You are on page 1of 8

Collision and Grounding of Ships and Offshore Structures – Amdahl, Ehlers & Leira (Eds)

© 2013 Taylor & Francis Group, London, ISBN 978-1-138-00059-9

Collision tests with rigid and deformable bulbous bows driven against
double hull side structures

I. Tautz, M. Schöttelndreyer, E. Lehmann & W. Fricke


Hamburg University of Technology (TUHH), Hamburg, Germany

ABSTRACT: Evaluation of ship collision safety is commonly carried out with a striking ship assumed to be
rigid. This approach is in line with classic safety assessments taking a worst case scenario into account. It is
in addition to that justified because the majority of ship-ship collisions occur without significant damage at
the striking vessel. Nevertheless considering stiffness of the striking vessel might have noticeable influence
on absorbable collision energies. Safety level could be increased significantly when design of bulbous bows
is carried out also under the aspect of good crushing behaviour regarding collision load. However up to now
experimental verification of this context is missing. This paper reports about collision experiments with rigid
and deformable bulbous bows driven against double hull side structures that have been carried out on the test
facility of the Institute for Ship Structural Design and Analysis of TUHH. Actual validation status of numerical
calculations is presented based on test results also with regard to material behaviour.

1 INTRODUCTION

Experimental investigations of ship collisions that con-


sider as many mechanical effects as possible are in
general highly complex. Thus experiments usually
concentrate on the inner mechanics of one collision
partner only, whilst the other one is considered to be
rigid.
Yamada (2006) tested bulbous bows of a fishing
Figure 1. Experimental program.
ship with 500 in gross tonnages in true scale. Amdahl
(1983) scaled down and simplified bulbous bow struc-
tures as tubes with circular and elliptical cross-section.
Both authors crushed their specimen by driving them collision forces and energies as well as to material
against rigid plates. behaviour of the used steel grade.
Ship collision tests with rigid indenters driven Effect of the side structure filled with granulate
against ship like side structure models are known from is presented in a separate paper by Schöttelndreyer
Peschmann (2001) or Karlsson (2009). Compared to (2013) who compares CE-1 with CE-3 and gives
Karlsson, Peschmann used a larger test model but did additional information with regard to the material
not have laboratory conditions because his tests were behaviour of the granulate.
carried out in open air with floating models. CE-4 is scheduled for January 2013; results have
In contrast to the above mentioned publications, not been available for this paper.
experiments presented in this paper consider stiffness
of both collision partners in a series of systematic vari-
ations. Test models of bulbous bow like indenters in
2 EXPERIMENTS
rigid and deformable configuration are driven against
models of ship side structures. Side structures repre-
2.1 Collision test rig
sent a conventional longitudinal stiffened double hull
design. Stiffness of the side structures is varied by Experimental investigations presented within this
filling them up with granulates.These models are com- paper consider a right angle collision. This scenario is
bined in an experimental program with a total number well established in evaluation of ship collision safety.
of four collision tests shown in Figure 1. It has in addition to that a lot of advantages regarding
This paper focuses on the comparison of CE-1 and the execution of the tests. Experiments are limited to
CE-2. Collision test rig, models and test procedure inner collision mechanical effects caused by bulbous
are described. Results are discussed with regard to bow indentation. Ship motions are not considered.

93
thickness: 14.2 mm; flange: Ø 740/Ø 940 × 10 mm,
bolt circle: Ø 876/4 × Ø 22 mm).
The test model of the struck ship is a conventional
double hull side structure. The model area is welded
within a massive supporting frame that represents the
boundary conditions of the surrounding structure. The
model area is symmetric about all coordinate planes.
The side structure is connected on both sides in x-
direction to supports with four connection rods each.
These supports also accommodate forces in z-direction
with a total number of four pressure load cells, one on
each model corner.
Supports and hydraulic cylinders are fixed perma-
nently to both longitudinal girders. Thereby a closed
flow of forces is provided.
If not indicated otherwise all descriptions refer to
the coordinate system shown in Figure 2. Its origin is
located in the intersection point of all three symme-
try planes of the ship side structure model area. Please
note that this definition does not correspond with com-
mon naval architectural annotation of ship coordinate
systems.
The whole test rig is symmetric about x-z plane and
y-z plane.
Displacements, forces and strains are measured on
several positions of the test rig and monitored by a
multi-channel data acquisition unit with up to 118
channels (Autolog 3000 with three boards CM3410,
Peekel-instruments). Fifty five channels are used to
monitor elastic behaviour of the test rig. Fifty eight
channels are in variable use for strain gauges on the
models. The five remaining channels collect the most
Figure 2. Collision test rig – total and detailed view.
important data: collision force and displacement of
the bulbous bow. The displacement is measured in z-
direction by a wire-actuated encoder with a measuring
length of 2000 mm from the aft end of the bulb to a
Test models that represent the interesting area of
point fixed in space. Thus a clear measuring quantity is
the collision partners are scaled down with a fac-
gained that is independent from elastic deformations of
tor of about 1:3. This scaling was chosen to provide
the test rig and from changing configuration of interim
comparability to other experiments e.g. published by
pieces between bulb and cross-beam as well. This is
Peschmann (2001).
important to provide correct comparability with simu-
Figure 2 shows the test rig of TUHH, Institute for
lations that consider neither cross-beam with cylinders
Ship Structural Design andAnalysis. Collision process
and longitudinal girders nor interim pieces. Collision
is carried out in vertical moving direction. The test rig
force is interpreted as the sum of the measured forces
is mounted on two longitudinal girders (17 m long, 2 m
at the pressure load cells described above.
high).
The whole test procedure is monitored with eight
Collision forces are applied by four servo-hydraulic
cameras from different positions. Recording of all sig-
driven cylinders. They are mounted on the longitudinal
nals is with chronological synchronism. Thus very
girders with an x-distance of 10 m and are connected
detailed analysing of most important incidents is
with the cross-beam. The maximum loading capacity
possible.
is 4000 KN.
The test model of the striking ship is a rotational
symmetric bulbous-bow-like indenter. It is connected
2.2 Test models – side structure
to the cross beam with z-axis as axis of rotation.
Hydraulic cylinders are limited to 400 mm regard- Design of the test models is derived from ship
ing the maximum range of displacement. Larger class ConRo 220 built by Flensburger Schiffbau –
displacements are realized by using appropriate Gesellscharf mbH, Germany. The model area repre-
interim pieces between the bulbous bow and the sents the longitudinally stiffened double hull structure
cross beam. Interim pieces are available in sev- surrounding the lower hold of this RoRo-vessel.
eral lengths and are manufactured out of a spi- Dimensions of the structure are described in Fig-
ral welded pipe with flanges (pipe according to ure 3 and Table 1. They have been kept constant for all
DIN EN 10220, external diameter: 813 mm, wall four collision experiments.

94
Figure 4. Detailed view on grids on outer and inner shell
plating.

Figure 3. Top view on side structure, dimensions in [mm].

Table 1. Dimensions of side structure.

Component Dimensions [mm]

Outer an inner shell t = 4 mm, Grade A


Web frames t = 5 mm, Grade A
two manholes, no stiffeners
Longitudinal frames HP 140x7, Grade A
Model height (z-length) 900 mm
Supporting frame t = 20 mm, Grade A36
Figure 5. Rigid test model of bulbous bow, dimensions
in [mm].

For orientation on outer and inner shell a rectangular


grid was applied in the contact area of the indenter
(Figure 3, details: Figure 4). Edge lengths of the grid
are as follows.
X-length: 66.67 mm (1/12 × web spacing)
Y-length: 70.00 mm (1/4 × long. frame spacing)
Collision tests CE-1, CE-2 and CE-3 are addition-
ally equipped with a grid of circles. The circles are
marked with template and scriber. Thus quite accurate
measurements of diameters even in the contact area are
possible when circles become ellipses caused by large Figure 6. Test model configurations of bulbous bow.
plastic deformation. Circles with 20 mm and 50 mm in
diameter have been used according to Figure 4.
in diameter, 30 mm wall thickness. Material thickness
Consistent numbering allows clear identification of
increases smoothly in the shaped fore end up to a value
each circle and quadrant.
of 150 mm in the bulb tip. The model is fitted with sev-
eral flanges to mount it at the cross beam (flange A), to
vary the model length in the cylindrical part (flange B)
2.3 Test models – bulbous bow
and to replace the rigid tip with a deformable one
Outline of the bulbous bow is based on the geome- (flange C).
try of pre-tests with deformable bulbous bows driven Experiments with deformable bulbous bow are car-
against rigid walls described by Tautz (2010). In order ried out with a sheet metal cap, 3 mm thick, welded on
to get a rigid, rotational-symmetric indenter, the bul- a flange ring that matches to flange C. Slight differ-
bous bow was manufactured as a turning work piece ences to the geometry of the rigid tip had to be accepted
out of a forged round bar welded on a pipe 813 mm because of manufacturing reasons (Figure 6).

95
Figure 8. History of collision forces of CE-1 and CE-2.

Figure 7. Adjusting the history of forces by classification


of incidents using the example of CE-1.

2.4 Test procedure and data processing


Tests are carried out displacement-controlled with
a velocity of 0.2 mm/sec. This velocity is assumed
to be small enough to regard the test procedure as
quasi static. Sample rate of the multi-channel data
acquisition unit is 2 Hz. The test is interrupted by sev-
eral unloading operations to insert additional interim
pieces between bulbous bow and cross-beam. Addi-
tional interruptions at certain loaded conditions are
carried out to visually analyse the deformed test
model, failure modes, crack pattern, etc. Each test
Figure 9. Cross-section of schematically illustrated
demands an effort of four to five working days with deformed shape just before crack initiation (*) at outer shell
all interruptions (preparation and post processing not for CE-1 and CE-2.
included).
Force displacement curves are carefully processed
regarding significant changes in the run of forces also Points 1: outer shell rupture occurs at an unexpected
using extensive video-data. Thus almost every sig- early stage at CE-1 and is followed by continu-
nificant change in the run of forces can be clearly ing increase of force. This early outer shell rupture
dedicated to a precise incident of the test and is is caused by local bending around the longitudinal
classified as follows: stiffeners in the contact area (Figure 9). The outer
shell rupture of CE-2 occurs at a displacement about
– Incidents of major relevance like crack opening of 420 mm later than in CE-1. Penetration is significantly
outer and inner shell, collapse of stiffeners or web- less sharp. Contact forces are well distributed on shell
frames. and stiffeners.
– Incidents of minor relevance like changes in the Points 2: after crack initiation of outer shell lon-
path of cracks, crack openings at stiffeners or other gitudinal stiffeners reach their ultimate load after
structural elements except outer and inner shell. significant increase of displacement in CE-1. Collapse
– Incidents of negligible relevance like un-reloading of stiffeners in CE-2 is compared to CE-1shortly after
cycles or temporary decrease of forces caused by rupture of outer shell. During the further procedure
interruptions of the test in loaded condition at a forces are caused just by friction between bulbous bow
fixed value of displacement. and parts of the side structure until reaching points 3.
For subsequent scientific work original test data is Points 3: bulbous bow (or deformed parts of the
adjusted by ignoring all incidents of negligible rele- outer hull, only CE-2) comes in contact with inner hull
vance. Figure 7 gives an impression of this approach. respectively longitudinal stiffeners of the inner hull.
All subsequent descriptions will use adjusted curves Points 4 (only CE-2): local crack initiation at inner
only. shell occurs within the range of transversal intersection
(web/stiffener). This effect is supposed to be negligible
and seems to be avoidable by appropriate application
3 RESULTS of collar plates.
Points 5: rupture of inner shell occurs at larger
3.1 Effect of deformable bulbous bow displacement in CE-2 compared to CE-1.
Points 6: collapse of stiffeners occurs. Tests are
The effect of a deformable bulbous bow will be stopped shortly after this incident.
explained by a detailed description of the force- Regarding collision energies shown in Figure 9 it
displacement curves of CE-1 and CE-2 (Figure 8). becomes obvious that the test with deformable bulb

96
Figure 10. Collision energy versus displacement of CE-1
and CE-2. Figure 11. Strain measurements at 20 mm-circles. Division
in groups exemplary for CE-1 in the range of crack initiation.
(CE-2) absorbed around twice the energy of CE-1 at
the point of inner shell rupture (Point 5 in Figure 8).
Inner shell rupture occurs at significantly larger dis-
placement which would also lead to larger penetration nearby (154 measurements) and group C that are direct
of the fore body over water. Thus true-scale energy neighbours of Group B (351 measurements).
difference may differ from experiment. Energy differ- The chosen grid is in general only suitable to res-
ence at outer shell rupture is even more impressive olute plastic strain distribution quite roughly. To get
but should not be generalized because of the large more accurate information forming limit diagrams
differences in failure mode between CE-1 and CE-2 have to be determined with additional metal forming
(Figure 9). experiments defined in DIN EN ISO 12004. Hogström
(2009) published results of these kind of tests with
NVA steel grade, a mild steel which is also used in the
3.2 Strain measurements at circles model area of the side structure (A-Grade, classified
by Lloyd‘s Register). In order to produce necking and
Outer and inner shell of the test model is equipped fracture at different strain states six different specimen
with a grid of circles that is described in chapter geometries were tested. Corresponding strain states at
2.2. All circles have been applied on the plates previ- fracture are described as points in the principal strain
ous to model production. Variations in circle diameter space for each geometry. Necking is described by a
may occur caused by welding heat or by individual limiting curve based on the Bressan–Williams–Hill
handling of template and scriber. Thus every circle criterion proposed by Alsos (2008) and fitted to the
was photographed with a measuring tape after all test results by Hogström (2009).
welding works have been finished (photos of 20 mm- Figure 11 shows the measured strain state at the
circles in groups of four circles per photo). Subse- 20 mm-circles in the range of crack initiation for CE-1
quent processing of relevant photos with CAD-tools in comparison with the data published by Hogström
delivered reliable values for diameters in undeformed (2009).
condition (d0 ). Obviously measurements with the above mentioned
According to Hasek (1973) principal strains can be method are significantly smaller than those carried
determined from measured lengths of the principal out by Hogström. This can be explained by the well-
axis of ellipses (d1/2 ) as follows: known dependency of measurements from the chosen
reference length described e.g. by Ehlers (2009).
Although the chosen method is quite rough com-
pared to optical measurement procedures it is never-
theless accurate enough to resolute decrease of strains
with increasing distance from the crack. It is suitable
Measurements have been carried out with a calliper to derive some statements regarding failure criteria for
with extra fine jaws. Accuracy of measurements is FE-calculations with thin shell elements of an edge
within a range of some few tenth millimetres. length greater than five times the thickness:
For this paper only 20 mm-circles of CE-1 and CE-
2 have been processed. No significant differences in – Strain state seems to be of negligible influence, fail-
measured strains could be observed between the tests ure strain seems to be dominated by first principal
or the location of the circles on outer respectively inner strain in general.
shell. Measured circles are classified according to their – Failure strains seem to be in the range of or even less
distance to a crack in three groups: group A for circles than necking strains received with high resolution
that are divided by a crack (84 measurements), group methods (Highest measured values 29.4%, 28.2%,
B that are not divided by a crack but that are located 27.8%).

97
Figure 12. Geometry of specimens for uniaxial tensile tests.

Figure 13. Determination of True Stress Strain relationship


3.3 Stress-strain relationship using the example of structural element shell (CE-1).
During manufacturing process of the models material
pieces of structural elements were retained for sub-
sequent material tests. For this paper uniaxial tensile
tests according to EN-ISO 6892-1 have been carried
out. Material pieces from shell, longitudinal stiffeners
and web frames have been tested; geometry of speci-
men is shown in Figure 12. Tests were carried out with
a MTS 810 Material Test System. Straining of the spec-
imens is measured with an extensometer 634.25F-24
(MTS) with a gage length of 50mm. Movement speed
of the machine was 0.04 mm/s. Material tests for tested
structural elements were repeated minimum five times
with small scatter. Engineering stress-strain relation-
ship was determined as arithmetic average over all tests
per structural element (ESSC-Test, see Figure 13).
True stress-strain relationship was determined
by using the power law relationship proposed by
Hollomon (1945):
Figure 14. Validation of TSSC-FEM with implicit simula-
tion of tensile test using the example of struct. elem. shell
(CE-1).

where σ is the true stress, ε is the true (logarithmic) are plotted as reaction forces divided by initial cross-
strain, K is the material’s strength coefficient and n is section. Thus an engineering stress-strain relationship
the strain hardening exponent. was determined with the FEM calculation (ESSC-
True (logarithmic) stress-strain relationship from FEM) that is comparable with results from tensile
tensile test (TSSC-Test, see Figure 13) is used to deter- tests. Good agreement was achieved for all tests; one
mine material parameter K and n by a least square example is given with Figure 14.
fit within the range of uniform elongation. Values up
to the point when yielding is securely terminated are
disregarded for the approximation. True stress-strain
4 NUMERICAL SIMULATION
relationship for FEM-simulations (TSSC-FEM) is a
combination of TSSC-TEST in the range of yielding
4.1 Model description
and Formula (2) for larger strain values.
TSSC-FEM data was used for an implicit FEM- Numerical simulations are carried out with a FE-
calculation of the tensile test in order to check appli- model whose geometry is shown in Figure 15.
cability of the above mentioned power law relationship Bulbous bow is represented as a rigid body driven
also for higher strain values. Therefore a FEM-model against the side structure with constant velocity. Shell
with solid elements was used (four elements through elements with a mean edge length of 20 mm are used.
the thickness). Side structure is modelled according to design
Engineering strain was determined by plotting dis- drawings with shell elements. HP-profiles are rep-
placements of control-nodes located on positions iden- resented as L-Profiles with appropriate moment of
tical with measuring points of extensometer. Stresses inertia. Mean shell element edge length is 34 mm;

98
Figure 15. Geometry of FE-model with boundary condi-
tions (Only bulb tip is displayed for better overview).

aspect ratio is almost equal to unity in the model area.


LS-DYNA contact type “automatic single surface” is
used with a static friction coefficient of 0.23 (based on
considerations described in 4.3). Nonlinear material
behaviour is considered by LS-DYNA material type Figure 16. Reaction force a) and collision energy b) of
*MAT 123 (modified piecewise linear plasticity) with experiment and calculation (CE-1).
true stress strain relationships from chapter 3.3. Failure
is represented with an equivalent plastic failure strain Figure 16 shows reaction forces and collision
criteria being dependent from element edge length energy of CE-1 in comparison between simulation
l and thickness t that was proposed by Peschmann and experiment. Although a comparably simple fail-
(2001) for shell elements with a thickness less than ure criterion is used simulation is a good match to the
12 mm as follows: experiment.
Differences in the maximum values of forces at
the point of shell rupture are significant but seem to
be acceptable with regard to repeat accuracy of the
tests and the good agreement in the History of energy.
Side structure is connected to supports with LS- Better correlation is assumed to be achievable with
DYNA trusses of type 3 in order not to transmit any finer meshes and more sophisticated failure models.
moments. Supports are meshed with shell elements For example local bending that is responsible for outer
(mean edge length: about 80 mm). Fixation of the sup- shell rupture in experiment cannot be represented by
ports at the longitudinal girders is considered with the chosen approach.
boundary conditions and trusses according to Fig- Although some compromises are to be made with
ure 15. Elastic material (*MAT 001) is defined for regard to details of fracture mechanics, the simulation
supports and trusses with standard parameters for mild is validated quite well by the experiment. Collision
steel. energy at the point of inner shell rupture is calcu-
All shell elements are defined with four noded lated with high accuracy and is the most important
quadrilateral Belytschko-Lin-Tsay formulations parameter in the evaluation of collision safety.
with five integration points through their thickness.
LS-DYNA Version 971/ R6.1.0 are used. 4.3 Comments on validation process
The mesh size in the model area was chosen on purpose
as coarse as described above. True-scale simulations
4.2 Results
shall be possible based on the presented results without
Karlsson (2009) carried out collision tests with two excessive adaptation of the mesh acuteness. For that
similar test models of a ship side structure. He achieved reason validation of the simulation was carried out
a repeat accuracy of about 10% regarding the peak val- without any changes regarding element sizes.
ues in the history of forces. It is assumed that repeat As a result of several calculations with different
accuracy of the tests presented in chapter 2 is in com- values for the coefficient of friction it can be stated
parable range or even worse because complexity of that consideration of friction is a must. It is just negli-
the rig and the models is even higher compared to gible for very small displacements previous to outer
Karlsson. Differences between calculation and exper- shell rupture. Standard value of 0.1 for the coeffi-
iments have to be evaluated with regard to this cient of friction (steel to steel) delivered significantly
background. lower forces than described in Figure 16. Because

99
roughness of contact surfaces, which is high partic- the cross-beam and two supports for the test-plant as
ularly at broken edges, the coefficient of friction has well as the test models.
to be increased to values of 0.23 to 0.3 to get good
correlation with test results. A value of 0.23 was cho-
sen because this was experimentally approved also by REFERENCES
Karlsson (2009).
Although the possibilities of the test rig allows quite Alsos HS., Hopperstad OS., Törnqvist R. and Amdahl J. 2008.
Analytical and numerical analysis of sheet metal instabil-
accurate positioning of the models several calcula-
ity using a stress based criterion. International Journal of
tions have been carried out to check the influence of Solids and Structures 45: 2042–2055.
changes regarding the first point of contact. Assum- Amdahl J. 1983. EnergyAbsorption in Ship-platform Impacts.
ing a maximum eccentricity of 10 mm bulbous bow Report No. UR-83-34. Trondheim: The University of
was positioned with values of 0, 2, 5 and 10 mm in Trondheim
all possible combinations regarding positive x- and Ehlers S., Varsta P. 2009. Strain and stress relation for non-
y-coordinates. Changes in the history of forces were linear finite element simulations. Thin-Walled Structures
found to be negligible. 47:1203–1217.
Hasek V. 1973. Über den Formänderungs- und Span-
nungszustand beim Ziehen von großen unregelmäßigen
Blechteilen. Report No. 25, Institut für Umformtechnik.
5 CONCLUSION Stuttgart: Universität Stuttgart.
Hogström P., Ringsberg JW. and Johnson E. 2009. An exper-
A collision test rig and procedure for comparably large- imental study of the effects of length scale and strain state
scaled test specimen is presented. Collision tests with on the necking and fracture behaviours in sheet metals.
deformable and rigid bulbous bows driven against ship International Journal of Impact Engineering 36(10–11):
side structures are described and corresponding effects 1194–1203.
are figured out. Significant increase of about 100% Hollomon JH. 1945. Trans. AIME 162: 268–290
in absorbed energy is achieved with deformable bul- Karlsson UB., Ringsberg JW., Johnson E., Hoseini M. and
Ulfvarson A. 2009. Experimental and Numerical Investi-
bous bow compared to a rigid one. These tests are
gation of Bulb Impact with a Ship Side-Shell Structure.
integrated in an experimental program with a total of Marine Technology 46(1): 16–26
four tests that will be finished in spring 2013. Numer- Peschmann J. 2001. Berechnung der Energieabsorption der
ical validation of the first test has been carried out Stahlstruktur von Schiffen bei Kollision und Grund-
with experimental results and additional examinations berührung. Dissertation. Hamburg: Hamburg University
of material strains in the plastic region. Results show of Technology.
good agreement between measurement and calcula- Schöttelndreyer M., Tautz I., Fricke W. and Lehmann E.
tion. More results are presented by Schöttelndreyer 2013. Side Structure filled with multicellular glass hol-
(2013) and by the end of 2013 when research work low spheres in a quasi-static collision test. Proc. of 6th
Int. Conference on Collision and Grounding of Ships
will be finished.
(ICCGS). Trondheim: Norwegian University of Science
and Technology
Tautz I., Schöttelndreyer M., Fricke W. and Lehmann E. 2010.
ACKNOWLEDGEMENT Experimental Investigations on Collision Behaviour of
Bow Structures. Proc. of 5th International Conference on
The work presented in this paper was performed Collision and Grounding of Ships; June 14–16; Espoo,
within the research Project ELKOS, funded by Finnland: 179–183. Ehlers S, Romanoff J. (eds).
German Federal Ministry of Economics and Technol- Yamada Y. 2006. Bulbous Buffer Bows: A Measure to Reduce
ogy (BMWi) under project no. 03SX284B.The authors Oil Spill in Tanker Collisions. Dissertation, Lyngby:
Technical University of Denmark.
are responsible for the content of this paper and wish
to thank for supporting this project. The authors’ grat-
itude is particularly addressed to German shipyard
Flensburger Schiffbau-Gesellschaft which delivered

100

You might also like