Marine2005 Paper a&A

7/31/2019 Marine2005 Paper a&A

1/13

INTENTIONAL GROUNDING OF DIASBLED SHIPS

Hagbart S. Alsos*

and Jrgen Amdahl

Norwegian University of Science and Technology, NTNUDepartment of Marine Technology, NO-7450 Trondheim, Norway

*[email protected]

[email protected]

Key words: Ship grounding, Finite element applications, Convergence studies

Abstract: The paper presents a model to determine the consequences of intentional

grounding (structural damage, rupture, potential flooding or oil spill). The

grounding process is decoupled into rigid body motion and resistance to penetrationof the bottom structure. The latter is calculated on the basis of predetermined

resistance - penetration relationships for a tank section with the non-linear finite

element code LS-DYNA. The focus is here put on the physics and modelling issues of

the problem, and concentrates on the initial state of a ship grounding.

The work is part of a project devoted to development of a decision support system,

which will provide guidance to the ship master and the various levels of decision

makers on board and on shore on how to operate a ship once a potentially critical

damage has occurred: i.e. how to manoeuvre in critical waters with loss of propulsion

or damaged manoeuvring systems, how to operate / navigate to limit sea loads, such

that hull damage (e.g. due to collision, grounding, overloading of degraded structure)

does not propagate to a critical level for the entire ship, and, as a last resort, to

assess the consequences of running the ship aground. The project is funded by the

European Community under Framework Programme 6.

1. INTRODUCTION

Both historical and recent events have made it clear that ship grounding and collision

represent significant hazards. This applies both with respect to loss of human lives,

severe environmental consequences and economical loss. The most typicalconsequence of ship grounding and collision is oil spill as exemplified for instance by

the Sea Empress which spilled 65,000 tons of crude oil at Milford Haven harbor,

Wales, in 1996. In 1997, the Nakhodka, a Russian tanker carrying 19,000 tons of

heavy fuel oil broke into two during storms in the Sea of Japan with severe

environmental consequences.

If counter measures are not taken the risk associated with maritime transport

especially in Northern European waters is likely increase substantially in the future.

Plans are being drafted for developing large facilities in North Russia for depositing of

nuclear waste from European power plants. This implies transport of such waste alongNorwegian coast also during extreme weather conditions in wintertime. The


2/13

expanding oil exploitation in the Barents sea is projected to increase current transport

volumes from passage of one VLCC tanker per week at present, to one VLCC per day

in 2007. It is evident that the exposure to oil spill from stranded tankers increases

considerably. This is all the more serious as the Barents sea is one of the last

unspoiled marine environment in Europe and hosts one of the worlds richest and

most pristine marine ecosystems. Arctic areas are particularly vulnerable to pollution,

due to local climatic conditions and the fragility of the food chain mechanisms. A

large oil spill could cause dramatic consequences to the wildlife in this area, both sea

birds, mammals and fish stocks.

The adoption of double side hulls in oil tanker has been recognized as an effective

countermeasure to prevent a disastrous damage induced by collisions. However, when

considering that ocean-going vessels are increasing not only in size but also in speed,

the threat of disastrous collision accident should be further mitigated also on the

responsibility of striking ships.

2. PROJECT OVERVIEW

Present ships are equipped with extensive systems for sensing and monitoring.

Monitoring of propulsion and maneuvering systems are standard, and hull strength

monitoring systems are becoming more and more common on new vessels. However,

modern sensing and monitoring systems provides a wealth of data, and the number of

alarms triggered if a system fails can make it difficult to identify the actual source of

the problem (alarm inflation). Furthermore, the systems focus on each system as

stand-alone unit, and do not convey an overall picture of the risk level for the ship as awhole, making it difficult to perform an appropriate assessment of the situation.

Finally, very few systems provide guidance to the ship master or the crew on how to

operate the ship if one or more critical systems have failed.

In order to meet this challenge a joint industry project funded by the European

Community under Framework Programme 6 is launched. The project is entitled:

Decision Support System for Ships in Degraded Conditions (DSS_DC). managing the

project.

The objectives of the project are to:

Develop an efficient on-board Decision Support System (DSS) for handling of

ships in degraded condition.

Develop simulation and guidance modules for mastering a ship in heavy seas

in the main emergencies: Loss of propulsion, Damage to maneuvering

systems, Collision / hull damage, and Grounding

Develop efficient systems for crisis assistance and decision support from on-

shore command centers and vessel traffic control centers, based on direct

information about technical condition of the ships systems. This includes

systems for automated shipshore transfer of condition data for the on-board

systems.


3/13

Extend on-board sensing and monitoring systems with modules and models

for Technical Condition Management to assess the capacity of supporting

functions like e.g. power generation.

Establish tools for consequence assessment of intentional grounding. This

objective is particularly addressed in the present paper

The final outcome of this effort will be prototype installations of the developed DSS

system and Man Machine Interface (MMI) on board one passenger vessel and one

cargo vessel.

The Idea of Intentional Grounding

After a critical event, such as collision, explosion or grounding, the worst outcome

might be that the vessel ends up by breaking, sinking or stranding in an uncontrolled

manner. The Prestige accident demonstrated that ship wrecks potentially may leak oil

for years, thereby representing a continuous pollution threat. Alternatively, a last

resort in a critical situation may be to run the ship aground in a controlled manner insheltered waters. However, running the ship aground is by itself a risky undertaking.

At present, very few tools are available to the relevant decision makers to assess the

consequences of such a decision.

A part of the project is dedicated to the development of simple procedures and tools

for consequence analysis of ship grounding. The tools will use information from the

ships sensing and monitoring systems as input (hull monitoring system, loading

computer etc.) together with simplified numerical models for hull girder strength.

Selected grounding scenarios for typical ship types will be calculated priori to

establish parametric formula for estimated damage and contact force between ship andthe sea floor. The final assessment will account for still water loading, dynamic

loading from waves and tide, as well as local contact forces from the sea floor.

3. INTENTIONAL GROUNDING

Ship grounding is a very complex process. The grounding force cannot be estimated a

priori. It depends on the resistance to penetration of the obstruction into the bottom of

the ship, and the indentation depends again on the rigid body motion of the ship

(heave, pitch, and roll) as it travels over the obstruction. Furthermore, the groundingdamage will often involve deep crushing and tearing of bottom plating, stiffeners and

girders over a substantial part of the ship length. Rigorous analysis of such problems

calls for non-linear finite element methods (NLFEM). However, the size of the

problem makes this a tremendous task. A relatively fine model may take days to run

through with today's CPU capacity. Including the time consumed to make the model,

an extra week of intense work can be added. Then there is the problem of fixing bugs

and making the simulation run smoothly. If at all successful, the process may take

weeks to complete. Thus, simplifications have to be made. An operational decision

support system requires the grounding simulation to be performed within seconds.


4/13

This will be achieved by:

Decoupling the rigid body motion and the resistance to penetration of the

bottom structure.

Predetermined resistance - penetration relationships for the typical ship bottom

structures

A Simplified Computer Model

A simplified computerized model will be established based on a rigid body system

with interacting loads and forces in a six degree of freedom system. It is convenient to

distinguish between motions in the vertical plane; heave, pitch, roll and the horizontal

plane; surge, sway, yaw. The coupling between them is weak and can be disregarded

for grounding analysis. The vertical motion may be determined from hydrostatic

considerations using a ship load and stability calculator, including the grounding

contact force. In the present project the loading computer SHIPLOAD will be used as a

platform for the whole system.

The necessary input to this analysis comprises:

ship hydrodynamic data

ship light-weight distribution

load condition, including bunker situation and potentially flooded

compartments

initial draught and trim (which may be calculated from the above information)

hull girder load effects (stresses) from monitoring system (if known) this

may alternatively be estimated from the above information

Intentional grounding is as slow process, thus transient dynamic forces may be

neglected. The effect of grounding is represented as a concentrated force

corresponding to the instantaneous indentation of the sea floor into the ship bottom

for a single point contact or a series of patch loads if grounding takes place over a

large area. At a given time the estimated grounding force is input to the load

calculator. The updated mean draught, trim and roll angle in the next time increment

provide the necessary information to calculate the next level of indentation of the sea

floor into the ship bottom, and hence the new grounding force. This is illustrated in

figure 1.

Figure 1: Model of the grounding process, [7]

Fv

FhFh

Fv


5/13

The load and stability calculator will also assess the hull girder load effects i.e.

global bending moment and shear force, including the effect of the grounding force.

These will be evaluated against the available hull girder resistance in shear and

bending. Any degradation of the resistance due to bottom damage is accounted for.

The resistance to penetration of the bottom structure will be calculated by

phenomenological numerical formulations which have been validated against test data

and results from non-linear finite element analysis (NLFEA). As the bottom structure is

fairly regular shaped over the cargo area, it is only necessary to model a part of the

ship. In this case two tank lengths have been modelled. The resistance to penetration

is analysed for a constant indentation by first pushing the obstruction the selected

distance into the bottom, followed by motion along the ship. This provides

information of the vertical and horizontal component of the contact force in the

steady-state phase. Analyses are performed for a large number of indentation levels.

Having established the resistance for various indentation levels, the actual resistance

based on the indentation calculated in the rigid body motion analysis can bedetermined by means of interpolation.

The harmonization of ship classification rules ensures that ships are built fairly

consistently. Within each size category of a ship type, different structural

arrangements can be classified into a small number of characteristic groups. For each

group and sea floor characteristics the resistance to penetration will be established as

described above.

In addition to calculating the grounding damage, the potential degradation of the hull

in stranded condition will be analysed, taking into account the effect of tides, waves,possible outflow of cargo, and/or flooding of tanks. This may also be used to assess

the force required to pull the ship of the off the ground.

The outcome of the grounding simulations is information of:

Likely damage of the ship bottom due to grounding

rupture of cargo tanks

amount of cargo spill

hull girders stresses to be evaluated against ultimate hull girder resistance

prediction of potential damage escalation if the ship remains in stranded

condition, taking into account weather forecasts. This includes hull girderloads and strength assessment

4. GROUNDING SCEARIOS

A factor of paramount importance is the sea floor conditions at the stranding site. In

general the conditions may vary from soft bottom (clay, sand) to sharp, rigid

obstructions like sharp pinnacles. Obviously, it will be extremely beneficial if a survey

of the sea floor characteristics at potential grounding locations is performed and

assessed with respect to functionality. Some locations which are particularlyfavourable with respect to intentional stranding have been identified by the Coastal


6/13

Directorate in Norway. No detailed survey of the sea floor has however been

undertaken. Preparation of the sea floor in order to smooth the surface and remove

unfavourable obstacles may also be envisaged. The actual conditions should constitute

the basis for the modelling of the sea floor in the bottom damage simulations.

The environmental conditions at the grounding site should also be mapped so that

their effect on the hull girder can be evaluated. Tidal variations are obviously very

important, because large changes in trim and/or draught of a stranded vessel may

result in excessive hull girder forces. Even if the grounding preferably should take

place in sheltered waters, wind and waves actions may induce significant forces on the

hull girder. Shallow water effects may also have to be considered.

Grounding Scenarios

As there exist very little information on the sea bed conditions, a few general

scenarios have been worked out. The problem is assumed to be quasi static, thus the

ship will crushed onto the sea bottom slowly without any motions in the horizontalplane. In the first stage of the project, local damage to tanks and the hull section due

to penetration have been analyzed. Overall damage, as loss of structural resistance in

hogging and sagging, are not treated here. This will be dealt with later on in the

project.

The problem have been simplified by defining three different indenters which can be

combined in different ways. These are the rock, reef and shoal scenarios. Sea bottom

profiles are relative in size, and need to be related to the actual ship size. The rock is

assumed to be much smaller than the ship itself, only 1/6 of the total ship breadth,

while the larger reef equals to 1/4B, where B is the ship width. The largest indenter,the shoal, covers more than half the ship breadth, see figure 3. For each penetration

object, a set of standardized positions is defined. On the basis of these scenarios, a

data base will be generated, which can be used for interpolation for intermediate

obstructions. All in all, 10 different cases have been determined:

Crushing at transverse bulkhead, longitudinal CL: rock, reef, shoal

Crushing at transverse bulkhead, 1/4B from longitudinal CL: rock, reef

Penetration of mid tank section, at the longitudinal CL: rock, reef, shoal

Penetration of mid tank section, 1/4B from the longitudinal CL: rock, reef

(a) (b) (c)

Figure 3: The grounding scenarios, (a) shoal, (b) reef, (c) rock.


7/13

It is a great simplification to limit the number of scenarios to 10 cases. However, the

bottom conditions are a big question itself, so simplifications may very well be as

accurate as any other more complex wild guess.

Friction is another part of the problem. A rock may be rue, have an highly irregular

shape, and covered by marine growth. A unique friction parameter is therefore hard to

establish. Grounding simulations show however that friction plays a minor role in the

total energy account. Therefore, with the basis in literature, [5] it may be reasonable to

assume a friction factor close to 0.3.

5. FINITE ELEMENT ANALYSIS

The contact forces at grounding are determined from finite element analysis, with the

explicit FE code LS-DYNA. It is commonly accepted that for transient and large

deformation analysis, explicit solution algorithms are best suited. As Kitamura [1]discusses, the great advantage is that numerous elements can be handled, without

having to deal with large stiffness matrices. The disadvantage is reduced accuracy due

to the explicit formulation itself. This can however be corrected by reducing the time

step or refining the mesh.

Finite element analysis put into a collision and grounding aspect, have undoubtedly a

great potential. However, the method is tremendously CPU demanding, and so

analysis may last for days or weeks if, the mesh is dense enough. This makes

modeling caution important. One way to avoid unnecessary computation time, is to

find an adequate mesh with respect to accuracy, which will catch the structuralbehavior and energy response.

Mesh Convergence Study

Grounding problems usually implies large deformations, fracture and material non-

linearities. Ideally, a very dense mesh should be applied in order to catch distortions

and instabilities. However, these damaged areas are fairly local, compared to the rest

of the ship which remains in an elastic / low plastic deformation region, where coarser

meshes are adequate. A combination between dense mesh in the contact area, and

coarse discretization in boundary zones, may be a way forward to achieve accuracy.

However, the nature of the explicit solver, is such that it is conditionally stable withrespect to time increments. Hence, very small elements will decrease the time steps

and extend the simulation time. Care must therefore be exercised, both in avoiding too

coarse meshes due to reduced accuracy, as well as too fine meshes causing

unnecessary CPU demanding simulations. For this reason, mesh convergence studies,

of highly deformed structural members should be performed. In this way, a mesh can

be found which balances the demands for both acceptable solution time and accuracy.

In the grounding scenarios described earlier, intersections between bottom girders and

transverse frames experience extreme plasticity. These are positioned vertically,

between the inner and outer bottom plating, creating a violent folding pattern duringgrounding. Figure 4 shows parts of the bottom section during grounding with a reef.


8/13

Seen from a global position. figure (a), the mesh is observed to be very dense. When

closing in to figure (b), it is seen that the elements which are 22 cm of length are on

the very limit to be able to give a physical representation of the panel folding. In order

to avoid overly stiff response and non-conservative capacity predictions, it is

important to find an adequate mesh through convergence studies.

(a) (b)

Figure 4: Crushed bottom structure, showing (a), the girder-frame arrangement for

the left tank without inner the bottom, and (b) closeup deformation.

A convergence study has been carried out for parts of the bottom structure,

represented by the intersection between a transverse frame and a longitudinal girder.

The model, which constitutes a cruciform is being evenly crushed in the vertical

direction, while symmetry conditions are assumed at the free edges. The structure is

2680 mm high, 15 mm thick, and is modeled with four different mesh sizes, the 30t,

15t, 7.5t, and 4tmesh, where tis the plate thickness. As seen from figure 5, the coarse30t mesh, gives a very different folding mode compared with denser meshes. The

result of this is a much too stiff response.

Imperfections are important in buckling analysis, and affects the results initially by

governing the folding shape. It is however reasonable to assume that they play a minor

role in the crushing process of a large and complex structure such as a ship section.

This is because deformation of one part will distort neighboring pars, and thereby

introduce new imperfections as the structure deforms.

(a) (b)

Figure 5: Bottom frame-girder intersections crushed in the vertical direction, (a) the

coarse 30t mesh, and (b) the 7.5t mesh.


9/13

Furthermore, a study on the effect of vertical stiffening of the transverse girder has

been performed. Simulations show, a dramatic increase of dissipated energy. In

addition, all models seem to exhibit a single folding shape. It is likely that the extra

constraints provided by the vertical stiffeners reduce the number of low energy modes,

which yields the resulting folding effect, figure 6. Much of the same effect was

witnessed by Amdahl and Kavlie [6] where the side structure of the cruciform was

fixed in the vertical direction, which gave birth to large shear forces in the vertical

direction and tensile forces in the horizontal direction. Again, extra constraints

resulted in fewer folding modes, which leads to faster force convergence.

(a) (b)

Figure 6: Bottom web-structure with vertical stiffening, showing (a) the 15t mesh,

and (b) the 4t mesh.

Figure 7 shows that energy convergence is achieved with element sizes denser than 15

t. It is however easy to be fooled by the smoothness of energy curves. A comparisonwith the force-displacement curves given in figure 8, shows that a mesh size of 15t

behaves too stiff and is on the very limit of what can be recommended. For large ship

models, the 7.5t mesh is too costly. Another mesh have for that reason been tested.

Positioned between the 15tand 7.5tmesh, the new 10tmesh shows good convergence

qualities, figure 7 (b) and figure 8, without being too CPU expensive.

It is interesting to observe that coarser meshes experiences the stiffening effect from

self contact later than dense meshes. This is due to the problems of representing the

folding shape at large deformations. Self contact will thereby occur much later in the

process, which gives completely wrong results. From this it is reasonable to conclude

that a mesh size of 15t= 225 mm is at the very edge of what can be acceptable with

respect to accuracy. Smaller elements such as the 10 t are therefore recommended in

such structures.


10/13

(a) (b)

Figure 7: Displacement-energy curves, showing (a) the relation between stiffened and

unstiffened cruciforms and the relation between mesh sizes. Figure (b),

illustrates the convergence in energy due to mesh refinement of the stiffened

cruciform.

Figure 8: The load displacement development of different mesh sizes, for the stiffened

cruciform.

The Ship Model

Several ship models have been made. All of them are limited to two tank sections of

an oil tanker, depending of which scenarios to be simulated. Familiar for them all, is

that they are held in place by to rigid transverse frames at each beam end, figure 9.

These are allowed to rotate about their neutral axis during deformation. For grounding

taking place in the middle of an oil tank, the model is represented by one finely

meshed tank section, and two coarser half tanks on each side. In the case of crushingof the transverse bulkhead, the problem is modeled as shown by figure 9 (a).


11/13

(a) (b)

Figure 9: Simple grounding models of the hull beam seen from the side, showing (a)

reef crushing into the the transversal bulkhead, (b) penetration of mid tank

section.

The ship hull is 42.5 m wide with a center longitudinal bulkhead, the length of one

tank is 32 m. Using the 15t mesh creates a model with the total number of 500000elements. Seen from the global perspective the model is very large. However, it is not

able to give a correct representation of the bottom section response. By applying the

10tmesh and increasing the number of elements, the accuracy problem can be solve,

but then at a higher CPU cost.

Material and Fracture

The ship hull and frames consist of two different grades of steel. Side and deck

sections are built by mild steel, while the bottom section is dominated by a higher

strength steel with a yield strength of 360 MPa.

A challenge in the analysis of large deformation problems, such as grounding, is to

predict fracture initiation and crack propagation. This is commonly done by removing

elements when a critical effective strain is exceeded. The method is very simple, yet

one of the few methods that exist in many FEA codes. Utilizing effective strain as a

fracture criterion is however a very rough generalization of the true behavior. This is

due to the fact that the criterion is constant, while the plastic strain at fracture is

different for various states of stress, such as compression, shear and tension. A

constant effective strain criterion may therefore introduce fracture at the wrong stage

of deformation. It may even lead to rupture in a pure compression state, and must

therefore be used with care.

A variety of damage models are available today. Many of them are very complex and

require a wide range of test parameters. Material data for ships are however very

limited, and often minimized to a few parameters such as yield stress, peak stress and

fracture elongation. A damage criterion should preferably be simple, and require a

minimum of parameters found from tests. Trnqvist [5], have successfully combined

two damage criteria which rely upon one test parameter only, given by Crockcroft -

Latham [1] and Rice and Tracey [3]. When combined, they cover a wide range of

stress states, and because its non complex formulation, the damage model often

referred to as the RTCL criterion, will be implemented into a user-defined materialmodel in LS-DYNA, for applications in the grounding problem.


12/13

It is a general problem for FE methods when predicting fracture that element

representation of the structure tend to be highly mesh size dependent. That is, the

local stresses and strains will for different mesh sizes undergoing large deformations,

be different. Fracture models based solely on one element strain or damage values

may for other element sizes lead to offset tolerances and completely wrong fracture

predictions. For ship structures, the element size tend to be very large, often more than

15 times the plate thickness. It is then hard to catch instabilities such as localizing of

stresses and strains. Simonsen and Trnqvist [4] proposed a solution to this, by

determining a relationship between damage and mesh size. This can be implemented

as a subroutine into LS-DYNA such that different sizes of elements will fail at the

correct time.

6. CONCLUSIONS

The basic ideas behind the intentional grounding project have been outlined. Theambition of the work is to predict the consequences of grounding with such a

precision that decision whether a disabled ship may be safely stranded (integrity

maintained, no or minimal oil spill), can be made with a high degree of confidence. In

addition to being reliable the calculations must be carried out within seconds in order

to allow for fast evaluation of alternative actions, to maintain integrity of the vessel.

The idea is therefore to combine load deformation relations from finite element

analysis with fast simplified rigid body calculations. For that reason finite element

accuracy is very important. It is however necessary to find a balance between accuracy

and simulation time, especially for large models such as ship models. The way

forward may therefore be to perform studies of convergence of critical parts of theship structure. Such a study has been outlined for a stiffened cruciform.

REFERENCES

[1] Crockcroft, M., G. and Latham, D., J., 1968, Ductility and the Workability of

Metals,J. Ist. Metals, 33:96

[2] Kitamura, O., 2002, FEM Approach to the Simulation of Collision and

Grounding Damage, Marine Structures, vol 15, pp 403-428

[3] Rice, J. and Tracey, D., 1969, On the ductile enlargement of voids in triaxial

stress fields,J Mechanics & Physics of Solids, v 17, n 3, pp. 201-217.

[4] Simonsen, B., C. and Trnqvist, R., 2004, Experimental and Numerical

Modeling of Ductile Crack Propagation in Largescale Shell Structures,

Marine Structures, Vol. 17, pp. 1-27.

[5] Trnqvist, R., 2003, Design of Crashworthy Ship Structures, PhD. Thesis,

Department of Mechanical Engineering, The Tech.Univ. of Denmark


13/13

[6] Amdahl, J. and Kavlie, D., Experimental and Numerical Simulation of

Double Hull Stranding, DNV - MIT Workshop on "Mechanics of Ship

Collision and Grounding" DNV Hvik, Oslo, 1992

[7] Amdahl, J., Kavlie, D., Johansen, A., Tanker Grounding Ressistance,

PRADS, 1995

Documents

Marine2005 Paper a&A