecting it, and subsequently, the calculations of contact forces and interface energies are not performed at all. Alternatively, significant numerical penetration of the master structure into the target may take place, and subsequently, the calculated contact forces and energies become not very accurate. For the simulation presented in this paper, the ship is forced to move forwards as explained above (and shown in Fig. 5). The numerical results are presented as follows: Calculated accelerations at selected nodes in the bow of the ship are given in Fig. 7a. von Mises stresses are given in Fig. 7b. The ship hull yielding is not considered, a linear elastic stress-strain model is chosen. von Mises stress distributions were already given in Fig 6c and 6.

Figure 7c. Deviatoric stresses, selected elements in the bow. (Time, s, and Stress =MPa)
ANALYSIS AND CONCLUSIONS In numerical modeling, assumptions and approximations are made about the geometry, material types and material properties, external loading (or motion) and the boundary conditions. These assumptions may have ramifications on the final results. Therefore, the conclusions drawn in this paper are limited to the assumptions and conditions pointed out above. In all probability, in order to investigate the effect of these approximations and assumptions on the accuracy (and validity) of the results of any numerical simulations, full sensitivity and parametric analysis is needed.

526

Figure 7d Pressure levels at selected elements in the bow (Time, s, and Pressure =MPa)
It is beyond the mandate (objective) of this paper to present the analysis of sensitivity and the effects of various parameters on the final results. That mandate is being carried out by the authors, and the results will be published in future publications. Within the framework of the mandate for development of a numerical model for ship collisions, it appears that combining CFD and CSD to solved the problem and obtain impact loads is very much possible. The complexity of the interactions between fluid and structure and the complexities involved in the contact and interactions between two structures in hydrodynamic environment make search for an analytical solution, almost impossible to find. Numerical methods offer the general and versatile solution for most complex problems. The results presented here are valid only for 1 velocity case scenario high-speed impacts. The magnitude of stresses and their distribution appear to be very acceptable. However, the validation of the numerical results against some experimental work is critical to develop much need confidence in the numerical modeling in this kind of problems. Unfortunately, up to now, there is no small-scale experimental data that can be compared to the results given in this paper. ACKNOWLEDGEMENTS The dedication of the staff from the computer systems group at the institute is greatly appreciated and valued. This work is made possible only with their help. REFERENCES Derradji-Aouat A. (1994). Ice Loads on Conical Piers Numerical and Development of Design Equations. 7th Cold Regions Engineering Specialty Conference. Canadian Society of Civil Engineering, Edmonton, pp. 243-262. Derradji-Aouat A., Sinha N.K., and Evgin E. (2000). Mathematical modeling of monotonic and cyclic behavior of fresh water columnar S-2 ice. Journal of Cold Regions Science and Technology, Elsevier Pub. No. 31, pp. 59-81. Derradji-Aouat A. (2000). A Unified Failure Envelop for Isotropic Fresh Water Ice and Iceberg Ice. Proceedings of the 19th Offshore Mechanics and Arctic Eng. OMAE2000, New Orleans, LA, USA.

P/A section, PDF file #1002. Derradji-Aouat A. (2001). On the development of the Marine Dynamics Virtual Laboratory. ASME-PVP-2001. The 5th Int. Symposium on Emerging Technologies in Fluids, Structures, and Fluid-Structure Interactions, state of the art methods, Atlanta, Georgia, USA. July, 22- 26. Derradji-Aouat A. Juva M. and Hermanski G. (2001). FE Model for CPF using CAD to ANSYS IGES Imports/Exports. 18th Canadian Congress of Applied Mechanics. CAN-CAM-2001, St. Johns. Newfoundland, Vol. 2, pp. 345-346. Derradji-Aouat A. (2002). A Failure envelope for saline ice in the brittle regime. Proceedings of the 17th International Symposium on Okhotsk Sea and sea ice, Mombetsu. Japan, February 24 - 28, 2002. Glykas A. and Das P. K. (2001). Energy conservation during a tanker collision, Ocean Engineering Journal, Vol. 28, No. 4, pp. pp. 361-374. Kitamura O. (1997). Comparative Study on Collision Resistance of Side Structure, Marine Technology, Vol., 34, No. 4, pp. 293308. Kitamura O. (2000). Buffer Bow Design for the Improved Safety of Ships. Ship Structure Symposium, June 13-14, 2000, Pentagon City, Virginia, USA.Simonsen B. C. (1998a). Ship Grounding on Rock I. Marine Structures Journal, Vol. 10, No. 7, pp. 519-562. Kuroiwa, T., Kawamoto, Y., Kusuba, S. And Stillman, D. (1995). Numerical Simulation of Collision and Grounding of Ships, Proceedings of the International Conference on Technologies for Marine Environment Preservation, Vol.1, pp.66-70 Simonsen B. C. (1998a). Ship Grounding on Rock I. Marine Structures Journal, Vol. 10, No. 7, pp. 519-562. Simonsen B. C. (1998b). Ship Grounding on Rock II. Marine Structures Journal, Vol. 10, No. 7, pp. 563-584. Tangborn A., Kan C. D. and Tangborn W. (1998). Calculation of the size of the iceberg struck by the oil tanker Overseas Ohio. The 14th IAHR Conference, symposium on ice, New York,. Vol. 1, pp. 237-241. Terndrup P. and Zhang S. (1998). On Impact Mechanics in Ship Collisions, Marine Structures Journal, Vol. 11, No. 10, pp. 429449. Wevers L. J. and Vredeveldt A.W. (1999). Full scale Ship Collision Experiments 1998. TNO-report # 98-CMC-R1725. TNO building and research, Delft, The Netherlands.

527

Potrebbero piacerti anche