Marine-2013 - 78-Structural Desin of A Bulbous

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 11

Structural Design of a Bulbous Bow with Regard to Collision Safety

V International Conference on Computational Methods in Marine Engineering


MARINE 2013
B. Brinkmann and P. Wriggers (Eds)

STRUCTURAL DESIN OF A BULBOUS BOW


WITH REGARD TO COLLISION SAFETY
I. Tautz*, M. Schöttelndreyer*, J. Gauerke*, E. Lehmann* and W. Fricke*
*
Hamburg University of Technology (TUHH)
Schwarzenbergstrasse 95c, 21073 Hamburg, Germany
e-mail: tautz@tuhh.de, web page: http://www.tuhh.de/skf

Key words: Structural Design, Crashworthiness, Ship Collision, Bulbous Bow, Buffer Bow

Abstract. Dimensioning of bulbous bows is based on conservative and usually empirically


developed Classification Rules. Furthermore their structural design is subordinate to
manufacturing aspects. Thus bulbous bow structures are mostly overdimensioned and could
be regarded as rigid in case of a ship collision. Weakening bulbous bows stiffness results in
noticeable influence on absorbable collision energies. Anyhow this approach is limited
because usual operating loads like hydrostatic pressure or slamming loads are still to be
considered. The paper describes a design approach for bulbous bows that aims at the best
compromise between preferably weak regarding collision load on the one hand and preferably
stiff under operating load on the other hand. Linear and nonlinear FEM-calculations based on
ultimate load theory are combined in order to estimate stability of weakened designs.
Examinations are accompanied by nonlinear explicit FEM-calculation of a ship collision
scenario taking into account stiffness of striking and struck ship.

1 INTRODUCTION
In the overwhelming majority of cases, ship-ship collisions lead to damages that are more
severe at the struck than at the striking ship. At worst this leads to the loss of the struck ship
with all well-known hazards for crew, passengers and environment. Compared to that damage
at the striking ship is of minor relevance or even negligible. This issue leads to considerations
regarding alternative design of bow structures comparable to the development of crumple
zones in automotive industry.
A first proposal was made by Cheung [1]. He introduced a so-called soft bow design of a
forecastle structure being stiffened with systems of tubes arranged in parallel rows. Although
he estimated energy absorption of his design with quite rough assumptions, he described the
essential principle of function already applicable. If the strength of the soft bow is less than
the struck ship side strength, the soft bow will be crushed. In the contact area of the collision
soft bow flattens to a certain extend. By that collision process will be less sharp and more
kinetic energy can be absorbed until fracture of the side of the struck ship.
In order to gain detailed understanding of this principle, highly nonlinear problems like
large deformation, contact, plasticity and failure have to be solved. Hence, extensive research
on the field of soft bow design was carried out not before a high development status of the
finite element method was reached. In this context two consecutive research projects are to be
highlighted that have been carried out in Japan from 1991-2006. Results have been published

901
I. Tautz, M. Schöttelndreyer, J. Gauerke, E. Lehmann, W. Fricke

amongst others by Kitamura [3], Yamada [8] and Endo [2] with promising results. They
describe a so-called buffer bow design for a bulbous bow which is transversally stiffened and
optimized with special regard to minimized plate thicknesses and minimized inner supporting
structure. Design of the buffer bow is in line with Classification Rules. Thus bow hooks in
comparably close distance are mandatory in the bulb tip. That means reducing axial crushing
resistance of the bulbous bow is limited especially in the area where first contact occurs. Thus
flattening of the bulb tip is not improved. However this is one helpful effect to increase
energy absorption capabilities of struck and striking ship as well. Detailed research regarding
that issue was carried out by the authors since 2008 and published in [6] - [8].
This paper presents a procedure to find a structural design that leads to a flattening of the
tip and that is acceptable for Classification Societies as well. The basic idea is to replace the
conventional bulb tip by a kind of cap, fitted on a watertight tank boundary about one or two
frame spaces aft of the bow tip. The scantlings of this unstiffened cap are proposed to be
developed on basis of a comparison design study independent from existing rules. The cap
may be regarded as an appendix to the ship hull not being relevant for the vessel’s strength
integrity. Anyhow, the strength reduction of the cap is limited because usual operating loads
still have to be considered. Unfortunately Classification Rules do not provide explicit design
loads on bulbous bows for direct calculations. Thus, in cooperation with Lloyd’s Register
(LR), four explicit load cases were proposed representing most important operational
conditions. Additionally one collision load case was defined according to the rules of
Germanischer Lloyd (GL). All load cases were applied to a conventional bulbous bow design
of a ship new building that was approved by LR as a reference. In order to compare the results
all calculations have been repeated using bulbous bow designs with a less strengthened cap as
described above. Results have been evaluated by comparison of stresses and deformations to
the reference structure.
conventional design design with unstiffened
(Model 1) cap (Model 2)

Figure 1: Investigated design variants, outer shell blanked on starboard side

902
I. Tautz, M. Schöttelndreyer, J. Gauerke, E. Lehmann, W. Fricke

2 COLLISION SIMULATION
In order to determine scantlings of the unstiffened cap, that lead to the desired flattening of
the bulb tip, several collision simulations have been carried out for a rectangular, central
impact. Struck and striking ship are identical, with same displacement and draught. These
assumptions follow rules of GL in parts. Both vessels are modelled according to design
drawings of ship class ConRo 2 0 built by Flensburger Schiffbau-Gesellschaft mbH,
Germany.
Simulation model considers the bulbous bow of the striking ship and the double hull of the
lower hold of the struck ship according to Fig. 2. Remaining ship mass of striking ship is
assigned to all nodes at the aft boundary of the bulbous bow. These nodes are coupled for
motion in x-direction, all other degrees of freedom have been fixed. An initial velocity of
5 m/s is given to all nodes of the bulbous bow. Model boundaries of the struck ship have been
fixed in translational degrees of freedom as shown in Fig. 2. No rotations have been fixed.
Model of the struck ship exhibits a web frame spacing of 2400 mm, five web frame spacings
have been considered. Aft and fore end of the struck ship model are fixed in y-translational-
degree of freedom. Penetration occurs in the middle of struck ship model. Calculations have
been carried out with LS-DYNA, Version 971/ R6.1.0.

upper deck
Striking ship: Struck ship:
only bulbous bow is modeled only double hull of lower hold
is modeled.

disp.z=0

main deck
aft bound. bulb. bow: disp.x=0
• remaining ship mass
assigned to nodes.
• nodal constraint for
motion in x-dir. v=5m/s lower hold

x double bottom

230 232 234 236 disp.x=0


Figure 2: Collision scenario and model boundaries

For both vessels shell elements are used - four noded quadrilateral Belytschko-Lin-Tsay
formulation with five integration points through their thickness. Reduced integration was used
for most parts (Type 2) except the outer shell of the bulbous bow where fully integrated
elements were used (Type 16). HP-profiles are represented as L-Profiles with appropriate

903
I. Tautz, M. Schöttelndreyer, J. Gauerke, E. Lehmann, W. Fricke

moment of inertia (struck ship) or as flat bars with added beam-elements for the bulb (striking
ship). Mean shell element edge length is about 80 mm for the bulbous bow and 100 mm for
the struck ship, aspect ratio is almost equal to unity. LS-DYNA contact type “automatic single
surface” and “automatic surface to surface” is used. Nonlinear material behaviour is
considered by LS-DYNA material type *MAT 123 (modified piecewise linear plasticity) with
standard parameters for mild steel. A true stress strain relationship for mild steel, determined
from tensile tests as described in [7], is used. Failure is represented with an equivalent plastic
failure strain criteria being dependent from element edge length l and thickness t that was
proposed by Peschmann [4] for shell elements with a thickness less than 12 mm as follows:
t
ε f = 0.1 + 0.8 (1)
l
And for shell elements with a thickness greater than 12 mm as follows:
t
ε f = 0.08 + 0.65 (2)
l
Failure is considered just for the struck ship. As experiments show [6], [8] failure is
negligible for bulbous bow structures collapsing in progressive folding mode. Thus no failure
criteria is applied for the striking ship model.
Calculations have been carried out with a striking ship consisting of model 1 and model 2
shown in Fig. 1. Model 2 was calculated with varying shell thickness. Results have been
evaluated regarding the degree of deformation just before rupture of outer shell (Fig. 3).
1
LCAP
0.8 Lcrushed
Lcrushed / LCAP

0.6

0.4

0.2

0
0 1 2 3 4 5

model 1 model 2, cap without any stiffening


shell=16mm shell=16mm shell=15mm shell=14mm shell=13mm shell=12mm
Figure 3: Degree of deformation just before rupture of outer shell for all calculated variants

904
I. Tautz, M. Schöttelndreyer, J. Gauerke, E. Lehmann, W. Fricke

As a result of the collision simulations it can be stated that the maximum crushed length
Lcrushed is about 80% of the length of the unstiffened cap LCAP. A design with unstiffened cap
but with shell thickness equal to the conventional design leads to minor degree of deformation
before rupture of outer shell. Although deformation will increase during subsequent collision
process, further decrease of shell thickness was examined because this study aims the best
possible flattening of the tip at a preferably early stage. A shell thickness of 14 mm seems to
meet this target quite well and will therefore be taken for further investigations under
operating loads.

3 PRESSURE LOADS (LC 1-3)


The design of a bulbous bow with unstiffened cap can be accepted from a structural point
of view in general if it is added as an appendix to the ship hull. Integrity of the remaining
bulbous bow can be provided by Classification Rules. In principle the cap could then be even
left away although this is not a realistic option because of hydrodynamic requirements.
Although requirements of the cap can be formally detached from existing rules, its strength
has to be evaluated in an adequate manner. Thus operating loads have been proposed in
collaboration with Lloyd’s Register as load cases for direct calculations. This chapter
describes hydrostatic and hydrodynamic load assumptions.
Calculations have been carried out with ANSYS Mechanical APDL Release 14.0. Bulbous
bow is discretized as described in Chapter 2. ANSYS Shell Element Type 63 is used (fully
integrated, four nodes), bulbs of stiffeners are represented by Type 8 (truss). Linear elastic
material behaviour with standard values for mild steel is used. Aft boundary of the bulbous
bow is fixed in all degrees of freedom.

3.1 Hydrostatic Pressure (LC 1)


Bulbous bow is exposed to hydrostatic pressure dependent from draught and from a
possible increase due to wave crest. According to rules of LR wave-crest-amplitude can be
calculated based on the given rule length L=184,51 m as follows

C w = 7,71 ⋅10 −2 ⋅ L ⋅ e −0,0044 L = 6,32m (3)

Assuming the increase due to wave crest to be constant over the draught T, hydrostatic
pressure for each shell element is calculated according to its vertical coordinate z as follows

p hs = ρ ⋅ g ⋅ [(T − z ) + C w ] (4)

3.2 Slamming (LC 2)


Rules of Lloyd’s Register provide a quite detailed description of bow flare wave impact
pressure and wave impact pressure on other parts of the side shell plating close to and above
the design waterline (Part 4, Chapter 2, Section 4.2.1). Definition is dependent on x- and z-
position and angle of slope αp, βp and γp described in Fig. 4.

905
I. Tautz, M. Schöttelndreyer, J. Gauerke, E. Lehmann, W. Fricke

Figure 4: Notation for angle of slope αp, βp and γp

Based on the input of these geometric parameters slamming pressure can be determined as

(
Pbf = 0,5 K bf ⋅ Vbf2 + K rv ⋅ H rv ⋅ Vrv2 ) (5)

where Kbf is a hull form shape coefficient for wave impacts (dependent from βp), Vbf is the
wave impact velocity (dependent from αp, x- and z-position), Krv is a hull form shape
coefficient for impact due to forward speed (dependent from αp), Hrv is a relative wave
heading coefficient (dependent from γp) and Vrv is the relative forward speed (dependent from
γp). Described dependencies from z-position find indirect expression by determining a
probability of a wave impact. Z-position is measured from waterline and has direct influence
only on this probability in a squared form. Thus it can be assumed, that formula (5) can also
be applied for slamming loads affecting the bulbous bow, although it is underneath the
waterline.
The described geometric parameters have been determined for each shell element of the
outer shell and an element specific pressure was calculated according to formula (5). Figure 5
gives an impression of the resulting pressure distribution in comparison with the calculated
hydrostatic pressure.
8
Centerline and section at #235
Pressure of LC2 displayed in forepart
7
Hydrostatic Pressure (LC 1)
6
z-coordinate a.b. [m]

5 Slamming at #235 (LC 2,3)


values match the shape

2
#235

0
0 200 400 600 800 1000 1200
Pressure [kN/m²]
Figure 5: Pressure distribution for load cases 1-3, slamming displayed exemplarily at #235

906
I. Tautz, M. Schöttelndreyer, J. Gauerke, E. Lehmann, W. Fricke

3.3 Unilateral pressure (LC 3)


The bulbous bow can be exposed to unilateral pressure e.g. when rolling and/or pitching in
heavy seas. That kind of condition is considered by applying LC 2 described in Chapter 3.2
unilaterally. Rules of GL (Part I, Section 9, A 5.3.3) propose a formula for a constant
unilateral pressure which could be an alternative but is not investigated in this paper.

3.4 Results
In general results of model 2 (unstiffened cap) are evaluated focusing the relation to values
of model 1 (conventional). In some extent evaluation of the differences is more meaningful
than evaluation of absolute values.
Results have been plotted and evaluated for bottom, middle and top layer for all principal
stresses separately for outer shell and inner stiffening system. Inner stiffening and global
bending behaviour of the bulbous bow is not affected in a problematic manner by changes
between model 1 and 2. Thus results are just displayed for a part of the model in front of
#234. Stresses are shown according to LR’s common practice without nodal average,
integration point results have been transferred to nodes without any extrapolation. Main
effects are discussed as follows according to results shown in Fig. 6.
For model 1 it can be seen that stress level in the bulb tip is quite low compared to panels
located behind #236. Stress level in model 2 only increases in the lower part of the cap where
curvature is less. In average it reaches values comparable to the panels behind #236.
Regarding local maxima, values are higher. In the upper part, where curvature is high, stress
increase is somehow surprisingly moderate or even negligible. This can be explained with
regard to theory of curved shells.
Stress level for hydrostatic pressure is moderate. Slamming and unilateral pressure leads to
somehow confusing high stresses for an elastic analysis. Hence model 1 is in line with rules,
evaluation is still possible. However focus has to be limited to the relation between model 1
and 2.
By classifying described effects as local ones, permissible stresses according to Rules of
LR are σperm = 120 N/mm² for bending stress and σperm = 150 N/mm² for equivalent stress.
Detailed stress evaluation with regard to permissible stresses was carried out at positions
shown in Fig. 7 for model 2. Equivalent positions behind #236 for model 1 have been used as
an appropriate reference. Calculated stress is normalized with permissible stress.
For hydrostatic pressure (LC1) model 2 fulfils the requirements although significantly
higher values can be observed compared to model 1. Results of LC2 and LC3 are just
comparable when introducing a load-dependent scaling factor found to be fscale = 6.2 for the
permissible stress. This approach is only valid for comparison purpose of course.
Nevertheless some statements regarding the evaluation of LC2 and LC3 are possible.
Stress increase at LC2 and LC3 comparing model 1 with model 2 is in the range of the
increase at LC1. Thus – in some degree - it can be drawn the conclusion that LC2 and LC3 do
not account for additional effects that could not be evaluated with LC1. This conclusion is
only valid if/because the changes in design from model 1 to model 2 are limited to a
comparably small area at the bulb tip. Even with this limitation it is not of general validity.
LC2 and LC3 might lead to additional effects when analysing bulbous bows of significantly
differing shape.

907
I. Tautz, M. Schöttelndreyer, J. Gauerke, E. Lehmann, W. Fricke

LC1: Hydrostatic Pressure


0 deformation scale
14 factor: 120

29
von Mises stress [N/mm²]

43
58
72

87 max. val.: max. val.:


101 90 N/mm² 128 N/mm²

116
model 1 model 2
130

LC2: Slamming
0 deformation scale
93 factor: 20

187
von Mises stress [N/mm²]

280
373

467
max. val.:
560 540 N/mm² max. val.:
838 N/mm²
653
747
model 1 model 2
840

LC3: Unilateral Pressure


0 deformation scale
79 factor: 20

158
von Mises stress [N/mm²]

237
316
394
max. val.:
473 525 N/mm² max. val.:
707 N/mm²
552
631
model 1 model 2
710
Figure 6: Stress contours for calculated variants and models, displayed in front of #234 only

908
I. Tautz, M. Schöttelndreyer, J. Gauerke, E. Lehmann, W. Fricke

1
LC1 LC2 LC3 Model 1 Model 2
fscale = 1 fscale = 6,2 fscale = 6,2
0.9

0.8
σ / fscale*σperm

2,3 1 4
0.7

0.6

0.5

0.4 Position: 1 2 3 4
σ: max. tension σ: max. v. Mises
σperm = 120 N/mm² σperm = 150 N/mm²
Figure 7: Detailed stress evaluation at certain points normalized to permissible values

4 IMPACT LOADS
Impact loads consider certain peak loads that could act on the bulbous bow e.g. caused by
contact to the anchor, its chain or floating objects. These kinds of loads are high but narrow
bounded contact loads that might lead to a certain plastic deformation. This might be
acceptable when load bearing reserves are considered in appropriate manner. Thus additional
FE-calculations have been carried out using models described in Chapter 3 with following
changes. Nonlinear material behaviour is considered with bilinear behaviour being ideal
plastic after reaching yield stress. ANSYS Shell Element Type 181 is used (reduced
integration, four nodes). Aft
boundary - fixed in all degrees of
freedom - is at #234. Thus the model
is four frames shorter than for all
other investigations.
Impact loads are considered as
area loads on four shell elements
resulting in nine loaded nodes shown
in Fig. 8. According area of impact is
almost equal to 160x160 mm².
Positioning of impact loads was
carried out aiming to cover almost all
possible draught conditions and
aiming to hit the bulb in the middle Figure 8: Impact load positions for ultimate load evaluation

909
I. Tautz, M. Schöttelndreyer, J. Gauerke, E. Lehmann, W. Fricke

of a stiffened panel (model 1) and at the bulb tip. In order to get a clear evaluation of results
calculations have been limited to a total number of six positions. Evaluation was carried out
by plotting applied load against displacement normal to the loaded areas. Reaching yield
stress at top- / bottom layer is depicted with a circle (Fel). Ultimate load is marked with a
rhomb and is defined in accordance to a beam with rectangular cross section as
F pl = 1,5 ⋅ Fel (6)

All results are compared with the load-displacement curve of a supported flat panel
1600mm x 620mm x 16,5mm as a reference. Chosen dimensions are typically in the area of
the bulbous bow. It is assumed that no larger ultimate load capacity is required than provided
by the reference field.
Results are shown for positions on the side of the bulbous bow in Fig. 9. For all positions,
ultimate load of model 1 is greater than provided by the reference panel. Model 2 leads to
significant decrease of ultimate load which is acceptable for positions 2s and 3s. Ultimate load
of model 2 at position 1s is about 25% less than the reference. Positions at the front are
significantly greater for all positions and both models. Thus position 1s at model 2 is the only
result where ultimate load decrease reaches a significant undercut compared to the reference.
According to this background this might be acceptable. The fact that position 1s is located
quite low with regard to possible impacts supports this conclusion.
300

250

200
Load [kN]

150

Position 3s, model 1


100 Position 3s, model 2

Position 2s, model 1


Reference: Supported panel Position 2s, model 2
50 1600x620x16,5
, Position 1s, model 1
F_pl
Position 1s, model 2
F_el
0
0 2 4 6 8 10 12 14
displacement [mm]
Figure 9: Load-displacement curves for positions shown in Fig. 8

10

910
I. Tautz, M. Schöttelndreyer, J. Gauerke, E. Lehmann, W. Fricke

5 CONCLUSION
An alternative bulbous bow design providing good crushing behavior in case of a ship
collision is presented. Dimensioning of the alternative design is a challenge because it is
formally separated from Classification Rules. Thus a procedure for dimensioning is proposed
that has been harmonized with comments of Lloyd’s Register. Three pressure load cases and
one impact load case are proposed for direct calculations. Results confirm the fact that
conventional design of the bulbous bow used as a reference is quite conservative. Thus
softening of parts of the structure as described in the paper is permissible in a certain manner.
It was shown that for hydrostatic pressure, acceptance of the softening was verified with
absolute values. For all other load cases comparative evaluation with the conventional design
was carried out with promising results.

5 ACKNOLEDGEMENT
The work presented in this paper was performed within the research Project ELKOS,
funded by German Federal Ministry of Economics and Technology (BMWi) under project no.
03SX284B. The authors are responsible for the content of this paper and wish to thank for
supporting this project. The authors’ gratitude is also addressed to German shipyard
Flensburger Schiffbau-Gesellschaft that provided the FE-model of the struck ship and
drawings of the bulbous bow.

REFERENCES
[1] Cheung L.: A soft bow for ships, European Shipbuilding No. 3, 52-53, 1969
[2] Endo H, Yamada Y. Performance of Buffer Bow Structure against Collision - On the
Effect in Preventing Oil Outflow. Marine Technology, Vol. 45, No.1, January 2008
[3] Kitamura 0.: Buffer Bow Design for the Improved Safety of Ships, IMO, MEPC45/INF.5,
Annex, 2000
[4] Peschmann J. 2001. Berechnung der Energieabsorption der Stahlstruktur von Schiffen bei
Kollision und Grundberührung. Dissertation, Hamburg University of Technology:
Hamburg.
[5] Schöttelndreyer, M.; Tautz, I.; Kubiczek J.M.; Fricke, W. & Lehmann, E. 2011. Influence
of bulbous bow structures on their collision behaviour. In C. Guedes Soares, W. Fricke,
Advances in Marine Structures: pp.381-390, London: CRC Press Taylor & Francis Group
[6] Tautz I, Schöttelndreyer M, Fricke W, Lehmann E. 2010. Experimental Investigations on
Collision Behaviour of Bow Structures. In: Ehlers S, Romanoff J. Proc. of 5th
International Conference on Collision and Grounding of Ships; June 14-16; Espoo,
Finland. 179-183
[7] Tautz I, Schöttelndreyer M, Fricke W, Lehmann E. 2013. Collision tests with rigid and
deformable bulbous bows driven against double hull side structures. To be published in:
Proc. of 6th International Conference on Collision and Grounding of Ships; June 17-19;
Trondheim, Norway.
[8] Yamada Y. 2006. Bulbous Buffer Bows: A Measure to Reduce Oil Spill in Tanker
Collisions. Dissertation, Technical University of Denmark: Lyngby (Denmark)

11

911

You might also like