Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
ULF KARLSSON
ULF KARLSSON, 2004
ISSN 1101-0614
CHA / NAV / R-04 / 0090
Department of Naval Architecture and Ocean Engineering
Chalmers University of Technology
SE-412 96 Gteborg
Sweden
Telephone +46 (0)31-772 1000
Abstract
Bow-doors are one of the most vulnerable parts on a RoRo-vessel. At the same time, they are
necessary for the profitability of todays RoRo-vessels.
The minimum design rules for bow-door systems, stated in the International Association of
Classification Societies (IACS) S8, constitute the basis for the class rules.
In this work, the rules of IACS S8 have been assessed regarding the structural integrity for
bow-doors of the clam door type.
The work is based on previous damages, compared to the design rules, so as to find
weaknesses in the rules. Damages were mapped and weak points in the rules as well as in the
design and construction stages were identified. For the found problem areas part analyses
were performed.
Fractures and cracks dominate as the cause of damage for bow-door systems. In most of these
cases cracks in welds are involved. Fatigue proved to be a major cause for the development of
fractures and cracks.
Assuming the bow-door as a rigid-body and distributing the forces equally on each support, as
prescribed by IACS S8, proved to be an approach that leads to inadequate results. The
deformation of the door causes the loads on the supports to differ significantly and the real
pressure distribution subjects the lower supports to much higher loads then the upper ones.
It is common with gaps between the supports in the doors and the corresponding ones in the
hull. These gaps will considerably affect the load distribution on the supports.
In certain situations, bow propellers may excite bow-door arms into resonance cycling. This
would give rise to a large number of high stress cycles and might even cause fatigue failure.
Keywords:
bow-door; IACS S8; damage; analysis; safety; arm; structure; ship design; fatigue.
Acknowledgments
Many people, companies and organisations have contributed with information and work in
this project, both people in the project group, reference group one [1], and from outside,
reference group two [2]. Great thanks for all of your support, information and advice. As you
all know, without you it would not have been possible to fulfil this project.
Especially, I would like to thank my supervisor Anders Ulfvarson for his knowledgeable
advice, guidance and always positive and good mood.
Erland Johnson and Gunnar Kjell at the Swedish National Testing Institute, SP, for their
measuring and analysis work of the bow-door arm.
Jan-Ove Carlsson and Pontus Dhalstrm at MacGREGOR RoRo-Division and Olle Thomsson
at Lloyds Register, for all the time and effort they put into this project.
I would also like to thank Vinnova and the Swedish Maritime Administration for funding this
project and Swedish Maritime Administration for their participation in the project.
To all at the Department of Naval Architecture and Ocean Engineering, it has really been fun;
even if the workload has sometimes been heavy, the good atmosphere makes one cope.
II
Table of Contents
Abstract .. I
Acknowledgements II
Table of contents ... III
1. Introduction ...
1.1 Background
1.2 Aim of the Project .
1.3 Limitations ...
1
1
1
2. Method ..
4
5
10
12
15
16
17
18
20
21
21
22
23
23
25
29
7.1 Introduction ..
7.2 Rules ..
7.3 Required Strength of the Supports according to IACS S8
7.4 Required Strength according to the Finite-Element Calculation .
7.5 Application of a Measured Pressure Event .
29
30
31
33
37
41
8.1 Introduction
8.2 Real Arm Verses the Model .
8.3 Load Cases ..
8.4 Bow Propellers
8.5 Fatigue Analysis ...
41
42
44
47
51
9. Conclusions
52
10. Recommendations
54
54
54
55
56
III
1. Introduction
1.1 Background
Accidents with many casualties due to failure of bow-door systems and several additional
incidents, are the cause for concern of the safety of bow-door systems.
The problem in these cases has been visors and bow-doors that, for some reason, have been
open or opened at sea.
This project is limited to bow-door systems of the clam door type, partly because no visors
have been constructed since the loss of M/S Estonia, at least not in Europe, and partly to limit
the size of the project in order to be able to conduct analysis with enough depth.
1.3 Limitations
The descriptions and information provided in this report are limited to RoRo-passenger ships
with bow openings of the bow-blade type, or clam doors, i.e., two doors that are either
attached to the hull via arms and open horizontally outwards, or two doors that are directly
attached to the hull with hinges and open down and outwards. It does not consider the inner
watertight door, the surveillance system or the crews management of the system.
2. Method
This licentiate thesis is a summery report based on four part-, or project reports. These are:
- Pre-Study of Bow-Door Systems CHA / NAV / R-04 / 0085
- Structural Damage Analysis of Bow-Doors CHA / NAV / R-04 / 0086
- Stress and Fatigue Analysis of a Bow-Door Arm CHA / NAV / R-04 / 0088
- Supports of Bow-Doors according to IACS S8 CHA / NAV / R-04 / 0089
The reports are all registered at the department of Naval Architecture and Ocean Engineering,
Chalmers University of Technology, Gteborg, Sweden.
The results from the analyses are based on verifiable facts. The research material, or the
chosen bow-door system, have a common design and are representative for most of the bowdoor designs of this type. All bow-door systems are, however, designed according to the same
basic rules, comprised within the IACS S8 rules.
The research method used consists of six steps, as described in diagram 2.1.
1. Pre-study of bow-door systems
2. Structural damage
analysis of bow-doors
5. Validation
6. Recommendations
Diagram 2.1. Schematic diagram of the used research method.
2
Step 1
The purpose of Step 1 is to gather information of different types of bow-door designs, the
design method and the design rules they are to comply too. It also describes some incidents
and damages that have previously occurred to bow-door systems. This information constitutes
the base for the continuation of this project.
The information was gathered through interviews with representatives of companies
designing bow-door systems, ships crews and other people knowledgeable in this area [1, 2],
the study of different types of bow-door systems, damage reports [3, 4], and of the IACS S8
rules [5], which the class rules are based on.
Step 1 consists of Chapter 3,Pre-Study of Bow-Door Systems.
Step 2
In this Step, we try to find structural weaknesses in these kinds of systems, based on previous
damages that have occurred. The result will give information of which parts that demand
further investigation.
The goal with the damage analysis has been to investigate what kind of structural damage that
has occurred on bow-door systems and surrounding hull structure, where the damages have
occurred and how common they are.
All relevant ships with bow-doors in a specific classification society are part of the analysis.
Step 2 consists of Chapter 4, Structural Damage Analysis of Bow-Door Systems.
Step 3
From Steps 1 and 2, items demanding further analysis were identified. Five items were found.
Step 4
Of the five items found in need of analysis, four have been analysed.
These are:
- Fractures and Cracks (Chapter 6)
- Load Distribution on Bow-Door Supports (Chapter 7)
- Gaps between the Plates of Supports (Section 7.4.3)
- Stress and Fatigue Analysis of a Bow-Door Arm (Chapter 8)
One item remains to be analysed:
- The rules for the hull structure around the doors
Step 5
In Step 5, the results from the part-analyses were validated with regard to the IACS S8 rules
and to design aspects.
Step 6
Finally, both rule-based and design-based recommendations are made regarding changes of
the active rules and regarding designing issues.
One of the most vulnerable areas of RoRo-vessels is the bow-door [2]. However, bow-doors
are essential to the profitability of this type of ship. The main reason is the large cost of
having a ship in port. The limiting factor regarding time-in-harbour is loading and unloading
from a RoRo deck. The best solution today for fast loading and unloading is to let the vehicles
drive through the ship. To make this possible, there needs to be a bow-door and an aft ramp.
Wide lanes through the bow are also important, as this makes it easier and faster to drive
trailers in and out from the ship.
The alternative to bow-doors is having very wide vessels in which vehicles can turn around.
This would work on ships trading on longer routes, where the ship stays in harbour for a
longer time and where loading and unloading is not the limiting time factor. If the vessel is to
go on a shorter route, the time-in-harbour is generally short, and loading/unloading becomes
the limiting factor.
Another reason to mount bow-doors on vessels, even if they are going to trade on a longer
route, is to make the vessels all-round, to be able to traffic more routes.
The second-hand value is also considered, as it is easier to sell a ferry with bow-doors than
without. The cost of mounting bow-doors on a ship afterwards, without the ship being
prepared for it, is about 50% higher than to install them during construction of the ship.
The customers wish of fast unloading is also an issue to consider.
Door
Arm
Hinges
Hinges
y
z
x
Symbols
Figure 3.4. Bow-door system with arms.
Cleating/Locking device
Support
Stopper
Guide
Icebreaking cylinder
Surveillance
camera
2
4
10
Ramp
Heaters
Photo 3.5. Stitch image of a bow-door system with arms (inside view).
Doors
The two doors are to have at least the same strength as the surrounding bow-hull, according to
IACS S8.4 [5], and the rules for the door structure are the same as for the foreship structure.
Manoeuvring arm
Each door is attached to an arm that carries the door during opening and closing and in an
open position. The arms have the shape of a box beam since they must carry heavy loads in
the form of a bending moment and torsion. They are attached to the doors at approximately
where the centre of gravity of the doors is.
The door is attached to the arm with two hinges, of which only one hinge carries vertical
loads, and the arm is attached to the hull, also with two hinges, of which only one hinge
carries vertical loads.
Hydraulic manoeuvring cylinder
The arm, and thereby the door, is manoeuvred by a hydraulic cylinder. Usually the hydraulic
cylinder is attached to the side of the hull and at the rear end of the arm.
Rubber seal
Along the edge of each door is a rubber seal. In the hull and in one of the doors is a coaming
in which the rubber seal is compressed at closing. The purpose is to keep the bow-doors
weather-tight. There are support edges on each side of the rubber seal to prevent it from
becoming too compressed, and to shelter it from water pressure and ice.
Supports
The doors open outwards and the forces from the sea are therefore in the closing direction.
These forces are transferred from the doors into the hull via supports on the doors and the
corresponding supports in the hull. Each door has 5-7 supports, positioned around the door
frames.
The supports transfer forces in different directions, depending on their position.
The supports mainly consist of thick steel plates welded to the doors and hull at structurally
strong points and strengthened with brackets etc.
Stoppers
Between the doors, there are stoppers with the purpose of transferring loads between the
doors. The stoppers consist of plates welded on structurally strong points on the doors.
Guide
The task of the guides is to guide the doors at opening and closing. There can be one or two
guides mounted in the front and upper part of the bow-hull and/or on the tank top. The guides
consist of guide rails in the hull and guide-wheels on the doors, which follow the rails and
guide the doors during opening and closing.
Icebreaking cylinders
There are one or two icebreaking cylinders for each door, positioned on the tank top and/or on
the front top of the doors. Their purpose is to help push out the doors, especially when they
are frozen stuck.
Locking for open position
In a fully open position, doors are locked by one hydraulically manoeuvred rod, or hook,
each. These are positioned on the side of the hull and lock through an eye-plate on the arm.
Manoeuvring sequence
Below is an example describing a typical manoeuvring sequence during opening and closing.
Opening the doors
1. Locking pins are manually pulled out from the cleating/locking devices;
2. Hydraulic pressure is turned on;
3. The cleating/locking devices between the doors and the hull are unhooked;
4. The cleating/locking devices between the doors are unhooked;
5. The icebreaking cylinders push the doors out;
6. The manoeuvring cylinders push on the arms and open the doors;
7. In a fully open position the doors are locked with the locking for open position;
8. The ramp is lowered;
9. The hydraulic pressure is turned off.
Closing the doors
1. Hydraulic pressure is turned on;
2. The ramp is raised;
3. The locking for open position unlocks;
4. The manoeuvring cylinders pull the arms and close the doors;
5. The cleating/locking devices between the doors cleat the doors towards each other,
compressing the rubber seal, and finally lock the doors in a cleated position;
6. The cleating/locking devices between the doors and the hull cleat the doors against the
hull, compressing the rubber seal, and finally lock the doors in a cleated position;
7. Hydraulic pressure is turned off;
8. Locking pins are manually pushed into the cleating/locking devices.
10
Some design companies always carry out a finite-element analysis of doors and of the arms
and their hinges; sometimes even of other items, for instance the fore bottom support.
Other design companies do not perform a finite-element analysis, except when they design the
steel structure for the bow-doors.
There are many design solutions to bow-door systems, among other things for how loads
transfer into the hull. This can differ greatly between ships.
In the designs, the ramp in a folded position usually constitutes the watertight bulkhead.
The steel material used is generally AH-36. DH-36 is chosen when thicker steel is to be used,
which is a softer and more ductile material and thus less fragile and gives a better margin for
brittle fracture.
Experience and continuous improvements is the method used for the development of bowdoor systems and elimination of critical points.
Responsibility
The design companies offer one years guarantee for the bow-door system.
The design companies have full responsibility for the bow-door systems they deliver, even if
the classification society in question has approved them.
Sometimes a second party, for instance shipyards in China or Korea, delivers the construction.
This splitting of the design/construction results in a less reliable construction and the question
of responsibility becomes unclear.
Feature bow-door systems
The opinion of the future bow-doors are somewhat divided. Some design companies believe
that the speed of Ro-Pax ships in the future will increase and that the bow shapes thereby will
become sharper. The radius, or the roundness, of the bow will thereby decrease.
They also think that the bow opening will be narrower with only one lane, due to the sharper
bow. Two lanes would make the bow-doors very long.
Others believe that future bow-doors will be wider and longer and made for two lanes through
the bow. They do not think that the speed of ships in the future will be higher than todays
high-speed crafts with a speed of up to 27-28 knots.
They agree about the fact that lightweight materials will not be used in the future, but that the
materials they use today will also be used in the future. They think that the bow-doors
attached directly to the hull will be less common in the future. Faster ships have larger stem
angels, or sharper noses, and these kinds of doors cannot be mounted on sharp noses, as they
would go into the water on opening.
11
12
CH = 0.0125 L
1
for L < 80 m
for L 80 m
flare angle, defined as the angle between a vertical line and the tangent to the side
shell plating, measured in a vertical plane normal to the horizontal tangent to the
shell plating at the point on the bow-door, /2 aft of the steam line on the plane h/2
above the bottom of the door, as shown in Figure 3.7.
entry angle, defined as the angle between a longitudinal line parallel to the centreline
and the tangent to the shell plating in a horizontal plane, measured at the same point
as .
13
Fx, Fy and Fz
acting on both doors,
0.7 Fx and 0.7 Fz acting on both doors and 0.7 Fy acting on each door separately,
The design external forces are applied at the centroid of the projected areas and the
reaction forces can be distributed uniformly on the effective securing and supporting
devices, irrespective of their position. In reality, the lower parts of the doors, and their
supports and securing devices, will be subjected to much larger forces than the upper
parts of the door.
3.4.3 Others
There are no special rules for the bow-hull structure depending on the size of the doors.
If the speed of ships increases in the future, the ships would become more slender. This would
require longer bow-doors to be able to have the same lane-width through the bow. The rules
for the hull structure around the doors might then be inadequate.
14
15
16
4.2 Method
Structural damages on the ships bow-door systems were searched for in the classification
societys database. The search period covered all of their lifetime.
Information on the ships was gathered in a table showing when the ship was built, how old it
was at the different damage occasions, which parts that had been damaged, what kind of
damages there were and where the damages had occurred.
The main structural parts in a bow-door system are:
1.
2.
3.
4.
5.
6.
Door structure
Bow-hull structure
Cleating/Locking devices
Hinges
Arms (if present)
(Supports)
Since it in the damage reports not has been possible to specifically distinguish damage on
supports, these probably appear as damages on doors or hulls; they are not included in the
analysis.
A summarizing of the damages for each of these main parts was made. This shows in what
kind of detail the damages had occurred, the number of damages, the type and location of the
damages and at which ship age they occurred.
17
35
61
30
7
4
Age
25
20
6
15 2 1
15
3
3
1
3 11
10
5
0
1
9 11 13 15 17 19 21 23 25 27 29 31 33 35 37
Ship
Undamaged
Ships
47%
(8.2 years)
Damaged
Ships
53%
(20.5 years)
47 damages
25
39
28
2
In Diagram 4.3 the percentage distribution of the damages is shown together with the average
age of the ships when the damages occurred.
ARM
1%
(15.0 years)
HINGE
20%
(19.6 years)
CLEATING/
LOCKING
28%
(13.9 years)
HULL
18%
(15.9 years)
The distribution of the 141 damages on the main parts shows that damages on doors are the
most common with 33% of the cases, closely followed by damages on the cleating/locking
devices with 28%. Hinges and hull have fewer damages, 20% and 18%, respectively. Arm
damages were only reported twice.
The average age of the ships, when the damages occurred on the different parts, varies
between 13.9-19.6 years.
19
Seal chanel
11%
(18.3 years)
Stiffener
33%
(22.7 years)
Frame
33%
(23.1 years)
Bracket
42%
(17.0 years)
Stiffner
50%
(8.8 years)
Bracket
11%
Shell plate (6.3 yeras)
11%
(21.0 years)
Pins/
Bearings
36%
(19.4 years)
Attachment/
Bracket
64%
(17.0 years)
Locking
pin/eye
37%
(17.8 years)
Attachment/
Bracket
63%
(14.9 years)
General for all main parts is that fractures and cracks dominate as the cause of damage.
Most common are fractures and cracks in stiffeners, brackets, attachments and in the door and
hull frame. Fractures and cracks were the cause in 70% of all damage cases and in most of
these cases, cracks in welds were involved.
Other found damages were; buckled, deformed, torn and worn.
For the door and the hull, most damages (80-90%) occurred in the aft parts. For the hull, there
were three times more damages in the upper parts than in the lower parts. For the doors the
damages between the upper parts and the lower parts were in equal numbers.
Damages in the cleating/locking devices occurred in 5 out of 6 occasions in the lower-placed
devices. Damages for the cleating/locking devices located at the centre were 4 times more
common than damages in the outer devices.
Hinges placed low had 2.6 times more damages then hinges placed high.
20
4.5 Arms
Only two cases of damage in arms were found. These were found on the same ship and on the
same occasion (See Section 4.6).
As mentioned before, it has not been possible to establish how many of the ships that have
their doors hinged to arms. An estimate is that approximately 40-45% of the ships in this
analysis have this arrangement. It would then mean that one out of 16 ships with arms have
suffered damages in an arm.
21
22
Number of Fractures/Cracks
14
12
10
0
1
11
13
15
17
19
21
23
25
27
29
31
60
Number of Fractures/Cracks
50
40
30
20
10
0
1
11
13
15
17
19
21
23
25
27
29
31
3
x38 = 14 damages
8
This is the relevant proportional value for the number of damages at age 25, considering all
ships to have reached this age.
As can be seen in Diagram 6.2, the number of damages has increased with the ships age and
are, especially for higher ship ages, much higher then in Diagram 6.1.
23
Summing up all damages for the first 10 years of age, for the age between 11-20 years and
between 21-30 years, the diagram will be as follows:
FRACTURES / CRACKS
Number of Fractures/Cracks
250
220.9
200
150
100
50
45.8
1 - 10
38.8
11 - 20
21 - 30
The first 10-year period shows somewhat more damages than the second, 11-20 year period.
One possible explanation could be initial faults in the construction, such as bad welds,
construction errors or even poor design. These kinds of damages will be discovered and
repaired in the ships early years. The number of damages would subsequently decline, as it
does in the next period between 11-20 years.
For the last 10-year period there are several times more damages found than for the earlier
two periods.
A filled line has been drawn through the columns in the diagram, illustrating how the
damages would occur.
If the cause of the fractures and cracks were due to random overloads, the distribution of the
damages would follow a straight increasing line, as the dotted line in the diagram shows, or at
least look something like that. The solid line, however, does not follow this line at all. Instead,
the line goes up in the first period, down in the second and increases thereafter rapidly in the
last period. This suggests fatigue to be a major cause for the development of fractures and
cracks.
It is of course difficult, from this analysis, to say how large a share fatigue has, or how large a
share overloads has, concerning the development of fractures and cracks. Most certainly both
fatigue and overloads are to blame, probably combined in many cases, but we can definitely
conclude that fatigue is a major cause of the development of fractures and cracks in bow-door
systems.
24
25
Opening/closing
During opening and closing of the doors, large stress ranges occur in hinges and arms, see
Chapter 8, Stress and Fatigue Analysis of a Bow-Door Arm. This causes low-cycle fatigue.
Since doors in general are opened and closed only 1/2 - 2 times a day, the number of these
cycles is low.
Loading/unloading
Loading and unloading will generally not give rise to fatigue in any of the main parts, see
Chapter 8.
Thrusters
When the thrusters are running, they often cause the whole bow of the ship to vibrate,
especially if the ship is moving at the same time at which the flow through the thruster tunnels
will be disturbed. In normal cases, this will only induce small stress ranges (Chapter 8), which
can cause high-cycle fatigue, but it will affect all parts and the number of these cycles is quite
high, between 1.5 15 million cycles a year depending on how often a ship enters a harbour.
The load cases described here will affect the main parts differently.
Table 6.4 shows how these load cases mainly can affect the main parts regarding high and
low-cycle fatigue. There will of course be exceptions, since all bow-door designs are
different.
Small stress ranges might give rise to high-cycle fatigue, marked High in the table.
Large stress ranges might give rise to low-cycle fatigue and are consequently marked Low
in the table.
Load cases
Number of
Cycles/Year
Hull
Waves
Large
High
High
Cleating/
Locking
High
Hinge
Arm
Flare slamming
Small
Low
Low
Low
Opening/closing
Small
Low
Low
Low
Loading/unloading
Large
Thruster
Large
High
High
High
High
High
Table 6.4.Type of fatigue the load cases mainly can cause in each main part.
26
Below, diagrams for damages due to fractures and cracks have been made for each main part
(except Arm) recalculated and summarised for each 10-year period as before.
Door and Hull
HULL
DOOR
45.6
50
140
120.2
Number of Damages
Number of Damages
120
100
80
60
40
20
45
11.9
12.6
1 - 10
11 - 20
40
35
30
25
20
15
10
11.9
5
0
7.6
21 - 30
1 - 10
11 - 20
21 - 30
As can be seen in table 6.4, doors and hulls can normally only be subjected to low-cycle
fatigue at flare slamming, for which the numbers of cycles per year are low. The load cases
loading/unloading and opening/closing will not affect the door and the hull, but the load
cases waves and thrusters can give rise to a large number of high-fatigue cycles.
In Diagrams 6.5 and 6.6, the first two 10-year periods have about the same number of
damages but during the last period, the number increases rapidly.
During the last period, between 20-30 years, the ships would have been subjected to more
than 100 million high-fatigue cycles, but only a few thousand low-fatigue cycles. If a few
thousand low-fatigue cycles had been the main reason for the development of the cracks,
these stress cycles have more or less to be overloads, as there are so few of them. The cracks
would in this case been distributed more evenly over time.
This indicates that high-cycle fatigue is the main cause for the development of cracks in doors
and hull. Nevertheless, low-cycle fatigue contributes and can be the main reason for the
cracks in the first two periods and for some of the cracks in the third period. The large
increase of cracks in the last period is, however, due to high-cycle fatigue.
As can be seen in Diagrams 6.5 and 6.6, doors have a much higher number of damages in the
last period compared to hull. A probable explanation for this is that hulls are stiffer than doors
and thus not subjected to as large stress ranges as doors are.
27
Cleating/Locking devices
CLEATING/LOCKING
20
18.6
18.6
Number of Damages
18
16
14
12
8.6
10
8
2.5
6
4
2
0
1 - 10
11 - 20
21 - 30
Cleating/locking devices can be subjected to lowcycle fatigue both from flare slamming, and from
opening/closing. The number of cycles they induce
is, however, low. Waves and thrusters will
induce a large number of high-fatigue cycles.
Loading and unloading will not affect
cleating/locking devices at all.
Of the 18.6 fracture/cracks (recalculated value) in
the first 10 year period (diagram 6.7), 17.6 were
found in four sister ships.
Three cracks were found in each ship, all at the same location, when the ships were between
7-9 years old. The cause of these cracks was obviously a design error. Removing these cases,
only one fracture/crack would remain in the first period. In the second period, we have 2.5
fractures/cracks and in the last period, 8.6 fractures/cracks. There would now be quite a linear
increase of cracks over time, suggesting that high-cycle fatigue not is a major cause of the
development of cracks in cleating/locking devices. The low number of fractures/cracks found
indicates that random overloads and possible low-cycle fatigue is the cause of the fractures
and cracks.
Hinge
HINGE
46.6
50
Number of Damages
45
40
35
30
25
20
15
10
6.6
7.0
5
0
1 - 10
11 -- 20
21 - 30
Arms
Arms will mostly be affected by the load cases: opening/closing and thrusters. Loading/
unloading, waves and slamming will not affect the arms very much, as long as they are
connected to the hull and to the door via a jointed coupling, which most of them are. Thrusters
will, in normal cases, induce high-cycle fatigue but can, as shown in Chapter 8, excite the arm
into resonance cycling, causing large stress ranges and thereby low-cycle fatigue. During each
opening and closing, the arms will be subjected to very large stress cycles, especially in the
attachments. The numbers of these are, however, few and they will generally not be able to
cause a fatigue failure during the ships lifetime.
28
29
7.2 Rules
For the first two steps, the rigid-body assumption and the direct calculation, the design
external pressure that is to be applied on the projected areas, are calculated according to IACS
S8.3 (see Section 3.4.2).
The arrangement and strength of supporting devices is to be designed with redundancy so that
in the event of failure of any single supporting device, the remaining devices are capable of
withstanding the reaction forces without exceeding by more than 20 per cent the permissible
stresses (S8.6.2).
The reaction forces applied on the effective supporting devices are to be determined for the
following combination of external loads acting simultaneously together with the self-weight
of the door (S8.6.2c):
Load case 1: Symmetric load case
The first load case is a symmetric load case, i.e., the load acts on both doors simultaneously.
Load case 2: Asymmetric load case
The second load case represents an asymmetric load case in which the doors cannot support
each other. The supports F1-F3 (see Figure 7.1), transferring loads between the doors, cannot
be considered here. In this load case, only 70% of the design load is to be applied.
The reaction forces from these load cases yield the minimum required strength of the
supports.
30
x, longitudinal direction
y, transverse direction
z, vertical direction
Figure 7.1. Supports with orthogonal constraints (dots) and degrees of freedom (arrows).
Constrained
Direction
x
y
z
A
X
B
X
D1
X
X
X
X
Support
E
F1
X
X
Total
F2
F3
G
X
X
X
2
6
4
The design external pressure to be applied on the projected areas, calculated according to
IACS S8.3, is 166.4 kN/m2.
The design loads to be applied on the projected areas, shown in Table 7.3, are calculated by
multiplying the design external pressure with the projected areas.
Projected area
Design load (kN)
x
y
z
5431 8379 7678
Direction
x
-y
-z
A
0
1397
0
B
4526
0
0
C
0
1397
0
D
0
1397
2284
D1
0
1397
2284
E
0
1397
2284
F1
0
1397
0
F2
0
1397
0
F3
0
1397
0
G
4526
1397
2284
Table 7.4. Required strength (kN) of the supports,
according to the rigid-body assumption.
In Figures, 7.6 and 7.11, the resulting minimum required strength of the supports is shown as
grey lines.
32
A
X
O
B
X
O
D1
X
O
X
X
X
X
Support
E
F1
X
X
Total
F2
F3
G
X
X
X
2
6
7
Table 7.5. Constrained degrees of freedom for the supports in the FE-calculations.
All the supports are modelled as point constraints. Point constraints result in high local
stresses and therefore deformations. Thus, the supports are softer than they would be in
reality. This error causes a more uniform distribution of forces over the supports and is
therefore favourable to the rules.
Vertical movement of support G
Especially the vertical direction of the top support on the centreline (in this work Support G)
is often not constrained in simulations, as the deck above the bow-door has a much lower
stiffness than the tank top. Other supports are then assumed to take the forces in a vertical
direction. However, the superstructure of the ship analysed in this work reaches over the
position of the bow-door, presumably making the area over the Support G a lot stronger than
on other ships. Therefore, constraining Support G in all three directions is appropriate.
33
The reaction forces obtained by the FE-calculations are compared to the required strength
according to the rigid-body assumption in Figure 7.6. The load cases describing failure of
Support G is specially marked in dark grey, since the stress for these cases is extreme.
For cases where the door is fully supported, the reaction forces for the symmetric load case
are shown as black lines and for the asymmetric load case with dotted black lines.
6000
5000
4000
3000
F [kN]
2000
1000
0
-1000
-2000
Required strength according to IACS S8
Range of reaction forces without failure in G
Range of reaction forces with failure in G
Symmetric reaction forces at fully supported door
Asymmetric reaction forces at fully supported door
-3000
-4000
-5000
D1
Support (x-direction)
Figure 7.6 a. Ranges of reaction forces [kN] of the FE-calculations in the x-direction, taking
all failure cases into account.
7000
6000
5000
4000
3000
F [kN]
2000
1000
0
-1000
-2000
-3000
-4000
-5000
D1
F1
F2
F3
Support (y-direction)
Figure 7.6 b. Ranges of reaction forces [kN] of the FE-calculations in the y-direction, taking
all failure cases into account.
34
14000
12000
10000
8000
F [kN]
6000
4000
2000
0
-2000
-4000
-6000
A
D1
Support (z-direction)
Figure 7.6 c. Ranges of reaction forces [kN] of the FE-calculations in the z-direction, taking
all failure cases into account.
Positive reaction forces imply pressure on the support while negative values imply tension.
The supports themselves cannot take tension, as they are realized as plate bearings. Therefore,
locking devices become necessary where negative forces occur. An exception to this is the zdirection of the supports A, B and C, situated at the bottom of the door. Due to their
geometry, they can transfer both positive and negative forces in the z-direction to a certain
extent.
It can clearly be seen that the distribution of forces over the supports is a long way from being
uniform. The forces do not only vary in wide ranges, they also have different signs. The cause
for this is the doors deformation. Even with the door fully supported, negative forces can be
found in the x-direction for Support G for the asymmetric load case and in the y-direction for
Support D for the symmetric load case.
In the rigid-body approach, only the active supporting devices having an effective stiffness in
the relevant orthogonal coordinate direction are to be included when calculating the reaction
forces acting on the devices. The supports that are not to be included, but in reality transfer
loads, are in this approach not dimensioned for it.
The loads when omitting failure of Support G
Special notice should be taken to Supports D and D1:
- They are assumed not to take any forces in the x-direction, since their direction of
possible displacement is mainly in the x-direction. However, the calculation reveals
that they take a great share of the forces in the x-direction. If, on the other hand, they
were to be taken into account for the simple calculation, this would lead to a required
strength of 1509kN according to IACS S8. The maximum load on support D is almost
twice that high.
- They are loaded with large negative forces in several of the redundancy cases. This
means that locking devices are necessary to prevent the door from moving outboard.
Even Support E has to stand some tension in the x-, and y-direction; this also calls for locking
devices.
35
It has to be checked, whether or not the strength of the locking devices according to the rules
is high enough to stand this tension.
The result shows that assuming the bow-door as a rigid-body and distributing the forces
equally between the supports, as prescribed by the IACS S8, is an approach that leads to
inadequate results.
Failure of Support G
Even though the door formally fulfils the redundancy criteria, because of its geometry, a
failure of Support G could cause massive overloads on the other supports. With the given
positions of the supports, especially Supports D and D1, but also Support B would have to be
equipped with massive securing devices in order to prevent the door from moving away at
these points.
A
B
C
D
D1
E
F1 F2 F3
G
Loadcase 1 7.7 1.0 11.8 9.2 12.8 9.7 1.8 3.5 7.0 32.2
Loadcase 2 4.3 3.5 12.4 5.9 11.0 6.3
279.5
Table 7.7. Displacement [mm] at removed supports in redundancy calculations.
The values are very low, compared to the average accuracy of welded structures of this size.
If the gap between the two plates of a support is higher than the value in the table, the support
is not able to take any forces, since the two plates can never make contact. The force then has
to be taken by other supports, of which some then might be stressed up to their redundancy
strength. If more than one support never makes contact, other supports can be overloaded.
According to the rules, the gap between supports is not to exceed 3mm. In reality, this is
difficult to achieve and even larger gaps than the above are common on ships [1].
These gaps are, most certainly, what affect the load distribution on the supports to the greatest
degree.
As all the supports in the FE-model are built with no gaps, and therefore assure a good
distribution of the forces, the maximum forces on the supports on the real door are likely to be
even higher than the simulated ones.
The hulls flexibility has not been modelled. This flexibility could reduce some of the
maximum forces.
36
The signals used from the pressure transducers were 10 Hz low-pass filtered pressure signals.
These pressure signals are, for our purposes, considered representative as pressure on the
bow-door since they reduce the local high-pressure peaks, which are not relevant to the larger
structural members. This is because the energy is small in pressure peaks of high frequency.
This energy will be absorbed, or damped out, locally in the doors plates and will thus not
affect the supports.
The measured average load is about the same as the design load, somewhat higher in the ydirection and a bit lower in the x-, and z-direction (see Table 7.9).
Measured Load
Design Load Case 1
Measured Load per Design Load
Direction
x
-y
-z
4451 kN 8684 kN 7232 kN
5431 kN 8378 kN 7678 kN
82 %
104 %
94 %
Table 7.9. Total load according to measurement compared with the design load.
37
For the event used in the calculations, the measured pressure signals are as in Table 7.10.
The highest pressure found is at the lower parts of the door, especially in the middle and aft
areas, P1-P3 and P6. The upper parts are not exposed to any pressure at all.
For this bow-door, a design pressure of 166.4 kN is to be applied according to IACS S8. In
areas P1-P3 and P6, this pressure is exceeded, in area P2 by more then 3 times.
P9 =
0
P6 = 227.7
P3 = 263.6
P8 =
0
P5 = 95.9
P2 = 524.9
P7 =
0
P4 = 95.1
P1 = 180.0
The reaction forces obtained are in Figure 7.11 compared to the required strength according to
the prescribed rigid-body assumption and the proposed FE-calculation, including failure of
Support G.
6000
5000
4000
3000
F [kN]
2000
1000
0
-1000
-2000
-3000
-4000
-5000
D1
Support (x-direction)
Figure 7.11 a. Reaction forces [kN] from the real pressure event, compared to the required
strength according to IACS S8 and to the FE-calculation in the x-direction.
38
7000
6000
5000
4000
3000
F [kN]
2000
1000
0
-1000
-3000
-4000
D1
F1
F2
F3
Support (y-direction)
Figure 7.11 b. Reaction forces [kN] from the real pressure event, compared to the required
strength according to IACS S8 and to the FE-calculation in the y-direction.
14000
Required strength according to IACS S8
Required strength according to FEM
Symmetric reaction forces at fully supported door
Asymmetric reaction forces at fully supported door
Reaction forces to symmetric pressure event
Reaction forces to asymmetric pressure event
12000
10000
8000
F [kN]
6000
4000
2000
0
-2000
-4000
-6000
D1
Support (z-direction)
Figure 7.11 c. Reaction forces [kN] from the real pressure event, compared to the required
strength according to IACS S8 and to the FE-calculation in the z-direction.
For the FE-calculations with the door fully supported, the required strength - according to the
FE-calculations - is shown with black solid lines in Figure 7.11 for the symmetric load case
and with dotted black lines for the asymmetric load case. As the measured pressure was
applied to the model while fully supported, the symmetric and asymmetric reaction forces
should not exceed the respective lines.
39
As can be seen, the reaction forces for many of the supports exceed the required strength
according to the FE-calculations. For some of the supports, the loads are even higher than the
required redundancy strength according to the FE-calculations. In some cases, the reaction
force is negative, i.e., the supports will not take any forces at all and, instead, the locking
devices are called for.
The result shows that even a direct simulation, as proposed in IACS S8, with the design
pressure applied uniformly to the door model, cannot ensure the bow-door from becoming
overloaded in real sea conditions. The pressure distribution has also to be taken into account.
40
41
2500
1510
Door
attachment
Eye-plates that carry
the vertical loads
Hull
attachment
The second upper eye-plate carries the vertical load from the door. Two plates under each side
of the eye-plate support this eye-plate. All the four eye-plates carry the horizontal load.
At the attachment arm/hull, the second lowest eye-plate carries the vertical load. On top, the
eye-plate is supported by two plates on each side. All the four eye-plates carry the horizontal
load here also.
The plating is 8 mm thick everywhere in the arm, except for the end plate at the attachment
arm/hull where the thickness is 10 mm and at both sides of the same attachment, where 15
mm plates are inlaid. The eye-plates have a thickness of 40-50 mm at the arm/hull attachment
and 30 mm at the arm/door attachment.
The material in the arm is A36 and has the following properties:
Yield stress, y = 355 MPa
Ultimate stress, max = 510 MPa
The arms weight is 3.9 tons, including attachment devices, such as hinges etc.
The mass of the door is 26.5 tons.
To stiffen the arm there are transverse
plates and stiffeners inside the arm,
see figure 8.3.
The model is made according to
drawings. It is, however, modified
with measurements taken at a control
measuring, as it showed that the
measurements of the real arms and
the drawings differed in some areas.
43
At sea
Loading/unloading
Opening/closing
Bow propeller
(door closed)
(door open)
(in between)
(door closed)
8.3.1 At sea
Earlier measurements [6] made on the ship showed that the acceleration on the foreship was
as its highest +4.5 m/s2 and -5 m/s2 at slamming in storm. This corresponds to a gravity of
+1.46 g and +0.49 g. The weather in the area can be even rougher than it was during this
measuring period, but the ships speed and course adjusts after the waves, and the foreship is
normally not subjected to higher accelerations than this.
In a closed position, the door rests on the bottom supports, and will not be able to affect the
arms downwards. Since there is no negative gravity, the door will not affect the arms upwards
either.
The frequencies of the accelerations at sea are low and one can assume that if the arm were
loaded with the corresponding g-forces, it would give the stress- increase/decrease in the arm
due to the accelerations. The stress difference between these two calculations should then
correspond to the highest stress range the arm is subjected to at sea.
FE-calculations have been made with the arm-model in a closed position, loaded with 1.46 g
and 0.49 g, respectively. This gave a maximum stress range of 0.41 MPa at the measuring
point.
The stress range is obviously small and, consequently, the ships motion at sea cannot cause
fatigue in the arms.
44
8.3.2 Loading/Unloading
During loading and unloading the hull is subjected to accelerations when vehicles drive
onboard or off the ship. The accelerations translate from the hull via the arms to the doors and
give rise to stress changes in the form of stress cycles with different stress amplitudes, i.e.,
stress ranges.
The stress ranges vary due to, among other things, the vehicles weight and speed. Vehicles
with hard rubber wheels give rise to higher accelerations, and thereby higher stress ranges,
than normal vehicles with air rubber wheels. Loading of a vehicle gives a somewhat higher
acceleration than unloading of the same.
[s]
(driving onboard)
(driving off)
(driving off)
(driving onboard)
(driving onboard)
(driving off)
45
Closing
During the closing event of the door, strain measurements carried out on the arm showed that
the stress at the measuring point decreased by about 34 MPa, from 55.6 MPa (calculated
value) at the fully open position to 21.6 MPa at the fully closed position. This means that 21.6
MPa is locked into the arm, at the measuring point, in the closed position.
The stress in the arm is constant during manoeuvring of the door from the fully open position
until it reaches the hull and the supports (Figure 8.6). Here the door starts to be drawn up on
the bottom supports at which the stress increases. The reason is that only the bottom part of
the door is in contact with the supports in the hull, which results in a twisting moment in the
arm. The arm starts to get unloaded, and the stress starts to decrease, when the hooks begin to
cleat the door into the fully closed position.
The largest stress occurs just before the hooks starts to cleat the door.
Opening
The stress increases in the arm when the hooks open. When the door is pushed out and off
from the bottom supports, the stress first increases rapidly and thereafter starts to decrease
until it hangs freely where some large stress cycles occur (Figure 8.7). During the rest of the
opening event, to the fully open position, the stress is relatively constant.
The door starts to
be pushed out.
At both the closing and the opening event, the largest half-stress range is 50 MPa. During
the measured events, there were a number of cycles with stress ranges in between 1-15 MPa
and some large cycles with stress ranges of 15-31 MPa.
Some of the stress ranges are quite large, but the number of these is small and during the
ships lifetime, the total number of these stress cycles will be relatively small.
46
47
Diagram 8.8. Stress at the measuring point while the bow propellers are running.
Diagram 8.8 shows von Mises stress at the measuring point in the arm when the bow
propellers are running and the ship leaves the quay.
As can be seen in Diagram 8.8, there are two occasions when the stress ranges are high,
between 300-370 seconds and 470-540 seconds. Apart from the large stress ranges of 9-18
MPa at the time 330 seconds, the stress ranges vary between 2-7 MPa at the measuring point,
with many cycles around 5 MPa.
Except for these two occasions, the measured stress is low, about 0.3 MPa.
[S]
Diagram 8.9. Acceleration in the hull while the bow propellers are running.
If we compare the stress in Diagram 8.8 to Diagram 8.9, showing the measured acceleration
in the hull during the same period, we find the acceleration in the hull to be only marginally
higher between 300-540 seconds. No acceleration peaks can be found at the same time as
during the two periods of high stresses found in Diagram 8.8. This indicates that the arm
could be subjected to resonance cycling.
48
In the Power Spectral Densities (PSD) diagram [7], for instance Diagram 8.10 showing the
power density as a function of the frequency of the arm for vertical motions, we find power
tops at the frequencies 6.8/7.5, 15.1 and 30.1 Hz. The frequencies at 7.5, 15.1 and 30.1 Hz are
multiples, which is typical for a forced vibration caused by a rotating device with some
unbalance. This unbalance is common for all propellers, which all have it in different degrees.
6.8/7.5 Hz
15.1 Hz
30.1 Hz
Diagram 8.10. Power Spectra Density (PSD) diagram of the arm for vertical motions.
A frequency analysis of different parts of the time history shows that in the parts with high
levels of strain (stress), the 30.1 Hz component is more dominating than in the time histories
with low levels. When the phase relationship between the two bow propellers is critical, the
arm vibrates at 30.1 Hz and at a rather high level, probably caused by resonance in the arm. If
the phase relationship is not critical, the motion is at a lower level but more complex and built
up by forced vibrations consisting of several frequency components [10].
The arms probably have a natural frequency at around 30 Hz. The two four-bladed,
unsynchronized bow propellers generate this frequency when they are running with a critical
phase relationship, causing resonance cycling in the arm.
49
The measured sequence in Diagram 8.8 is recorded when the ship leaves the quay. The ship
first started to back out, speed 5 knots with the bow propellers running, but with the blades
idle, i.e., without pitch. While backing, they started to turn the ship around with the bow
propellers (maximum pitch), during which large vibrations appeared in the ship until the ship
stood still, turning. While standing still and turning, there were no large vibrations. When they
had nearly turned around to the new course (they turned around nearly 180), they started to
go forward, during which large vibrations once more appeared until the bow propeller blades
were idle again. The forward speed was then 5 knots.
One theory is that when using the bow propellers with a large pitch on the blades, i.e., while
taking out effect from the bow propellers and at the same time moving forward or backwards,
the water flow through the bow propeller tunnels will be disturbed. This will cause the flow to
the bow propeller blades to differ, which causes unbalance in the propellers and large
vibrations as an effect. The disturbance frequency could then be: 3.8 x 2 x 4 = 30.1 Hz. (3.8
rotations per second for one propeller, 2 bow propellers, and 4 blades), corresponding to one
of the natural frequencies for the arm.
Disturbance
Figure 8.11. Flow through a bow propeller tunnel while going forward and turning with the
bow propeller.
Summary
The highest measured stress ranges for the different load cases are:
0.41 MPa,
3.2
7
50
The stress ranges for the load cases at sea and loading/unloading are very small and
probably cause no fatigue. Even for the case bow propellers, the stress ranges are normally
small, except when they excite the arm into resonance cycling, which causes large stress
cycles. At opening and closing, some stress ranges are very large, but the number of these
cycles are few.
Except for the case where the bow propellers excite the arm into resonance cycling, no other
load case is able to cause resonance in the arm.
50
51
9. Conclusions
Structural Damage Analysis of Bow-Doors
It is common with damages on bow-door systems, as about half of the ships in the damage
analysis have had reports of structural damages. The doors and the cleating/locking devices
had about 30% each of the found damages and the hinges and the hulls, about 20% each. The
arm had two reported damages.
General for all main parts is that fractures and cracks dominate as the cause of damage. In
most of these cases cracks in welds were involved.
Many of the cracks had developed due to poor detail design and construction, hard spots and
bad welding.
Damages on the doors occurred mostly in the aft parts. Between the upper and the lower parts
of the doors, no differences where found. For the hull, most damages occurred in the aft and
upper parts. Damages to the cleating/locking devices occurred mostly in the lower and centrepositioned devices. Hinges placed low had more damages then hinges placed high.
Most of the damage occasions in this analysis were of minor severity and represented no
danger for the ship, the passenger or the crew. However, on two occasions and for two
vessels, the damages led to severe incidents. Two, out of 38 ships, with an average age of 14.7
years, is a high proportion.
Fractures and Cracks
The distribution of fractures and cracks over the ships age showed for the last 10-year period,
between 20-30 years of age, a dramatic increase in damages. The damages were several times
larger than for the two previous periods, indicating fatigue to be a major cause for the
development of fractures and cracks.
In general, high-cycle fatigue appears to be the main cause of the development of cracks in
doors and hull.
For cleating/locking devices, high-cycle fatigue seems not to be the cause for the development
of cracks. Instead, overloads and possibly low-cycle fatigue, appear to be the reason.
For hinges, low-cycle fatigue due to the opening and closing event seems to be the main cause
for the development of cracks.
Arms seem rarely subjected to fractures and cracks, as there were only two fracture/cracks
found in one arm. In this case, the arm was rigidly attached to the door, which is a rare design
solution. However, arms can be excited to resonance cycling by the thrusters. This would
cause large stress cycles, which can result in a fatigue failure.
52
53
10. Recommendations
10.1 Rule-Based Recommendations
Load Distributions on Bow-Door Supports
A new rule concerning the design load is called for. It has to consider the deformation of the
door and the hull and to satisfy the fact that the pressure on a bow-door is not distributed
uniformly.
To achieve this, one can demand a direct simulation of the doors.
If the rigid-body assumption is to be used even in the future, the design pressure for
dimensioning of the supports has to be raised. It has to be large enough to cope with the
unknown load distribution, due to deformation of the door and the hull. This new design
pressure should, however, not affect the stiffness of the door, as it does today. To raise the
stiffness of the door would result in an even stiffer construction, with an even worse load
distribution.
The design pressure must also be distributed in a way where it takes into account the peaks at
the bottom and aft parts of the doors.
Stress and Fatigue Analysis of a Bow-Door Arm
It should not be allowed to include the arms in the supporting or locking arrangement, unless
a resonance and fatigue analysis has been carried out, insuring the arms structural integrity.
54
55
12. References
[1] Reference group 1
Hkan Torstensson, Professor
Hgskolan i Bors
School of Engineering
Erland Johnson, Ph.D.
Manager R&D
Swedish National Testing and Research Institute (SP), Bors
Building Technology and Mechanics
Gunnar Kjell, Ph.D.
Swedish National Testing and Research Institute (SP), Bors
Building Technology and Mechanics
Jan-Ove Carlsson,
General Manager
Conversion Department
MacGREGOR, RoRo Ship Division, Gothenburg
Pontus Dahlstrm, M.Sc. Naval Architect
Structural Analysis Manager
MacGREGOR, RoRo Ship Division, Gothenburg
Mikael Jogvall, M.Sc. Naval Architect
Steel Structural Engineer
TTS Ships Equipment AB, Gothenburg
Per Nordstrm, M.Sc. Naval Architect
Deputy Director of Marine Safety
Swedish Maritime Administration
Maritime Safety Inspectorate, Norrkping
Olle Thomsson, M.Sc. Tekn. Lic, Naval Architect
Lloyds Register, Gothenburg
Anders Ulfvarson, Professor
Department of Naval Architecture and Ocean Engineering
Chalmers University of Technology
56
57
[3]
[4]
The Joint Accident Investigation Commission of Estonia, Finland and Sweden (1997),
Final report on the capsizing on 28 September 1994 in the Baltic Sea of the RoRo
Passenger Vessel MV ESTONIA, 1997.
[5]
[6]
[7]
Kjell G. Johnson E. Karlsson U. Strain and Acceleration Measurements of a BowDoor Arm, Department of Naval Architecture and Ocean Engineering, Chalmers
University of Technology, R-04 / 0087, Gteborg, Sweden, 2003.
[8]
[9]
58