Sei sulla pagina 1di 40

SNAME Transactions, Vol. 106, 1998, pp.

1-40

The Strength and Reliability of Bulk Carrier Structures


Subject to Age and Accidental Flooding
Jeom Kee Paik, Member, Pusan National University, Pusan, and Anil K. Thayamballi, Life Member,
American Bureau of Shipping, New York

" A Capesize bulk carrier operating in the laden condition"

ABSTRACT

This paper is a summary of recent research and development in areas related to the strength,
structural safety and reliability of bulk carriers, undertaken by the American Bureau of Shipping
and the Pusan National University. The paper focuses on the accidental flooding and age related
strength degradation of" bulk carrier structures. The hull girder and local load redistribution
characteristics of bulk carriers subjected to accidental flooding are discussed. The ultimate
strength characteristics of corrugated bulkheads in a flooding situation are studied experimen-
tally, theoretically and numerically. Corrosion rate estimation models for bulk carrier longitudi-
nal strength members and transverse bulkheads are introduced. The collapse strength reliability
of the bulk carrier hull girder considering corrosion effects is evaluated. A study of the fatigue
performance of bulk carrier side frame structures is performed Recommendations are made with
respect to bulk carrier design and also needed future research.
1. INTRODUCTION ]Side Shell Failure ]l Hatch Cover Failure ] ] Va~gneCrack
I I

In the last decade, there have been several bulk car- '
Water Ingress I l
Partial Loss of
rier casualties including total losses. Over a thousand sea- Structure
Hold Vlooding I
farers have lost their lives in the related incidents. Some
of those vessels were lost under circumstances where the Transverse Bulkhead Increase of Hull Decrease of Hull
crew apparently had no time for calling a signal of dis- Failure Girder Loads Girder Strength
tress, meaning that such losses occurred quite suddenly
(Brown 1997). The possible causes of such losses can be 7 tu

categorized into the three groups, namely (a) loss of re-


serve buoyancy (or floating capability), (b) hull girder I Stab
breakage and (c) loss of stability, initiated perhaps totally
or partly by unintended water ingress into cargo holds,
see Fig. 1. Fig. 1 Bulk carrier loss scenarios
In a large proportion of the bulk carrier incidents re-
ported, the vessels were carrying iron ore or coal, the
former being one of the denser types of cargo, and the University. This research was initiated due to concerns
latter being one of the more corrosive. It has been indi- with the bulk carrier losses during the last decade.
cated that the age of most vessels concerned was over 15 In this paper, the nature of the bulk trade is briefly
years, and so significant defects related to corrosion and reviewed, with reference to bulk carrier size and types of
fatigue may have been present. Decrease of the residual cargo carried. A review of bulk carrier losses is made.
strength and / or increase of the applied hull girder loads The structural characteristics of conventional bulk carri-
in a flooding event may possibly lead to hull girder ers are investigated. Unintended water ingress into the
breakage (Paik 1994). Related to this, there is the possi- forward hold of bulk carriers has been stated to be a pos-
bility that part of the side shell forward could have been sible factor in bulk carrier losses. The degree of increase
lost due to a combination of circumstances, e.g., exces- in hull girder and local loads due to such water ingress is
sive corrosion and fatigue cracking damages together illustrated.
with perhaps the shifting of solid cargo due to roll in As previously noted, progressive flooding subsequent
rough weather. This would cause ingress of sea water into to the collapse of the corrugated transverse bulkhead in a
the cargo holds. Water ingress into the forward hold flooded forward cargo hold is thought to be a factor in
could occur also through failed hatch covers, as other some bulk carrier losses. Hence the ultimate strength
investigators have noted (Faulkner et al. 1996, Byrne & characteristics of corrugated bulkheads are studied using
Evans 1998). actual tests on mild steel corrugated bulkhead models. A
In such postulated accident scenarios as indicated in closed-form design formulation for predicting the ulti-
Fig. 1, even if the vessel could initially survive with one mate strength of corrugated bulkheads under pressure
of the compartments flooded, the ingress of sea water into loads is proposed, and the related modeling error is quan-
a cargo hold could amplify the applied loads, and also tified. A more refined prediction technique for strength
otherwise lead to vessel loss by progressive flooding through a special purpose nonlinear finite element
(particularly when the watertight transverse bulkheads are method is also developed.
insufficient to withstand the increased static and dynamic Corrosion and fatigue are the important factors af-
pressures). The relevant flooded load components may be fecting structural strength of bulk carriers over time. This
increased by pitch motion, whose severity can depend on paper reviews the corrosion protection of conventional
the cargo density and other characteristics of the vessel. bulk carriers and also presents pertinent results from re-
Progressive flooding after the collapse of corrugated cent investigations to establish probabilistic corrosion rate
bulkheads in a flooded condition is thought to be impli- estimation models for bulk carrier longitudinal structure
cated in some of the bulk carrier losses (RINA 1996, Fry- and transverse bulkheads using gauging data. By applying
stock & Spencer 1996). Also, with progressive hold these corrosion rate models together with the strength
flooding, the vessel in some cases could loose stability in formulae for transverse bulkheads previously noted, the
rough seas, potentially leading to capsizing (Turan & ultimate hull girder strength reliability of bulk carrier
Vassalos 1994). structures as the vessel ages is investigated. The fatigue
This paper is a summary of recent research and de- performance of bulk carrier side frame structures is then
velopment in specific areas related to the strength, struc- studied, as the related side shell failure is a potential
tural safety and reliability of bulk carriers, undertaken by mode of water ingress into the forward hold, and the loss
the American Bureau of Shipping and the Pusan National of the side shell is a factor that can significantly affect the

2 Strength of Bulk Carrier Structures


500 Panamax bulk carriers in addition also carry significant
450 [ Capesize bulk carrier ] quantities of grain and other cargo, see the bulk trade
400 , O t h e r _ share estimates in Fig. 2 (Clarkson Research 1996). In
fact, relatively more types of cargo now mow,~ in
35o
Panamaxes when compared to Capesizes.
300

250
2.2 Bulk Carrier Losses and Potential Causes
200 - Regarding bulk carrier losses at sea, Table 1 (HMSO
150 - 1995, LR 1997) is particularly telling. The table indicates
Ore
r~ 100 -
that over the period 1980 through 1996, there were 186
50-
total losses of bulk and combination carriers, with 1,278
lives presumed lost.
0 I I I I I I I I I I I I
To calculate the relative loss rates for various catego-
ries of bulk carriers, we used data from Clarkson Re-
search (1996) related to losses by deadweight, and also
Fig. 2.a Capesize bulk trade share estimates
the bulk carrier fleet breakdown by deadweight. The re-
sults, shown in Table 2, indicate that over the past 6
350 years, the Capesize loss rate has been typically greater
than the Panamax loss rate. One can argue about details
300 -
bul c i r, of these calculations, such as definition of loss, lack of
consideration of the age of the vessels at risk, and sensi-
250 -
tivity of the loss rates to the (small) number of losses in-
200 - volved, but it appears that a Capesize bulk carrier may be
more at risk than a Panamax. The table also shows the
150 -
average age of the vessels lost in each of the last six
100 -
years, indicating that an older vessel is at greater risk.
Another interesting factor in bulk carrier losses is the
50- type of cargo carried. Consider these mixture of news
Ore
items compiled from several Lloyd's List articles of 1995:
0 I I I I I I I I I I I I

Table 1 Losses of bulk / combination carriers and lives


Fig. 2.b Panamax bulk trade share estimates during 1980-1996

Year Vessels Lives


residual strength of the hull girder. 1980 19 162
Conclusions regarding the relative importance of 1981 l0 152
various factors affecting bulk carrier strength in a flooded 1982 2 31
condition are then drawn. Future study needs are noted, 1983 12 J"~
D

including potential flooding through failure of hatch cov- 1984 9 32


ers (a subject that is not considered in this paper). From 1985 12 27
the various studies, related implication,; for the bulk car-
1986 8 4
rier structural design process in general are emphasized.
1987 15 101
1988 4 39
2. NATURE OF BULK CARRIERS
1989 10 61
2.1 The Bulk Trade 1990 12 125
Today's bulk carrier fleet consists of about 5,000 ves- 1991 18 152
sels worldwide. Of these, on a deadweight tonnage 1992 5 28
(DWT) basis, nearly 50% are Handysize (10,000-35,000 1993 4 74
DWT), 20% are Handymax (35,000-50,000 DWT), 20% 1994 9 123
are Panamax (50,000-80,000 DWT), and 10% are 1995 19 84
Capesize (80,000 DWT and over). Iron ore and coal are 1996 18 50
the predominant types of cargo in Capesize bulk carriers. Total 186 1,278

Strength of Bulk Carrier Structures


Table 2 Bulk carrier loss rates and age at loss

Vessel Year 1990 1991 1992 1993 1994 1995


Lost 9 8 8 6 7 4
Handysize At risk 2,613 2,591 2,566 2,534 2,521 2,563
% Loss 0.34 0.31 0.31 0.24 0.28 0.16
Av. Age at loss 18 19 17 18 22 15
Lost 3 0 0 0 1 1
Handymax At risk 942 971 999 1,001 1063 1,147
% Loss 0.32 0 0 0 0.094 0.087
Av. Age at loss 16 0 0 0 13 10
Lost 2 4 1 1 0 0
Panamax At risk 783 796 790 815 862 910
% Loss 0.26 0.50 0.13 0.12 0 0
Av. age at loss 22 19 24 25 0 0
Lost 3 4 3 0 3 1
Capesize At risk 387 397 392 409 427 478
% Loss 0.78 0.10 0.77 0 0.70 0.21
Av. age at loss 19 15 21 0 23 9

"Industry is shocked out o f complacency. The Marika was Actual causes of a bulk carrier loss at sea in rough
lost with all 36 crew members and iron ore cargo in seas, where the vessel is typically not recovered, are diffi-
heavy weather off Newfoundland within days" of the start cult to pinpoint. In an IMO study which listed known or
of 1994... The 1ton Antonis was lost with her full 24-man conjectured causes of bulk carrier losses with fatalities
crew between Tristan daCunha and St. Helena while car- during 1990 through 1994, the general factors implicated
rying a cargo o f iron ore from Tubarao in Brazil to were:
China... The Panamanian flag Apollo Sea sank in heavy
weather with her 36-man crew shortly after leaving 9 Heavy weather
Saldanha Bay, South Africa, with a cargo of iron ore... 9 Water ingress
The KamarL also with a cargo of iron ore, was sunk pur- 9 Foundering or sinking
posely 75 miles off Rio Grande after sustaining severe 9 Hull, side shell or tank crack
structural damage in Atlantic storms... The bulk carrier 9 Vessel age
Christinaki sank during Atlantic storms after a hatch 9 Open or damaged forward hatch
cover failure and took her 27 crew members with her...
The dangers of iron ore as a cargo were amply demon- In view of all of the above discussion regarding bulk
strated in November when the Trade Daring broke her carrier losses, it may however be said that (a) the vessels
back and subsequently sank halfway through a loading lost generally start out being adequate to the demands of
operation at the Brazilian port of Ponta da Madeira. the sea, but may have become inadequate at some point
Loading at a rate approaching 16,000 tonnes an hour late in their life, (b) heavy weather and water ingress are
stressed the hull o f this 20-year-old vessel beyond break- implicated, and (c) water ingress may be through side
ing point... ". shell cracking or possibly forward hatch cover damage.
The picture is confused, but it is extremely unlikely that
A statistical study by BTCE (1993) in Australia, also each of these vessels somehow met an environment more
indicated that iron ore was a relatively common factor severe than its design extreme storm, so to speak. What is
associated with bulk carrier structural failures. But corre- more probable is that with passage of time, the structure
lation is not causation, and the role of iron ore in vessel or parts of it weakened, e.g., by corrosion, fatigue, or
losses is by no means certain. In fact as previously noted, increased levels of deformation. The sensitivity of the
iron ore and coal are major types of cargo in bulk carriers, structure to failure thus increased, or its damage tolerance
and so their presence in a proportion of lost vessels is not decreased, and accidental events such as water ingress
in of itself surprising. occurred leading to a progressive set of undesirable events

4 Strength of Bulk Carrier Structures


floors and girders. Together these three structural ele-
ments apparently account for a third of all local structural
damages, according to Shama. In addition, Ivanov (1993)
reports the following types of side frame structural dam-
age as typical:

1) At bracket connections of the side frame to the


wing and hopper tanks, fractures and potential detach-
ment of brackets occur frequently.
2) Side frames can often be found twisted, aided by
mechanical damage and corrosion.

The detachment of hold frame ends forward, and the


Q Hold Frame-Connection to upper & lower wing tanks and spread of that type of damage through a domino effect
side shell. involving load shedding to nearby corrosion / wear weak-
~) Boundaries of transverse bulkheads and bulkhead stools. ened structure might have been an important cause of
| Cross deck structure. bulk carrier losses. Anecdotally, it is known that in a
@ Hatch corners/hatch coaming brackets. number of vessels that survived to tell the tale, water in-
@ Localized cracking and buckling of web frames and break- gress occurred in hold No.l, through failure of the side
down of coatings in water ballast tanks.
shell structure between hopper and wing tanks, typically
| Inner bottom plating/hopper plating intersection.
near the aft bulkhead. A schematic of this type of failure
is shown in Fig. 4.
Fig. 3 Problem areas in bulk carrier structures
3. STRUCTURAL CHARACTERISTICS OF CON-
V E N T I O N A L BULK CARRIERS

To understand the structural characteristics of con-


ventional bulk carriers, some important properties of ex-
Broken~~ isting bulk carrier structures are now surveyed. These
structural parameter studies would be of value in judging,
describing and generalizing the nature of structural failure
behavior to be expected for given operational, extreme
~"""SddemP~otinffgelctSoen
s SadjPaPc~ b:its:Chcd I I~
and accidental load levels.

3.1 General Characteristics


Figure 5 shows the relationship between ship length
and deadweight for bulk carriers with a large proportion
of which are conventional. It is evident that there is a
definite relationship between ship length and deadweight,
with moderate degree of scatter. Typically the number of
holds is seven for the Panamax and nine for the Capesize
Fig. 4 A schematic of side shell damage class, with some variability. There are essentially three
types of holds in such vessels in so far as the structure is
concerned; i.e., ore holds, light holds and ballast holds.
as previously noted. When carrying dense cargo such as iron ore, an alternate
A brief look at local structural failures typical in bulk hold loading condition is common, with the light holds
carriers (that survived) is of interest. While forensic ex- empty.
amination of lost vessels is difficult, indirect information Figure 6 shows the relationship between ship length
from existing local failures experienced may serve as an and maximum hold length. It is seen from Fig. 6 that the
indicator of the major potential causes of vessel losses. maximum hold length tends to decrease with increase in
Figure 3 from a study by IACS indicates areas where lo- the ship length. It is of interest that the hold lengths in the
cal structural failures occur in bulk carriers. Of these, a Panamax are comparable to the Capesize, as shown in
study by Shama (1995) indicates that damage to side Fig. 6, but the Capesize hold areas and volumes are nor-
frames is the most frequent, followed by damaged bottom mally greater than those ofa Panamax (e.g., areas approxi-

Strength of Bulk Carrier Structures 5


400 2.8

O
DW=alL3+a2L2+a3L+a4 [ ~
a1 = 0.000014 | * .% 2,4
~" 3O0
o a2 ~0002337 ~.. .V
.T=
a3=0127091 I
E 0
a4=8451520 I * ~176176176 2,0 o~ O 0
r 200 o 0

(30 0 O
t~O 1.6 0
100 o

0
1.2 i i i i
120 160 200 240 280
0 I I I I I
100 150 200 250 300 350 L B P (m)

LBP (m) Fig. 7.a The relationship between ship length and double
bottom height for conventional bulk carriers
Fig. 5 The relationship between ship length and dead-
weight for bulk carriers 55
0
0
30 0
50- O

o o co o ~l 0000o
~n
28 0
O O "&
0 0 0 '~= O
45-
o .~ 0 0
26
o o ~ o E 0
g q~ o
24 40- O
o o OD
E 0 0
O
22
E 0 O
35 I I I I
20 120 160 200 240 280
O
L B P (m)

18 i i i i Fig. 7.b The relationship between ship length and bottom


140 180 ~20 260 300 plate slenderness ratio for conventional bulk car-
LBP (m) riers

Fig. 6 The relationship between ship length and maxi-


1.8
mum hold length for conventional bulk carriers

~ 1.6

mately 1.5 times for hold No. 1; i.e., most forward, and 1.25
times in the case of the other holds, in one particular set = O 1,4
O~ o
O
of vessels). Hence normally any postulated flooding of
the Capesize hold No. 1 is of more load consequence than o
O
flooding of the Panamax hold No. 1. O O
O
O
3.2 Hull Structures 1.0 So
Figures 7.a to 7.d show some important properties of
conventional bulk carrier hull structures of Handymax 0.8 I I I I
class or larger. The bulk carrier double bottom height 120 160 200 240 280
(and also width of flat part of inner bottom) increase re- LBP (m)
markably as the vessel becomes larger, see Fig. 7.a. As Fig. 7.c The relationship between ship length and hull
shown in Fig. 7.b, the slenderness ratio of outer bottom section modulus at bottom for conventional bulk
plating decreases (e.g., the bottom plate thickness increases) carriers

6 Strength of Bulk Carrier Structures


100 of aspect ratio and reduced slenderness ratio for bulk car-
rier plating between stiffeners. The plate aspect ratio in
80
8 the bottom and side shell is about 3.0 in way of longitudi-
o
nal framing (and aspect ratios 30 to 40 in way of trans-
o verse framing), while it is about 6.0 in the deck. It is also
60
seen from Table 3 that the reduced slenderness ratio of
o
o
longitudinal plating is less than 2.0, indicating that such
O
40 plating of bulk carriers is usually stocky in a buckling
o
0 sense. It should be noted that for long plate compression,
0
o if the plate slenderness ratio is less than about 1.9, buck-
20 0
0 ling normally occurs in the elastic-plastic or plastic re-
o
gime, in which the buckled plate will have little residual
0 0 ; i ; strength margin beyond first buckling, while thin plate
I00 150 20(1 Z,SO 300
elements with the plate slenderness ratios greater than
LBP (m)
about 2.5 will normally first buckle in the elastic regime,
Fig. 7.d The relationship between ship length and HTS with a fair amount of post-buckling strength reserve until
usage for conventional bulk carriers the ultimate limit state is reached.
Bending and torsional rigidities of stiffeners and
supporting members in bulk carriers were also surveyed.
with increase in the vessel size. The ratio of actual hull As would be expected, the rigidities of transverse mem-
section modulus to Rule required section modulus de- bers are usually larger than those of longitudinal mem-
creases as the vessel becomes larger, see Fig. 7.c, which bers. It was observed that the bending rigidity of sup-
implies that some vessels are built to satisfy the rule re- porting members is large enough compared to that of
quirements with little margin. Figure 7.d indicates the plate elements so that the relative lateral deformation of
usage of high tensile steel in conventional bulk carrier plate elements along the edges could be ignored in calcu-
hull structures. More than 70% of the hull structures in lating the plate buckling strength. However, the normal-
Panamax and Capesize bulk carriers can in some cases be ized torsional rigidity of supporting members; i.e., the
of high tensile steel. All statistics indicated are for a se- ratio of torsional rigidity of supporting members to the
lected cross section of vessels. bending rigidity of plating between supporting members,
Table 3 indicates typical ranges and average values is less than about 1.0 for longitudinal stiffeners and less

Table 3 Structural characteristics of plates between stiffeners in selected existing bulk carriers

a/b b / t~f~o / E
Structure
Range Average Range Average
Outer Bottom Plate 2.9-3.4 3.2 1.6-2.1 1.9
Inner Bottom Plate 2.9-3.4 3.2 1.3- 1.8 1.6
Bottom Floor 2.0-2.9 2.6 2.3-3.0 2.4
Bottom Girder 3.2-4.0 3.6 1.8-2.8 2.3
Side Shell 3.2-3.3 3.3 2.0-2.2 1.6
Deck Plate 4.7-6.7 5.7 1.0-2.0 1.6
Longitudinal Bulkhead Plate 3.2-3.3 3.3 2.2-2.4 2.3
Topside Wing Tank Bottom Plate 4.9-7.5 6.3 1.9-2.7 2.3
Topside Wing Tank Web 1.0-1.6 1.3 2.2-2.9 2.5
Hopper Bottom Plate 1.9-3.7 2.8 1.6-2.5 1.9
Hopper Web 1.0-1.6 1.3 2.0-2.7 2.5

Note: a = plate length, b = plate breadth, t = plate thickness o-,, = yield strength, E = Young's modulus

S t r e n g t h of B u l k C a r r i e r S t r u c t u r e s 7
than about 2.0 for transverse frames. This could indicate
that although the ship plating has a finite amount of rota-
tional restraints along the edges implying that idealized
boundary conditions; i.e., simply supported or clamped,
never occur, there is still the possibility of tripping failure
of stiffeners, particularly in the longitudinal direction
(which needs to be controlled by design).
The various obvious generalizations of this section
relate of course to strength, and thus are only indirectly
indicative of the structural performance considering loads
as well. Also, it should be noted that part of the conven-
tional bulk carrier side structure is transversely framed,
for which the limiting rigidities and slenderness would be
different from those noted above. In conventional bulk
carriers, each side shell flame may not form part of a
continuous ring; i.e., the transverse flame spacing in the
topside tanks can be different from that in the cargo hold,
which in turn can be different from the frame spacing in (a) Corrugated transverse bulkhead and other structure in
the double bottom, although the recently suggested dou- the cargo hold of a bulk carrier
ble skin bulk carrier design is a clear exception (Liu et al.
1995).

3.3 Corrugated Transverse Bulkheads


In addition to other types of structural components
discussed above, we surveyed corrugated bulkhead
structure characteristics as well. Figure 8 defines related
nomenclature. The typical corrugated transverse bulkhead
in existing bulk carriers is vertically corrugated and with-
out horizontal girders. Lower and upper stools with dif-
a/2 a/2
ferent proportions from each other are normally located at B
its ends. Most corrugated bulkheads have shedder plates
on top of horizontal lower stool plates. (b) Nomenclature for the corrugated bulkhead structure
Figures 9.a to 9.c show some important properties of models and analysis
corrugated bulkheads. The corrugation span normally
increases as the vessel size becomes larger, see Fig. 9.a. Fig. 8 Features of the analysis of bulkhead strength
(Example corrugation spans are 12 m for a Panamax and
16 m for a Capesize bulk carrier). The statistics of the
20
corrugation angle showed scatter, but they were in the
range of 55-90 degrees. The corrugation shape in the 0 0
ballast holds is typically rectangular, while in the ore or ~
E
16 0
0
light holds it is typically trapezoidal. 0
0
0 o
0
oo
o
It is also observed from the survey that the ratio of
the corrugation flange width to web width (web depth in 0 0

case of a rectangular corrugation) is in the range of ~:~ 8


0 0

0.633-1.630, with an average value of about 1.0, see Fig. o


9.b. The variation of thickness of corrugation flange or
4
web with increase in the vessel size was studied. The re-
duced slenderness ratios of the corrugation flange or web
0 i i i i
are in the range of 0.867-3.003 or 1.521-3.413, respec- laO ~so 2eo 260 300
tively and their average value is about 2.5 (with consider- LBP (m)
able scatter as may be expected because of differing local
pressures and related load sustaining needs). The column Fig. 9.a The relationship between ship length and corru-
(reduced) slenderness ratio of a single corrugation is in gation span for some conventional bulk carrier
the range of 0.3005-0.5855, and its average value is about corrugated bulkheads

8 Strength of Bulk Carrier Structures


2.0" 4. H Y P O T H E T I C A L BULK CARRIER STRUC-
T U R A L DESIGNS

1.5- For the purpose of illustrative structural performance


calculations we designed two hypothetical bulk carriers,
namely a 150,000 DWT Capesize and a 67,000 DWT
,--. 8~ o ~ Panamax. The number of holds is seven for the Panamax
1.0 -

0 o o o and nine for the Capesize. Both vessels are considered


0 0
O with heavy iron ore loaded in alternate holds.
=l~ 0
0.5-
The profile and principal particulars of the Capesize
are given in Fig. 10 and Table 4, respectively. The profile
and principal particulars of the Panamax are given in Fig.
0.0 I I I I
11 and Table 4, respectively. The vessel midship sections
140 180 220 '2.60 300 are similar to those of Fig. 24, but considerably fuller. All
LBP (m) vessel particulars shown in the figures and tables are
gross scantlings; e.g., including corrosion margin. The
Fig. 9.b The relationship between ship length and ratio of vessels are designed to the 1985 ABS Rules. Hence their
corrugation flange to web breadth for conven- gross scantlings are different in critical areas from what
tional bulk carrier corrugated bulkheads would be required today under the SafeHull based ABS
Rules (ABS 1995, 1997), but the reason why these ves-
sels are so designed is that they are more representative of
the existing fleet. The Capesize bulk carrier hull structure
3 0
O is made of H36 steel, but in the Panamax the structure is
mild steel, except for the deck region where H32 steel is
2.5 O
used. The design still water bending moment is applicable
0 O
0 to both hogging and sagging conditions as the Rules of
00
the time made no distinction between the wave induced
E~ 0
0 bending moment in hogging and sagging.
g~
The hold double bottom structure is stiffened by lon-
o 4 gitudinal girders and transverse solid floors. Much of the
0 net local loading (external minus internal pressure) on the
double bottom is transmitted to transverse bulkheads in
0.5 shear, but a portion is also carried by the bottom longitu-
dinal girders in bending. The longer the hold compared to
0.0 I I I I its breadth, the more important the girder system in
140 180 220 2(;0 300 bending. So in a relative sense, the girder system in the
LBP (m) Panamax carries a greater proportion of the net double
bottom load in bending than in the Capesize.
Fig. 9.c The relationship between ship length and corru-
gation section modulus for conventional bulk 5. H U L L G I R D E R AND LOCAL LOAD EFFECTS
carrier corrugated bulkheads DUE TO A C C I D E N T A L FLOODING

5.1 Hull Girder Load Changes


0.45, with relatively moderate scatter. The corrugation The hull girder loads can be decomposed into still
section modulus increases as the vessel size becomes water and wave induced load components. Each of these
larger, see Fig. 9.c. Note that the data in this figure are load components is usually calculated separately, and the
not normalized for material grade. total load is obtained by combining the two load compo-
The aspect ratios of the lower stool top plate and up- nents. Detailed distribution of the still water loads along
per stool bottom plate are in the range of 3.4-7.2 (with the ship length can be predicted by integration of the dif-
average of 5.2) and 2.7-5.6 (with average of 3.8), respec- ference between the weight force and the buoyancy force,
tively. Also, the reduced slenderness ratio is in the range using simple beam theory. Also, using strip theory, the
of 1.7-2.7 (with average of 2.2) for the lower stool top distribution of wave induced loads along the ship length
plate, and in the range 2.1-2.5 (with average of 2.3) for can be estimated. Ingress of sea water due to accidental
the upper stool bottom plate. flooding will cause changes to the hull girder load distri-

Strength of Bulk Carrier Structures 9


TYP ]
13.2m1
i i
' - ~ : =J-c : " ' - - i - "'" Z ~ -~.........'~ ,,..,..~ ~ , . ~ ~ :, .,,: : " " - - " """ "= , . c I
/
Hold9 Hold8 Hold7 Hold6 Hold5 Hold4 Hold3 Hold2 Holdl
)
L23.76mL23.76ml 23.76nLL23.76m, 23.76m1 23.76m1 23.76m ] 23.76m 125.52m
47 74 101 128 155 182 209 234 263 292
'F.S. 800ram Frame spacing=880mm F.S,
~1 ~
LBP=264m "~l~800m~

LOA=276m

Fig. 10 A profile of the Capesize bulk carrier

o**_:
oa
1.1.O,.)111 1,9.O,.-3111 1 1 ..~,,.)111

E mlmllmll ii6imm lmlmllllml


L 24.'8m Il 42
~1 ~
25m
72
25m ~l~ 25m
102 132
25m
162
25m
192
25m
222
22.5m ,Ijl l
249 253.5
J

. 780ram ~ Frame ~ = 8 3 3 m m
LBP=215m
LOA=220m

Fig. 11 A profile of the Panamax bulk carrier

Table 4 Principal particulars of two illustrative vessels

Item Capesize Panamax


Length, overall (m) 276 220
Length, perpendiculars (m) 264 215
Length, scantling (m) 262.3 213
Breadth, molded (m) 43.5 32
Depth, molded (m) 24.0 18.3
Draft, design (m) 17.6 12.0
Speed, design (knots) 14.1 15.5
Block coefficient, design 0.84 0.855
Max inner bottom pressure, hold Nos. 1,3,5,7 (9) 39.3 26.8 t/m 2
Max inner bottom pressure, other holds 24.4 14.3 t/m z
Hold aspect ratio (=hold length/breadth) 0.55 0.68 hold No. 1
Typical 0.78 Typical
Side frame end brackets Non-integral
lapped Integral
Deadweight (DWT) 150,000 67,000
Design SWBM (ton-m) 310,000 134,600
Material All H36 All MS (H32 deck)

10 Strength of Bulk Carrier Structures


bution in the intact case, and these effects can be similarly 16

included in the calculations. 12


To explore the possibility of hull girder distress due
to accidental flooding, the extreme hull girder load redis- L6 L2 L3
O O
tribution for a Capesize bulk carrier in the flooded condi- ~ ~ 4
o/)O
tion is studied (Paik 1994). The long term prediction of
~ 0
the wave induced loads was carried out using a strip the-
~ -4
ory based method with a probability level o f 10 -s as-
sumed for purposes o f defining the extreme load. As pre-
viously noted, a Capesize vessel has nine cargo holds, and
0'3 -12
in heavy ore load condition hold Nos. 2, 4, 6 and 8 in this
case are empty. The ballast and full load conditions in in- -16 ' I ' I l I ' l ' I I I 9 I ' I

tact (i.e., unflooded) situation are considered first. In ad- 0 2 4 6 8 10 12 14 16 18 20

dition, five cases of hold flooding are considered as indi- (A.P.) Station number (F.P)
cated in Table 5. For the flooded condition, sea water was
taken to be present in addition to the ore cargo. The hold Fig. 12.a Distribution o f still water bending moment in
volume up to the idealized cargo fill level is taken to have the assumed flooding scenarios (Paik 1994)
a permeability of 0.3, while the rest o f the hold space is at
20-
an assumed permeability of 0.95.
Figures 12.a to 12.d show the distributions o f still 16- . L(
water and wave induced hull girder loads for each flood- 12- /t

ing case. In the hypothetical flooding conditions L5 and . 3 -~ 8

L 7 the vessel is (as expected) found to be at risk due to 4


_~x
loss of reserve buoyancy, and hence the results o f these 0

two situations are not shown in the figure. Based on the -4

computed results, some observations can be made: -8

-12
1) It is clear that some cases o f hold flooding can
-16
amplify the magnitude o f the hull girder loads occasion-
ally even beyond the design values. Decrease of the re-
-20 ' ' ' 't ' I ' I ' I ' I ' I

2 4 6 8 10 L2 14 16 18 20
sidual hull girder strength (e.g., due to vessel local struc-
(A 1'3 Station number (F.P.)
tural loss) together with the increase of the hull girder
loads may possibly lead to hull girder breakage.
Fig. 12.b Distribution o f still water shear force in the as-
sumed flooding scenarios (Paik 1994)
Table 5 Hold flooding scenarios considered in the calcu-
lations (Paik ! 994) I L6
Probability level ~----~jL2
7 = .

Scenario
Load Condition and Holds Flooded 6
No. E~
5
L1 Ballast Condition / Intact
L2 Alternate Ore Load Condition / Intact 4

Alternate Ore Load Condition / Hold No.5 3


L3 Flooded e--
2
Alternate Ore Load Condition / Hold Nos.
L4 1
4 & 5 Flooded
Alternate Ore Load Condition / Hold Nos. ' I ' I ' I ' I ' I ' I ' I ' I ' I ' I
L5
4, 5 & 6 Flooded 0 2 4 6 8 10 12 14 16 18 20

Alternate Ore Load Condition / Hold No.l (A.P.) Station number (F.P.)
L6 Flooded
Alternate Ore Load Condition / Hold Nos. Fig. 12.c Distribution o f wave induced bending moment
L7 in the assumed flooding scenarios (Paik 1994)
1 & 2 Flooded

Strength of Bulk Carrier Structures 11


moment distribution in the intact condition is first calcu-
Probability level = 10-s L6,,,,,~ lated. Loaded hold No.1 is then considered to be
10 L2 breached, with open communication to the sea. For the
L3 resulting flooded condition, the still water shear force and
~Y" s L4 bending moment distributions are recalculated. Using the
above results together with the wave induced shearing
6 force and bending moment envelopes from IACS unified
requirements (Nitta et al. 1992), the total shearing forces
N 4 and bending moments along the vessels' length are esti-
mated. The load increases due to flooding are then esti-
2 mated, and related critical locations along the vessel
length are identified.
0 The vessels are considered to be loaded with heavy
2 4 6 8 10 12 14 16 18 20
ore. Three ore stowage factors were considered, namely
(A.P.) Station number (F.P.) 10, 14 and 22 ft3/ton corresponding to ore specific
weights of 3.59, 2.56 and 1.6 ton/m 3 (As an aside, ac-
Fig. 12.d Distribution of wave induced shear force in the cording to Mano et al. (1976), ore related inertial forces
assumed flooding scenarios (Paik 1994) have a significant influence on the double bottom stresses
for a bulk carrier for cargo densities greater than 2
ton/m3). Calculations are performed by standard hydro-
statics and longitudinal strength computer programs.
2) For the flooded conditions L5 and L 7 in which
We should note that the flooded case considered in
three holds amidships or two forward holds are flooded, this study is similar to the IACS UR S17 case. The hold is
respectively, it is seen that the vessel can founder since breached, water enters, and the vessel finds a new still
the draught exceeds the ship depth resulting from loss of water equilibrium position, at a new waterline that is dif-
the reserve buoyancy. It can be said that if more than two ferent from the old, which defines the amount of water
cargo holds are flooded in laden condition, the possibility ingress. Our study differs from premise of the IACS UR
of foundering could be significant. This is of course not S17 in that we assume that the entire lifetime extreme
unexpected because bulk carriers are ordinarily not de- wave induced load may apply in a flooding situation,
signed to such a (2 or 3 compartment) standard for while the UR more realistically uses a reduced (80% of
flooding or damage stability purposes. design) wave load in such a case. We should also note
3) For the flooded conditions L4 and L 6 in which here for the sake of completeness that
two holds amidships or one forward hold are flooded,
respectively, it is seen that the magnitude of extreme hull 1) We compared through comparative calculations
loads could become very large, potentially leading to hull (not shown) that flooding of hold No.2 instead of hold
girder collapse. In this particular case, it is thus evident No.1 would result in a less severe case in these vessels.
that if more than one cargo hold particularly forward part We did not consider hold Nos. 1 and 2 flooded, as it
is flooded the possibility of foundering due to hull girder would be unreasonable to expect these vessels to survive
collapse could be large even if the survival buoyancy is such a progressive flooding condition as previously men-
sufficient in the beginning stages of flooding. Moreover, tioned.
once flooding occurs in one hold, particularly forward 2) We also confirmed through comparative calcula-
part, progressive flooding into the adjacent holds by col- tions (not shown) that the sagging wave condition is of
lapse of transverse bulkheads is possible if the bulkheads less interest in these particular vessels, because the still
had not been previously specifically designed to with- water bending moment is of a hogging nature. We did not
stand such accidental flooding conditions. consider ballast (including heavy ballast) conditions in
this study. While relevant for design, service experience
The above considered flooding cases are perhaps un- does not support the view that bulk carriers are typically
duly severe in comparison to the design bases of the ves- lost in ballast.
sels concerned. Thus to judge the degree of stress in-
crease in the structure due to a more realistic flooding Still water draft and freeboard summaries for the ves-
scenario, a finite element analysis was performed for the sels in the intact and flooded conditions are given in Ta-
hypothetical Capesize and Panamax vessels of Table 4. ble 6. Note that in the table the minimum freeboard any-
The vessels are considered to be in the alternate hold where along the vessel length is shown. The results indi-
loading condition. The still water shear force and bending cate the following:

12 Strength of Bulk Carrier Structures


Table 6 Still water draft and freeboard summary

Item Capesize Panamax


Draft, intact (m) 17.18 13.01
Depth, molded (m) 24.00 18.30
Beam, molded (m) 43.50 32.00
Draft, flooded (m) (10 c f t / L T ) 20.61 13.66
Draft, flooded (m) (14 cft / LT) 19.73 13.43
Draft, flooded (m) (22 cft / LT) 18.66 13.33
Trim, intact (deg.) 0.06 aft 0.30 aft
Trim, flooded (deg.) (10 c f t / L T ) 1.07 fwd 0.58 fwd
Trim, flooded (deg.) (14 cft / LT) 0.78 fwd 0.26 fwd
Trim, flooded (deg.) (22 c f t / L T ) 0.43 fwd 0.13 fwd
Freeboard, intact (m) 6.61 4.86
Freeboard, flooded (m) (10 cft / LT) 3.90 3.97
Freeboard, flooded (m) (14 cft / LT) 4.68 4.56
Freeboard, flooded (m) (22 cft / LT) 5.60 4.82

1) The still water draft and trim changes due to from the aft perpendicular in the Capesize and 55% in the
flooding of hold No. 1 are significant in the Capesize, but Panamax. The location is near mid-hold, e.g., hold No.4,
not the Panamax. As expected, the denser the ore, the which is an empty hold in both vessels. A further study of
greater the changes, because more water is admitted into the tabulated results shows that:
the hold.
2) The changes in draft, trim and freeboard due to 1) The weight differentials due to hold No.l flooding
flooding are relatively greater in the Capesize vessel are 6,111 tonne for the Capesize and 2,479 tonne for the
when compared to the Panamax. This is because of the Panamax; i.e., the Capesize value is about 2.5 times the
larger hold volume flooded and its greater distance from Panamax value.
the trim center in the Capesize. 2) The flooded forward body still water shear force
3) The intact freeboard in the Capesize is 1.7 m (SWSF) and total shear force (TSF) increases are 1.6 and
greater than the Panamax, as would be explained by the 1.4 times the intact values for the Capesize, and 1.4 and
relationship of freeboard to the vessel beam. Interestingly, 1.3 times for the Panamax. Hence there is a significant
the flooded minimum freeboards in the two vessels are increase in SWSF forward, more so for the Capesize than
comparable (about 4 m), meaning that the relative loss in the Panamax (1.6 times contra 1.4 times).
freeboard (and hence the possibility, of green water ship- 3) The flooded mid-body still water bending moment
ping) is greater in the Capesize. (SWBM) and total bending moment (TBM) increases are
1.9 and 1.3 times the intact values for the Capesize, and
The longitudinal distributions of total shear force and 1.7 and 1.2 times for the Panamax. Hence there is a sig-
bending moment for the vessels in the intact and flooded nificant increase in SWBM amidships, by about the same
conditions are then obtained. The shear force and bending factor for the Capesize and the Panamax (1.9 times contra
moment conventions used are that a positive sign is asso- 1.7 times).
ciated with hogging bending moments, and a positive 4) The bending moment increase due to flooding
shear force corresponds to bow section up. From such amounts to 9.67 and 8.35 metric tonne-m per tonne of
data, the maximum and minimum values of the shear differential weight at the location of maximum total shear
force and bending moment anywhere along the vessel force in the Capesize and the Panamax vessels. The same
length can be obtained. increase at the location of maximum total bending mo-
From the results obtained, Table 7 shows that the ment calculates to 39.0 and 31.8 tonne-m per tonne of
largest total shear force value occurs at the aft bulkhead differential weight. Hence the bending moment increase
of hold No.l. Table 8 shows the largest total hogging influence coefficients at the locations considered are
bending moment occurs at about 63% of vessel length similar in both vessels (within 20%).

Strength of Bulk Carrier Structures 13


Table 7 Effects of flooding on shear over 25% vessel length forward

Condition SWSF (tonne) SWBM (tonne-m) TSF (tonne) TBM (tonne-m)


Capesize, Intact - 10,216 122,538 - 15,057 324,047
Capesize, Flooded - 16,327 181,639 -21,168 383,148
Difference -6,111 59,101 -6,111 59,101
Panamax, Intact -5,763 50,076 -8,880 166,551
Panamax, Flooded -8,242 70,774 - 11,359 187,249
Difference -2,479 20,698 -2,479 20,698

Table 8 Effects of flooding on bending moment over the mid-ship region

Condition SWSF (tonne) SWBM (tonne-m) TSF (tonne) TBM (tonne-m)


Capesize, Intact 2,071 253,542 6,062 773,200
Capesize, Flooded 1,899 491,920 5,890 1,011,578
Difference - 172 238,378 -172 238,378
Panamax, Intact 1,238 118,889 3,564 369,604
Panamax, Flooded 1,556 197,612 3,882 448,327
Difference 318 78,723 318 78,723

To obtain the magnitudes of stress increases in the seven holds, with hold Nos. 1, 3, 5 and 7 loaded. They
structures due to flooding, finite element analyses are estimated the related individual long term values of hull
then performed. For this purpose, the following envelope girder bending stresses and stresses due to bending of the
dynamic loads along the vessel length are calculated us- double bottom structure itself subject to internal cargo
ing Section 3 of the SafeHull bulk carrier guide (ABS loads and external hydrodynamic water pressures. One of
1995), now absorbed into the ABS Rules: the interesting points in their results is that the external
hydrodynamic pressure on the double bottom acts oppo-
9 Wave induced vertical bending moment site to the internal cargo inertial forces, resulting in partial
9 Wave induced vertical shear force cancellation of loads. The same conclusion has also been
9 External hydrodynamic pressure independently reached by Kawamura et al. (1986, 1987).
9 Internal cargo pressure Hattori et al. (1984) also estimated the longitudinal
distribution of the apparent long term correlation coeffi-
The use of the SafeHull formulae is a matter of con- cient between hull girder bending and double bottom net
venience. The effective vertical accelerations in the ves- stresses. Their result indicates that (a) in empty holds, a
sels are found to be about 0.3g in No.4 hold and 0.6g in zero apparent correlation coefficient is appropriate; i.e.,
hold No.l, with pitch amplitudes in the range of 7 to 9 the total extreme stress is decided by either the hull girder
degree. The resulting dynamic loads are then combined bending stress alone or the stress due to maximum exter-
with the relevant still water loads to obtain the distribu- nal pressure alone, and (b) in laden holds, a correlation
tion of total loads. coefficient of 0.5 is appropriate; i.e., the largest total long
In the process of defining the loads for finite element term stress is about the same as what one would obtain by
analysis, load correlation and load combination issues for considering a combined load case where the long term
dynamic loads need to be considered. In this regard, maximum hull girder stress is combined with 1/2 of the
some interesting observations by other investigators are largest maximum double bottom stress due to net double
now pointed out. Hattori et al. (1984) investigated the bottom pressure alone, or vice versa.
correlation of loads on the double bottom of a 216m bulk Another set of approximations to the long term dy-
carrier traveling in head seas at 15 knots. The vessel had namic load correlation information for ships is provided

14 Strength of Bulk Carrier Structures


by Cramer et al. (1995) where the load combination fac- location in hold No.4 and in the side shell at bulkhead
tor between hull girder and local effects is 0.6. The local between hold Nos. l and 2. The highest stress in the
effects in turn are calculated from the external and inter- Panamax occurs at a bottom location in hold No.4.
nal effects with a correlation coefficient of 0.5 amidships 6) The stresses in the flooded condition are more
to 0.6 forward. In our case, we used a similar rule (but likely to be within limits in the Capesize than in the
with 0.5 in place of 0.6). Panamax. In the Panamax, stress limits are likely to be
exceeded both at a bottom location of hold No.4, a side
5.2 Local Stress Changes shell location at the bulkhead between hold Nos. l and 2,
To investigate the local load and stress characteristics and a transverse floor in hold No. 1.
due to accidental flooding, three dimensional finite ele-
ment models were constructed for the Capesize and the Note that all stresses referred to are von Mises effec-
Panamax vessels. The finite element models are pris- tive stresses, which include both axial and shear effects.
matic, and cover a three hold extent of each vessel, with Also, while of interest, the last two points noted above
scantlings corroded l 0%, thus approximating the net ship. should be interpreted carefully. This is because while the
Required local increase in side shell thickness in way of Capesize stresses are higher than the Panamax stresses,
bulkheads is accounted for in the finite element model. the Capesize has a larger yield related safety factor than
The part of the model whose structural response is of the Panamax since the entire Capesi;,e is of H36 steel,
interest is the mid-hold. The models are loaded so that the while the Panamax is of mild steel except for its H32 steel
target bending moment at the middle of tile mid-hold and deck.
/ or target shear forces at the bulkheads of that hold are Based on the stress results obtained, the following
realized. In the bending moment and shear force cases the conclusions could be drawn regarding the onset of yield-
mid-hold is empty, and the adjacent holds are laden. In ing in the structure: (a) the deck and bottom of empty
the pressure load case the mid-hold is laden, and the adja- hold No.4, and the side shell close to the bulkhead be-
cent holds are empty. A heavy cargo specific gravity of tween hold Nos. 1 and 2 are the likely critical locations to
2.6 (stowage rate 14 ft3/ton) is used. The following con- govern design. Of these, the side shell location is the most
clusions could be drawn from an inspection of the results important, followed by the deck location, and (b) local
of related finite element analyses: shear reinforcements at the bulkheads (i.e., increases in
side shell thickness) can be important.
1) Among the three cases considered, the moment The above calculations consider the structure in way
case with hold No.4 empty is the most important, fol- of the maximum total bending moment and shear force
lowed by the shear case with hold No.2 empty. The pres- locations to be intact. Any actual loss of side shell will
sure load case, with hold No.1 laden, showed high dramatically affect the shear related safety factors for the
stresses only for the transverse floors of hold No. I. region adjacent to transverse bulkhead No.l. In an ex-
2) In magnitude, stresses in deck, bottom and side treme case, we would have most of the shear load previ-
girder in the Capesize are somewhat higher (about 10%) ously carried by one of the side shells transferred to the
than the Panamax, both in the intact and the flooded case. other, intact side. Since it is the side shells that carry most
3) In magnitude, stresses in the side shell in the of the shear, the shear related safety factor will then be
Capesize are 25 % higher than the Panamax in the intact clearly inadequate; i.e., the vessels will not normally sur-
case, and 35 % higher than the Panamax in the flooded vive such a situation if heavy seas are encountered unless
case. Due to local strengthening, the gross side shell specifically designed for.
thickness in the region of interest is 22.4mm for the
Capesize (in contrast to the 18ram elsewhere), and 22ram 6. U L T I M A T E STRENGTH C H A R A C T E R I S T I C S
for the Panamax (in contrast to the 17 to 18ram else- OF C O R R U G A T E D BULKHEADS
where). As a point of interest, the local strengthening of
side shell plating in the Capesize extends 5m either side The ultimate strength characteristics of corrugated
of the bulkhead, while in the Panamax, it extends 7.5m bulkheads in a flooded situation are also studied experi-
either side. mentally, theoretically and numerically. In the following,
4) Flooding of hold No.l increases stresses in the some important conclusions based on the related studies
deck and bottom by up to 20% in the Panamax and 30% are briefly summarized. The results noted should be use-
in the Capesize. The side shell stresses, on the other hand, ful for the design or corrugated bulkheads in general. In
are increased by 40% in the Capesize and 30 % in the addition, the data developed facilitates the quantification
Panamax. of modeling errors in simplified bulkhead strength for-
5) The stresses are within limits in the intact condi- mulations for both conventional design and structural
tion. The highest stress in the Capesize occurs at a bottom reliability assessment purposes.

Strength of Bulk Carrier Structures 15


6.1 Experimental Study may or may not be so in future novel designs. However,
Most previous tests related to collapse of corrugated related work on the collapse strength of corrugated bulk-
bulkheads were carried out on single corrugations. The heads under a combination of lateral pressure and axial
pioneering work on the collapse tests using corrugated compression so far appears to be unavailable.
bulkhead models (not single corrugations) was that of Thus the present authors performed a new series of
Caldwell (1955). He performed a series of tests on four collapse tests on nine mild steel corrugated bulkhead
aluminum alloy and four mild steel corrugated bulkhead models in order to make possible new contributions to
models subject to water pressure. The overall dimensions some of the above mentioned problems. All models were
of the corrugated bulkhead models were 900 mm in made by folding (or corrugating) fiat mild steel sheets
length and 560 mm in breadth. The plate thickness of the with the aid of a mechanical bending machine. The di-
bulkhead models was in the range 0.554-1.712 mm. It mensions of steel sheets before folding were 1,560 mm in
was found in these tests that the maximum deflection of length and 1,219 mm in breadth. Two kinds of corruga-
vertically corrugated bulkheads was "virtually constant" tion angle, namely 60 and 90 degree, were used, and so
across the width of the panel except near the vertical the final breadth of the test models was either 1,125 mm
edges. These test results thus showed that the behavior of for a 60 degree corrugation angle or 750 mm for a 90
a single central corrugation could be taken as representa- degree corrugation angle. The number of corrugations
tive of the whole bulkhead, provided the conditions at the (also called pitch) for all models is five. Three kinds of
sides of each corrugation are the same, which is usually plate thickness, namely 0.8 mm, 1.2 mm and 1.6 mm,
found to be the case except near the comers of the bulk- were tested. The a/c ratio for the models (see Fig. 8) was
heads where there is a slight increase in apparent torsional about 1.0. The slenderness ratio of the corrugation flange
stiffness of the flange. and web was in the range 1.7-3.0, and the column slen-
The Caldwell tests provide much useful information. derness ratio for a single corrugation is in the range
However, the corrugation angle in most models of Cald- 0.41-0.55 which correspond to the range for actual cor-
well was small compared to ships, their range being rugated bulkheads in ships. A detailed description of the
12-42 deg. except for one mild steel and one aluminum tests is given by Paik, Thayamballi & Chun (1997). Some
model having corrugation angles of 62 deg. This means noteworthy characteristics of the tests are:
that, strictly speaking, only one test datum for a mild steel
corrugated bulkhead model with a more representative 9 Testing of mild steel corrugated bulkhead models
large corrugation angle exists. More test data on mild 9 Testing of models with large angles of corrugation
steel corrugated bulkhead models with larger corrugation typical of ships
angles are thus desirable. 9 Investigation of corrugation angle effects on collapse
Most collapse tests on single corrugations or corru- behavior
gated bulkhead models were performed under water pres- 9 Investigation of plate thickness effects on collapse
sure. As a result, the post collapse behavior of test models behavior
could not be fully observed because the tests typically 9 Investigation of loading condition effects on collapse
needed to be stopped immediately after (and sometimes behavior
before) collapse of the models to prevent failure in the *Evaluation of collapse strength and post-collapse
test facility. As a loading method, use of compressed air behavior
would be another possibility which would make possible 9 Tabulation of results in a form suitable for verifica-
the loading of test models into the post collapse regime, tion of guidance procedures for ultimate strength cal-
so as to observe both post collapse as well as ultimate culation of corrugated bulkheads
limit state behavior.
In the uneven alternate hold loading condition for Through this experimental study on nine mild steel
bulk carriers carrying dense cargo such as iron ore, corrugated bulkhead models, the following conclusions
shearing forces will normally be imposed in vicinity of could be drawn:
the hull girder cross section corresponding to the trans-
verse corrugated bulkheads. Even if the static shearing 1) The deflection pattern at the ultimate limit state of
forces in this case usually are to induce axial tension in the models was very similar to that observed by Caldwell.
the corrugated bulkheads, depending on the sea state, the The conclusion of Caldwell that the behavior of a single
total shearing force may in some cases possibly act so as central corrugation can be taken as representative of the
to induce axial compression in parts of the bulkhead, whole bulkhead, is thus reconfirmed.
which in turn can reduce the ultimate strength of corru- 2) It was noted that the ultimate limit pressure for a
gated bulkheads under lateral pressure. While such axial particular model P60-1 with 60 degree corrugation was
compression may be insignificant in existing vessels, it roughly 40% of that for another particular model P90-1

16 Strength of Bulk Carrier Structures


with 90 degree corrugation. By making the corrugation global bending detbrmation) which is observed in thin
angle smaller, the overall breadth of a corrugation can be walled structures under large compressive loads, such as
made larger, resulting in apparent savings of material. those for instance in ship collision.
However, the test results indicate that the ultimate col- 6) The ultimate strength interaction relationship for
lapse pressure load for the 60 degree corrugated bulkhead the corrugated bulkhead models subject to combined ax-
may in some cases be only one third of a 90 degree bulk- ial compression and uniform lateral pressure was ob-
head. This would imply that in certain cases the corru- tained. It was seen that the ultimate collapse strength of
gated bulkhead of a trapezoidal corrugation profile with the corrugated bulkhead could in principle be quite sensi-
smaller angle can result in a heavier structure than that a tive to the presence of axial compression in addition to
rectangular profile with its larger corrugation angle in lateral pressure. Traditionally the design of corrugated
order to sustain the same magnitude of applied pressure bulkheads has been performed under lateral pressure
loads. loads alone. In bulk carriers carrying dense cargo such as
3) It was observed during the tests that the internal iron ore, an alternate hold loading condition is common
resistive force dropped rapidly after collapse, implying a and, as a result, large sectional shearing forces are applied
sudden load shedding phenomenon which is of course vertically at the vessel fore and aft locations of transverse
structurally undesirable. This observed behavior of the (corrugated) bulkheads. The still water sectional shearing
models may in part be due to the fact that after initial forces normally act as axial tensile loads, but the total
collapse of the corrugated bulkhead models, the applied loads (including wave induced shear) may in some cases
pressure tends to concentrate around the most deflected possibly act in axial compression depending on the sea
point, resulting in further acceleration of the deformation. state. Such axial compressive loads may significantly
Potentially counteracting this is the membrane action in affect the ultimate collapse strength of the bulkheads un-
the model, which if the edge conditions are stable, should der lateral pressure. While the relevance of this observa-
increase the internal resistive forces in a sense. While tion to actual ships needs further study, it is evident that
these aspects need additional investigation, it is for the the coexisting axial forces in actual ships is small, simply
time being prudent to assume that the post-ultimate col- because the related crushing behavior is not usually ob-
lapse of corrugated bulkheads is likely to be rapid. served.
4) With increase in the plate thickness the collapse
strength of corrugated bulkheads significantly increases 6.2 Theoretical Study
as well. Even for the thinnest corrugated bulkhead models As shown in Fig. 13, corrugated transverse bulkheads
under lateral pressure, sudden local buckling deflection of in dry cargo holds are usually designed to withstand three
plating (flange and web) did not take place before the load components, namely:
inception of collapse of the entire bulkhead. This would
imply that for purposes of predicting the collapse strength 9 Lateral pressure due to dry cargo and / or flooding
of actual corrugated bulkheads having relatively thick water
plates any effects of pre-existing deflection due to lateral | Carry-over bending moment resulting from overall
pressure may be ignored. It was also seen that with in- double bottom bending which is important in alter-
crease in the thickness of corrugation flange / web, the nate hold loading situations
ultimate bending moment for a single corrugation of 90 9 In-plane axial force due to the net double bottom
degrees increases at a faster rate than that for 60 degrees. pressure
This trend would imply again that barring any other prac-
tical considerations, a rectangular corrugation profile The last load component is a funclion of cargo mass
could be more efficient than a trapezoidal profile with its and draft, while the first two load components are mainly
smaller angle of corrugation. related to the cargo mass. The present authors derived a
5) As previously noted, the collapse behavior of the simple analytical formulation for the first cut prediction
corrugated bulkhead models under axial compression was of the ultimate bending moment of corrugated bulkhead
investigated for the two different corrugation angles. The cross section subject to lateral pressure loads (Paik,
tests were continued until crushing (folding) occurred. Thayamballi & Chun 1997). An approximate relationship
The test results showed that axial compressive loads between the bending moments and the applied pressure
would increase without occurrence of local buckling of loads at the ultimate limit state of corrugated bulkheads
flange up until the ultimate strength of the whole corru- was also derived (Paik, Thayamballi & Kim 1998).
gated bulkhead model was reached. Initially, unloading Figure 14 represents a single corrugation that is a
follows after the sudden collapse, but due to the internal proxy for the behavior of a corrugated bulkhead subject
contact of walls the internal forces may rise again. The to lateral hydrostatic pressure and end moments (e.g., due
latter is a typical crushing (folding) response (without to rotational restraints of adjoining structures). For sim-

Strength of Bulk Carrier Structures 17


FD It would be very desirable to properly evaluate the
end conditions for the corrugation in order to predict the
collapse strength more accurately. In this regard, based on
finite element studies for existing bulk carrier corrugated
bulkheads, Ivanov (1994) suggested simple expressions
for the rotational restraints at upper and lower ends of
corrugation in terms of the properties of the adjoining
structures.
In a design sense, rotational restraints at the upper
end may be practically regarded as zero, and further the
upper end support rigidity in the direction normal to the
pressure load direction should not be overestimated. On
the other hand, since the lower end is connected to heavy
adjoining structures (i.e., double bottom) one may be
Fig. 13 Applied loads on corrugated transverse bulkheads tempted to assume a clamped end condition for the corru-
gation. However, as will be discussed later, the simply
supported end condition might be more appropriate even
at the lower end of corrugation, in lieu of actually mod-
eling the effect of adjoining structures, but with the carry-
over moment due to double bottom bending otherwise
considered.
Conceptually, when both ends of the corrugation and
Mo(B,- L ., ')Mo any one point inside the span yield, a collapse hinge
,z mechanism is formed. Depending on the actual structure,
(a) Loading end conditions and loading, details of the formation of the
collapse mechanism will vary. A plastic hinge is formed
when the maximum bending moment reaches the ultimate
bending moment of the corrugation beam cross section,
with the corrugation flange / web in compression buck-
ling and the corrugation flange / web in tension yielding.
Based on a credible stress distribution over the cross sec-
tion of the corrugation at the ultimate limit state, and as-
suming that the entire material in compression of the cor-
(b) Elastic bending moment distribution rugation reaches its ultimate buckling strength while the
material in tension is in full yielding, we can derive (see
Fig. 14 A single corrugation beam under lateral pressure Paik, Thayamballi & Chun 1997) a simple ultimate
bending moment capacity formula of a single corrugation
by integration of the assumed stress distribution with re-
plicity, the corrugation is modeled as a beam with con- spect to the final neutral axis, the result being as follows:
stant cross section (and thus constant bending rigidity). It
is assumed that its ends can move freely axially, implying g2
that membrane stresses are not developed. The lateral M , = Crof A f g + CrowA w sin qb d
hydrostatic pressure distribution with a trapezoidal pat-
tern is assumed to vary linearly between the bottom and + cr,wA w sin qk ( d - g)2 + A/(d - g) (1)
deck structures.
The end moments shown in Fig. 14 arise from the
where
constraints against angular rotation of the corrugation
cross section at the junctures of the bulkhead and the ad- M , = ultimate bending moment
joining structure. They thus depend on the torsional ri- Af , A,, = section area of corrugation flange or web
gidity of the adjoining structures. The adjoining structures d = vertical height of corrugation web
such as deck stool, hatch end beams and upper stool g = final neutral axis at the ultimate limit state
brackets at the upper end and bottom stool, shedder plate
~b = corrugation angle
and double bottom structures at the lower end are the
main members contributing to the rotational restraints. troy, Crow= yield strength of corrugation flange or web

18 Strength of Bulk Carrier Structures


cr.f,ff.w = ultimate compressive strength of corrugation supported, and 2) the lower end clamped and the upper
end simply supported, with results as follows:
flange or web

In our study, the ultimate compressive strength of (1) Condition with both ends simply supported
corrugation flange or web (i.e., cr,f or a , . , ) was calcu-
lated by the formula (Paik & Thayamballi 1997a) M, = p---s {Xp2- Lxp) PB- ]gO
+ Lxt, - ~ ) (4)
-

or,, = [0.996+0.17fl 2] -~ (2)


O"o where
L
where X p "~ - - when PD =Ptt
2
~, = cr,r for flange I ~ ") ' 3 ' ' --
L p~ _ ~ PV~+ PD + PDP;~
= o-,w for web Xp--
P~ - PD 5
cro = Crof for flange
when P6, > PD
= Crow for web

orOan, (2) Condition with lower end clamped and upper end
f ~ simply supported
_a ~/~_~_~_ for web
= ~ W -- t w
M,, = xp{(L2 -Xp)PD + ( 2 L 2 - 3Lxt' + xap)Pz
t f , t., = thickness of flange or web (5)
6(2L - xp)
More recently, IACS adopted the following expres-
sion for predicting the ultimate bending moment capacity where xe is determined so that M,, takes the minimum
of the corrugation cross section, namely value.

M. = (0.5. Z;e "a,;e + Z,,," era,,,,) x I0 -3 kN-m (3)

2.0
where
Z;e,Z,, =section modulus of one half pitch corrugation,
in cm 3, at the lower end and mid-span of corru- 0 "90deg.] /
gations, respectively
cr,.;e,o-a,,, = allowable stresses, in N/ram 2, for the lower
end and mid-span of corrugations, respec-
~0
1.5

/o
tively
o 1.0
=or o -5 0.Smm
Go = minimum upper yield stress, in N/ram 2, of the ma-
terial
0.5

Both of the above noted formulae provide the ulti- //'-~ ~ 1.2mm
mate bending moments (not the ultimate applied pressure
loads) at the ultimate limit state for the corrugation cross
0.0 I I I I
section. To predict the corresponding ultimate applied 0.0 0.5 1.0 1.5 2.0
pressure loads it is necessary to derive the relationship 3
between the applied pressure loads and the bending mo- (Pu)experiment(kg/cm-)
ments at the ultimate limit state. The authors (Paik, Fig. 15 Comparison of the ultimate pressure loads from
Thayamballi & Kim 1998) derived such relationships for experiments with the new analytical formula,
two idealized end conditions, namely 1) both ends simply equation (1)

Strength of Bulk Carrier Structures 19


Table 9 Types of CORBHD/FEM models

Model Structures Included in Modeling


Stool Upper Lower Shedder Deck Double
Type Corrugation plate stool stool plate bottom
I 0 X X X X X X
II 0 0 X X X X X
III 0 0 X X 0 X X
IV 0 0 0 0 X 0 0
V 0 0 0 0 0 0 0

A comparison of the above derived ultimate strength


formula, equation (1), with the experimental results from
our corrugated bulkhead model tests under uniform lat-
eral pressure; i.e., p~ = PD, can be made, assuming that
the end conditions of the test models are simply sup-
ported. Figure 15 compares the relevant theoretical solu- Cross deck ,~ , Upper stool
tions with the corresponding experimental results. The
comparisons shown indicate that the derived ultimate
strength formula agrees well with the test results. For the "41 ,it
I'"l iqi
models with larger thickness of corrugation flange and ]1 i ,

web, the formula tends to underestimate the ultimate Corrugation - - " " : , , . '
pressure load for a single corrugation; i.e., it is somewhat
, q ,i
pessimistic in a design sense. (One of the main reasons 't I
for this could be the fact that the corrugation ends were
assumed to be simply supported). In any event, from such Shedder plate X ~
theory versus experiment comparisons, the related mod- Inner bottom
Lower stool X
eling error (prediction accuracy) of the formula can be ,,, / Outer bottom
defined, whether for structural reliability assessment or Bottom girder ~ , ~ / ~
any other purpose.

6.3 Numerical Study


A special purpose nonlinear finite element program
CORBHD/FEM for efficiently analyzing the progressive
collapse behavior of corrugated bulkheads in detail was Fig. 16 A CORBHD/FEM model with all adjoining
developed by the authors, for accurately taking into ac- structures (Model type V)
count the influence of the boundary condition due to ad-
joining structures, e.g., deck and double bottom structures
(Paik & Thayamballi 1997b, Paik, Thayamballi & Kim lower and upper corrugation ends in order to include the
1998). The program automates the structural modeling of deformation of the corrugation at the juncture of the ad-
the corrugated bulkheads and the adjoining structure. joining structure more smoothly. Five types of structural
Further, it uses special purpose finite elements for more modeling for the corrugated bulkheads are automated
efficiently analyzing the elastic-plastic large deformation with or without adjoining structure, as indicated in Table
behavior of such structures (Paik & Kim 1989). 9. Figure 16 shows a CORBHD/FEM model including all
As the representative extent of the analysis for corru- adjoining structures. As options the program can vari-
gated bulkheads, a half pitch of corrugation is taken. Web ously include the adjoining structure such as lower stool
and half-flange of the corrugation are modeled by a num- top plate, upper stool bottom plate, shedder plates, deck
ber of 8 and 4 node shell elements. In the longitudinal and double bottom which may significantly affect the
(vertical) direction of the corrugation, a total of 43 ele- (boundary conditions for the) corrugated bulkhead be-
ments are employed. A relatively fine mesh is used at the havior. As indicated in Fig. 17, two types of pressure load

20 Strength of Bulk Carrier Structures


into an adjacent hold of a bulk carrier can occur is by the
--~pol+-- rupture of a corrugated bulkhead and / or its boundaries
i
after the reference cargo hold has been flooded. There-
fore, it would be desirable for the structural analyst to get
information related to when and where rupture initiates
during the failure process fox"the bulkhead. With increase
in the applied pressure loads, the CORBHD/FEM pro-
I gram thus checks for the occurrence of first rupture at the
finite element level, considering that rupture occurs if the
total (accumulated) equivalent plastic strain reaches a pre-
defined critical rupture strain.
X To verify the accuracy of the CORBHD/FEM pro-
gram and also to investigate the influence of adjoining
structures on the collapse behavior of corrugated bulk-
ip. . heads, a bulkhead model P90-3 employing a 90 deg. cor-
Pressure may act on both of corrugation and inner rugation under uniform lateral pressure as tested by the
bottom or on corrugation alone. authors was analyzed by using the program.
Three types of structural modeling were considered.
Fig. 17 Pressure loading patterns available Since the original l:est models are composed of corruga-
tions plus minimal upper / lower stool plates (without
adjoining structures), the first type of the calculation
model was made to represent the original test model
(with-out adjoining; structures) by assuming the simply
>,~ 2.0
supported condition at both ends; i.e., no lateral deforma-
r tion and no rotational restraints. The second and third
types of the calculation model were constructed such that
v 1G-
the original test model is represented with artificial ad-
joining structures nominally similar to the double bottom
r
and deck of a vessel.
I/ In these initial calculations, the plate thickness of the
.~ 08 artificial adjoining structures was taken to be either the
M .... CORBHD/FEM same as or much larger than that of the corrugation and,
N /~" - -C- - Experiment I as a result the "adjoining structure" rigidity was either
O,,1
small or very large. In the former case it may be expected
that due to the weak artificial adjoining structures, while
O0 I t I I the rotation at the corrugation ends would to some extent
10 20 30 40 50
be restrained, out-of-plane lateral deformations could
Deflection at mid-span (ram) occur not only along the span but also at its ends. On the
Model with upper stool bottom plate and lower stool top other hand, in the latter case with the strong adjoining
plate only (simply supported condition at both ends) structures, out-of-plane lateral deformations as well as the
~) Model with all adjoining structures (small dimensions) rotation could in some instances be overly restrained at
@ Model with all adjoining structures (large dimensions) the corrugation ends.
Figure 18 shows these expectations to be true. It is
Fig. 18 Average pressure load versus deflection at mid- seen from the figure., that the numerical solution obtained
span, COPBHD/FEM versus experiment for the original model (i.e., without adjoining structures)
with simply supported edge condition agrees reasonably
with the experimental results. (There were minor differ-
patterns, one for trapezoidal pressure loads acting on cor- ences in bending stiffness between FEM and experiments
rugation alone, and the other for trapezoidal pressure due to the fact that in the tests lateral deformations oc-
loads acting on corrugation and uniform pressure acting curred even near the edges while in the analysis model,
on inner bottom panel are available. The water pressure they could not). In the calculations, the critical rupture
acting on the oute~" bottom panel is not specifically con- strain was assumed to 15%, but no rupture took place
sidered. until the model reached the ultimate strength. The results
One manner in which potential progressive flooding of Fig. 18 indicate that adjoining structures such as deck

S t r e n g t h of B u l k C a r r i e r S t r u c t u r e s 21
6.0

~ , x Design formula with B.C.2 (upper bound)


5.0
t ........................................
[ ~ Model type V

~ - Model type III with B.C.2


4.0
/ /~W/~ Design formula with B.C. 1 (lower bound)
ra0 . . . . . . . . . . . . . . . . . . . . . . . . .

3.0
/ / / ~ \ Model type III with B.C.1

/// ~\ ModeltypeIIwithB.C.1
2.0
/ \ Model type I with B.C.1

1.0
// [ B'CC'91"iBl~ rendsrsimply supp~
I B.C.2: Upper end simply supported & lower end clamped)
0.0 I I I I I I l I
50 100 150 200 250 300 350 400
Deflection at mid-span (mm)

Fig. 19 Maximum pressure load versus deflection at mid-span for a hypothetical Capesize bulk
carrier corrugated bulkhead, as obtained by CORBHD/FEM and design formulae

and bottom structures can affect the stiffness and strength corrugation ends even though the end rotations are
of corrugated bulkheads. If the adjoining structures are somewhat restrained, the resulting strength and stiffness
light, the lateral deflection at the ends can occur relatively may in some cases be different from those for the (usual)
freely, although the rotation at the ends may to some ex- simply supported edge condition. In general, this would
tent be restrained. As expected depending on the degree imply that in the direct strength calculations for corru-
of rigidity of the adjoining structures, the relative accu- gated bulkheads under lateral pressure, the end conditions
racy of theoretical and / or numerical procedures which for lateral deformation as well as rotational restraints at
model the corrugated bulkheads assuming idealized (e.g., both ends should be accounted for correctly.
simply supported) end conditions will vary compared to 2) The presence of shedder plates can raise the
reality. strength and stiffness of corrugated bulkheads, see Fig.
Subsequent to the test model related calculations of 19. The effects of the shedder plates on the ultimate
Fig. 18, the CORBHD/FEM program was applied to cal- strength of bulk carrier corrugated bulkheads may should
culate the ultimate strength of a transverse corrugated not be neglected.
bulkhead of the hypothetical Capesize bulk carrier when 3) Due to adjoining structures; i.e., deck and double
subject to the triangular pressure loading pattern for the bottom structures, both the stiffness and the strength of
corrugation alone; i.e., with PD = 0. Figure 19 represents corrugated bulkhead of actual bulk carriers can possibly
the ongoing relationship between the maximum pressure increase, meaning that rotational restraints due to the ad-
at the lower end of corrugation and the deflection at the joining structures can be large, even though the out-of-
mid-span of corrugation as the analyses proceeded with plane deformation at corrugation ends may be non-zero.
various model types indicated in Table 9. Based on the 4) A direct calculation model where both ends of the
results of the above noted calculations for the test model corrugation are simply supported is one possibility if we
and the Capesize bulkhead (see Figs. 18 and 19), the fol- wished to avoid modeling the adjoining structures. How-
lowing observations may be made: ever, since the lower part of the corrugation is in reality
connected to relatively rigid double bottom structures,
1) In a hypothetical case with small rigidity of ad- another possibility may be to assume end conditions
joining structures, lateral deformation can occur at the which are clamped at the lower end and simply supported

22 Strength of Bulk Carrier Structures


at the upper end. To establish such simplified direct cal- 7. CORROSION MODELS FOR BULK CARRIER
culation procedures and verify their relative accuracy STRUCTURES
needs further study, for which the nonlinear finite ele-
ment program from this study should prove quite useful. To establish cost effective maintenance and repair
5) In the calculations of Fig. 19, where the critical plans and also to predict the time dependent ultimate
rupture strain w~,s assumed to be 15%, no fracture was strength and related structural reliability of aging vessels,
indicated in any of the elements until the corrugated probabilistic corro,;ion rate estimation models for primary
bulkhead reached the ultimate strength. The potential for member regions are necessary. The corrosion rate of ship
progressive flooding into an adjacent hold is contingent hull structures is influenced by many factors including the
on such fracture, the crack formation related to which corrosion protection system and various operational pa-
apparently appears in the post-ultimate strength regime in rameters. The corrosion protection systems that may be
this particular case. Further study is needed to determine employed are coatings and anodes. The operational pa-
the generality of this particular observation, especially in rameters concerned include the percentage of time in
way of fillet welds. ballast, type of cargo carried, frequency and method of
tank cleaning, maintenance and repair, etc. The theoreti-
The ultimate pressure loads as obtained by the theo- cal prediction of the corrosion rates for the structural
retical formulae, e.g., equations (I) to (5), may also be members is not an easy task. An easier alternative is to
compared with the CORBHD/FEM solutions as shown in base the rate prediction on statistical analysis of past data
Fig. 19. The ultimate pressure loads indicated are ob- from comparable situations. In this section, significant
tained by substituting the ultimate bending moments; i.e., recent work on developing useful probabilistic corrosion
equations (1) or (3), into equation (4), for the case of the models for transverse bulkheads (by Yamamoto & Ike-
both ends simply supported condition, or into equation gami 1996) and longitudinal hull structure of bulk carriers
(5), for the case of the lower end clamped and the upper (by Paik, Kim & l_,ee 1998) is reviewed, and related re-
end simply supported condition. It is seen from Fig. 19 sults are summarized.
that the ultimate pressure loads of the hypothetical bulk Until a few years ago, the ore holds of bulk carriers
carrier corrugated bulkhead, as obtained by CORBHD / were normally uncoated. More recently, however, the
FEM considering the rigidity of real adjoining structures, related practice ha:~ changed, and bulk carrier hold re-
fi~ll somewhere between the two bounding approximate gions including transverse bulkhead and side frames are
solutions mentioned above. But note that the present now coated as indicated in Fig. 20. This aspect is impor-
CORBHD/FEM solutions have been obtained for the load tant to adjusting the corrosion rate models, where based
case of Fig. 17.b; i.e., pressure loads on the corrugation on the old practice, for consistency with the new. It also is
alone. If one considers the load case of Fig. 17.a with
inclusion of the water pressure acting on the outer bottom
panel in some manner, the ultimate strength can approach [_
the lower bound because the effectiveness of rotational
restraints of bottom adjoining structures may then de-
crease. From a strength point of view alone, the condition
with both ends simply supported would thus in some
cases be appropriate. Additional needed investigations are
continuing in this regard, to better assess the inclusion of J
the "carry-over" bending moment effects from adjoining Coating Coating ~'
I
structure within such a scheme. extent extent ~q
I
As a final point of interest regarding corrugated I
1
bulkhead strength, we should point out that a traditional 1
weak link in such bulkhead structures in some past bulk 1
carriers is the top plate of the transverse bulkhead stools,
which take connection stresses from the corrugated bulk-
head plating when subject to bending. In the past, experi-
ence has shown that through thickness properties (yield
and ultimate strength) of the stool top plate were in some
cases inadequate, e.g., 30 to 40% of the nominal values.
Thus currently, "Z" grade steel with acceptable through (a) Transverse Bulkhead (b) Side Frame
thickness properties is typically used for such top plating
exceeding a certain limit thickness. Fig. 20 Extents of coatings in bulk carrier holds

Strength of Bulk Carrier Structures 23


a reason why the models presented here need ongoing relationship between the depth of corrosion and the
refinement in the future. elapsed time after the generation of the pitting point:
The models summarized treat the time dependent
wear as corrosion related. There is evidence that such an r = q. U 2 (6)
assumption may be somewhat pessimistic for the early
periods of the stool where chance events related to grab
where r = depth of corrosion (ram)
damage may be more evident, in the sense that the corro-
T = elapsed time (years), and
sion rates predicted from the model exceed actual gaug-
q , c 2 = empirical coefficients
ing data.
Corrosion data and corrosion rate models such as
those presented here are potentially useful for many pur- Yamamoto & Ikegami provide analytical ways of es-
poses, including adjusting design corrosion margins to timating the time dependent diminution due to corrosion
better suit experience, and in predicting the reliability of using their model. However, the required probability dis-
bulk carrier structures as they age (The latter application tribution of the depth of corrosion as a function of the
will be discussed later). elapsed time can in practice be established in an alternate
manner by simulation wherein the coating life is sampled
7.1 Corrosion Model for Transverse Bulkheads first, the transition time is sampled next, and then the
A probabilistic corrosion model of bulk carrier trans- depth of corrosion is found as a function of the elapsed
verse bulkheads has been recently developed by Ya- and other time.
mamoto & Ikegami (1996). Their model is based on an In such calculations using equation (6), the exponent
analysis of survey records for 50 conventional bulk carri- c2 may be taken to be deterministic, the suggested value
ers of 10,000 DWT and over, placed in service after being 1/3-1.0 (Kondo 1987). The coefficient c 1 is taken
1986. There were about 1,100 thickness measurement to be a log-normally distributed random variable. Ya-
data for deep tank bulkheads, about 2,300 for watertight mamoto & Ikegami indicate that means and COVs of the
bulkheads and about 4,000 for the lower stool where cor- coefficient q are 1.105 and 0.5 for the deep tank bulk-
rosion was typically more severe. Because the resulting
head, 1.474 and 0.5 for the watertight bulkhead, and
model has been shown to relate well to actual time de-
1.289 and 0.43 for the stool regions, respectively. Their
pendent data for thickness diminution due to corrosion,
results indicate that during early vessel life, the corrosion
and it is the only such model thus far available for trans-
depth is small, but the associated COV is large. For in-
verse bulkheads in bulk carriers, relevant aspects of the
stance, at 5 years, the mean and COV of the corrosion
model are reviewed below.
depth are about 0.3 mm and over 2.0. Later in life, the
The rate of thickness reduction due to corrosion in a
mean increases while the COV decreases. For example, at
coated structure over time is conditional depending on (a)
15 years, the mean is about 3 ram, while the associated
loss of effectiveness of coating, (b) initiation of corrosion
COV is about 0.5. Essentially, it is early in life that the
damage, and (c) progress of the corrosion damage. The
coating breakdown and transition time statistics matter.
life of coating is the duration of the time between its ap-
Later, the variability in the depth of corrosion seems to be
plication and the time when it looses its effectiveness.
driven almost entirely by the variability in the parameter
The life of the coating may be considered to be a nor-
e I defining the wear function; i.e., equation (6).
mally distributed random variable. Yamamoto & lkegami
indicate the mean value of coating life to be 2.75 years The age dependent corrosion degradation statistics
for the deep tank bulkheads, 5.5 years for the ore hold calculated as above using the transverse bulkhead corro-
watertight bulkheads, and 1.5 years for the stool region. sion rate model suggested may also be readily applied for
The COV (i.e., coefficient of variation) of coating life is other related purposes, e.g., to predict the time dependent
approximately 0.4 (Emi et al. 1993). probability distribution of ultimate strength and the
The transition time; i.e., the duration between the time dependent structural reliability of corrugated trans-
time of coating effectiveness loss and the time of corro- verse bulkheads. Studies in this regard are ongoing.
sion initiation, is shown to be an exponentially distributed
random variable with mean of 3.25 years for the deep 7.2 Corrosion Model for Bulk Carrier Longitudinal
tank bulkhead, 2 years for the watertight bulkhead, and Strength Members
1.5 years for the stool region. Consistent with the expo- A probabilistic corrosion rate model for the longitu-
nential distribution, the associated COV of the transition dinal strength members of bulk carriers has been devel-
time is 1.0. oped by the authors (Paik, Kim & Lee 1998). The model
The progress of the pitting corrosion that follows is based on available statistical data for measured corro-
may be defined by a wear function which describes the sion damage of existing bulk carriers. A total of 7,503

24 Strength of Bulk Carrier Structures


i
0.15/0.18 ! 0.07/0 07
i

i
J \ o.o3/a15
i
0.06 / UPPER WING
UPPER S L O P 1 N ~ TANK SIDE SHELL
TANK

Bulk Carrier Corrosion Rates \


X
M E A N / S T A N D A R D D E V I A T I O N (ram/year)
i SIDE SHELL
0.09/ 0.1()k ~, //~xA r
] ~ 0 13/0 I0 ; LOWERSLOPING.A~I I

0.08,0.0, I , .141.13 / ~.=Tj.~.__j.~lr : ;t 0.03]__t7E SHELL

BOTTOM PLATE

Fig. 121 A probabilistic corrosion rate estimation model for bulk carriers (Paik, Kim & Lee 1998)

gauging data for 44 existing bulk carriers were used. Cor- computed results for the mean and standard deviation of
rosion rates were investigated for sixteen primary struc- the corrosion rate (e.g., uniform thickness reduction due
tural member regions in bulk carriers. to corrosion in mm per year) for each and every primary
As previously noted, the progress of corrosion would member region of a bulk carrier. Based on the results, the
normally depend very much on the degradation of anti- following observations can be made:
corrosion coating. Therefore, the corrosion model devel-
oped was divided into two parts, namely one that estab- 1) The corrosion rate for the boundary plating be-
lishes the life of' coating and another that describes the tween ballast water and cargo regions; i.e., the inner bot-
progress of corrosion, where it is assumed that the con'o- tom, hopper and top side tank bottom plating is higher
sion would start immediately after the loss of coating ef- than that for the bottom and bilge plating. This may be
fectiveness. It is known that the mean value of coating due to the fact that the former are exposed to both ballast
life in cargo holds is normally 5 to 10 years (Emi et al. and cargoes, while the bottom and bilge plates are ex-
1993). In fact, 5 years of coating life may be considered posed to ballast water alone as the external surface is
to represent an imsatisfactory situation, while 10 years normally fully coated.
would be representative of a relatively more desirable 2) The corrosion rate for the inner bottom plates is
state of affairs. Also, according to Emi et al. (1993), the relatively large. This may be due to the fact that the inner
COV of the normally distributed coating life is about bottom is subject to wear not only due to its being the
40%. boundary between ballast and cargo regions, but also due
Equation (6) may now be used to estimate the corro- to mechanical action of grabs, etc. and also accelerated
sion degradation of the longitudinal strength member local corrosion as a horizontal surface.
regions. In this case, the exponent c2 may also be treated 3) The corrosion rate of the side shell between the
as deterministic with the pessimistic value of 1.0. The hopper tank and the top side tank is slightly higher than
probability function of the coefficient c I is assumed to that of the shell in way of the hopper tank. This implies
that in some cases, the cargo concerned may be more
follow the Weibull distribution as indicated by Emi et al.
corrosive than ballast water, due in part to the humidity,
0993).
temperature and type of cargo (acidity) involved.
By analyzing, the measured corrosion data for longi-
4) Deck and side shell plating in way of top side
tudinal primary member regions of bulk carrier structures
tanks can be relatively heavily corroded compared to the
available, the sta6stics of the coefficient c I (e.g., depth of
other external surfaces. This may be due to more frequent
corrosion since cL~ = 1.0), namely the mean and standard occurrence of wet and dry cycles in the ullage areas of the
deviation, were established. Figure 21 summarizes the top side ballast tanks and the heat from the sun.

Strength of Bulk Carrier Structures 25


5) Most of the longitudinal stiffeners in a ballast tank 0.10 --

show quite similar corrosion rates except for the deck 0.98 --
longitudinal stiffeners. The relatively higher corrosion
0.96 --
rate observed for deck longitudinal stiffeners may be due
to similar reasons as for the deck plate, e.g., the heat from
0.94 -- sagging
=o
the sun. X 0.92--

0.90 -
.

. COLLAPSE S T R E N G T H R E L I A B I L I T Y OF 0.88-
B U L K C A R R I E R H U L L G I R D E R CONSIDER-
0.86 - repair
ING C O R R O S I O N
0.84 -
.......... norepair
A reasonably accurate and fast procedure for the reli- 0.82 -

ability analysis of aging vessels, based on the ultimate 0.80


' ' ' ' l ' ' ' ' l ' ' ' ' l ' ' ' ' l ' ' ' ' l ' ' ' ' l
collapse of the ship hull girder, and taking into account 0 5 10 15 20 25 30
the degradation of primary members due to general corro- Ship Age (year)
sion has been developed by the authors (Paik, Kim, Yang
& Thayamballi 1998, Paik, Thayamballi, Kim & Yang Fig. 22 Variation of the ultimate strength mean value with
1998). In outline, the procedure is as follows: The prob- age
abilistic characteristics of the corrosion rate for the mem-
ber regions are defined by using the corrosion rate model
presented in section 7. The time dependent ultimate ca- 10

pacity model of ship hulls under vertical bending is for-


mulated by using an analytical formula derived by Paik & 0.8
Mansour (1995), taking into account the effects of corro-
sion degradation for each primary member region. Local
load effects, e.g., due to normal pressure, are currently 9- 0.6-
Innerbo~x
~
not included. The probabilistic model for the extreme
vertical bending moment in the vessel life time (e.g., 20 0 0.4-
years) is established using a simplified direct wave load r~

prediction method developed by Loukakis & Chryssos-


tomidis (1975) and an IACS design guidance formula for 0.2-
still water bending moment (Nitta et al. 1992). The time
variant reliability index associated with hull girder col- 0.0 .... I .... I .... I''''1 .... I ....
lapse is then calculated by using the second order reli- 5 10 15 20 25 30
ability method (SORM). ShipAge(year)
As an illustrative example, the procedure developed
is applied to assessment of ultimate strength reliability of Fig. 23 Probability of first repair for selected members of
the hypothetical Capesize bulk carrier. Through the the hypothetical Capesize bulk carrier
analysis, the trends of section modulus and ultimate hull
strength with increase in vessel age can be investigated.
The probabilities of steel renewal due to corrosion for culation assumed to be subject to corrosion once it starts
different parts of the vessels can be predicted. The ex- (albeit at different degradation rates), and under the re-
pected reduction of the ultimate strength reliability level newal criteria used in these particular calculations, no
with vessel age can be obtained. renewal was necessary until the vessel was more than 20
In the particular calculations shown, it is considered years of age, see Fig. 23. The assumption of uniform cor-
that corrosion initiates 5 years from the time of new rosion is mostly unrealistic in usual ship structures, but is
building. Two types of repair conditions for heavily cor- certainly an extreme (worst case) possibility. The ne-
roded local members are included, namely either that they glected effect of spatial variability in corrosion may be
are repaired or renewed to original state, or they are not. significant, and thus no complete confidence may be
These obviously represent the two hypothetical extreme placed on the results which purport to pessimistically
decision limits. Figure 22 shows mean values of the time show more than 10% hull girder strength degradation
dependent ultimate hull strength of the vessel, with their during the vessel life (It is to be noted that IACS require-
members corroding at the rates indicated in Fig. 21. Each ments normally limit such degradation to 90% of the
and every member in the hull cross section is in this cal- minimum required hull girder section modulus).

26 Strength of Bulk Carrier Structures


Under the pessimistic assumptions used, certain parts In summary, the procedures developed should be
of the structure continue to remain in their corroded state useful for assessing the ultimate strength reliability of
even when other parts are renewed to their original state, aging bulk carriers taking into account the degradation of
because of relative differences in their renewal criteria. members due to corrosion. With further research and de-
This is why in ~he present calculations, the local plate velopment, which is ongoing, the procedure should also
renewals when they occur do not apparently return the be useful as an aid for the establishment of first principles
global strength parameters to their original values even in based cost effective maintenance plans for specific ves-
an average sense. These results may be expected to be sels involving corrosion. By effectively applying such
qualitatively different if the corrosion were to be local- technology at the new building stage, a design can be
ized to selected parts of the structure in a probabilistic rationally optimized for new building as well as life cycle
manner rather than considered to affect large parts of the costs.
entire structure uniformly. The procedures developed in
this paper could of course be applied using more realistic 9. F A T I G U E P E R F O R M A N C E OF BULK C A R R I E R
assumptions related to the spatial variability of corrosion, SIDE FRAME ,STRUCTURES
provided related data are available. In any event, the fol-
lowing conclusions can be drawn from our present stud- 9.1 Calculation Outline
ies: A study of the fatigue performance of bulk carrier
side frame structures is now made (Liu & Thayamballi
l) The section modulus and ultimate strength of cor- 1997), as side she]il failure is a potential mode of water
roded hulls can potentially decrease with time, but the ingress into a cargo hold. The example calculations ad-
degree of decrease is controllable through proper tech- dress the hypothetical Capesize and Panamax vessels pre-
nology application, inspections and steel renewal criteria. viously described in Figs. 10 and 11, respectively.
2) The effects of spatial variability in corrosion can The Capesize calculations are made for a draft of
be significant. The spatial variability of corrosion rates at 18.7m and speed of 14.7knots, while the Panamax calcu-
various points in :service time needs quantification. lations use a draft of 13.3m and speed of 15.5knots. The
3) The probabilistic corrosion rate estimation models envelope wave induced bending moments, shear forces,
for longitudinal strength members of bulk carrier struc- and external hydrodynamic pressures were calculated for
tures as previously described are useful for evaluating the the two vessels, from hold No.4 through hold No.1 for-
time variant hull girder strength reliability of aging ves- ward, using the load component formulae of the ABS
sels. The models, however, may need further develop- Rules (1997).
ment as more extensive corrosion data become available. The calculations showed that the external pressure
4) By appropriate renewal of corroding local mem- loads are larger toward the forward locations of the vessel
bers, the level of ultimate strength reliability of the hull is in comparison to midship. From mid-hold 2 to mid-hold
increased immediately after a repair. With properly se- 1, the waterline pressure anaplitudes were estimated to
lected renewal criteria relating in part to the minimum vary from 18 to 23 ton/m 2 for the Capesize, and 15 to 21
plate thickness required, or equivalently, the permissible ton/m 2 for the Panamax. The external dynamic pressures
plate thickness reduction of primary members subject to at the bilge were about 40% of the stillwater waterline
corrosion, and also in part to the required global hull values. The local draft used in the calculations was found
girder performance, a pre-specified level of structural to mainly affect the static and not dynamic pressures.
safety can be maintained. Based on the dynamic pressure load study, side shell
5) Consideration of the interacting local and global fatigue is checked in hold No.l, in the vicinity of the aft
strength degradation effects in a corroding hull is possible transverse bulkhead of the hold. Specifically the side
using the ultimate hull strength based reliability proce- frame lower connection details at the intersection of the
dure suggested. From such a procedure, one can obtain a side shell and the sloping plate of the lower (hopper) tank
nominal measure of likelihood of hull girder failure, with are of interest because that location is affected by wave
direct and explicit consideration of various load, strength profile changes in both the laden and the ballast condi-
and corrosion related uncertainties on a statistical basis. tions. It also happens to be a known location of side shell
6) The various applicable performance considera- failure in some vessels. The fatigue process at that loca-
tions must be fully taken into account in such calcula- tion can be considered to be driven primarily by external
tions. For example, local renewal limits may address local dynamic pressure if" one assumes that the ore in the laden
degradation in buckling performance. Present determinis- condition is left as poured, although in our fatigue analy-
tic criteria typically address these aspects through slen- sis we have included the additional effect of stresses due
derness ratio limits. It is noted that the calculations of Fig. to secondary bending of the double bottom structure
21 do not directly address local strength. arising from the clifferential (internal minus external)

Strength of Bulk Carrier Structures 27


S: Scantling, L" Laden, HB: Heavy Ballast, LB- Light Ballast
/ !

JI

i
Item Panamax Capesize L
Hold Frame (mm, max.) 833 880
Floor (mm, max.) 2500 2640 HB
Hopper Web (mm, max.) 2500 2640

LB __ Top S i d e T K W e b (mm, max.) 4167 5280 LB

I !

Panamax Capesize

Fig. 24 Location of side shell with respect to still water draft

double bottom pressures as well. 9 Weibull shape parameter

9.2 Loading Conditions for Fatigue Analysis Using these the long term distribution of the stress
To judge the alternating load magnitudes to be used range can be defined. Once that long term distribution
for fatigue analysis, an idea of the location of the struc- and the associated number of fatigue stress range cycles is
tural detail (lower end of side frame) with respect to the available, the fatigue damage may be calculated using the
laden and ballast waterlines is of interest. For the appropriate S-N curve for the structural detail being
Capesize, the mean still water draft in an alternate hold studied. In our study, we will pessimistically use a design
loading ore laden condition is 17.2m (even keel). The (e.g., lower-bound) S-N curve rather than a curve repre-
typical draft in a heavy ballast condition is 10.9m (10.3m senting mean or average fatigue capacity.
forward). In the case of the Panamax, the design draft is The extreme stress range in a nominal 20 year life is
12.2m, the scantling draft is 13m, a typical light ballast calculable from the corresponding extreme pressure
draft is 6.2m (4.5m forward), and a typical heavy ballast ranges. The extreme pressure range and hence the ex-
draft is 8.1m (7.6m fwd). These drafts are shown plotted treme stress range corresponding to the laden and ballast
in Fig. 24, which indicates that for all practical purposes, cases are first calculated, and the larger of the two taken
the still waterline in the vessels varies roughly between as the extreme stress range to be used in the fatigue
the top and bottom connections of the side frame. analysis. In our case, the larger stress was estimated to
The comparative fatigue analysis that we undertake is result from the laden pressure distribution, which is con-
of the simplified type used in SafeHull, with minor modi- sistent with the relative locations of the still water lines in
fications as noted earlier. To calculate the fatigue damage the laden and heavy ballast cases and also the fact that the
for a structural detail by such a method, we need to know laden and heavy ballast side shell extreme dynamic pres-
the following: sures are not vastly different. The pressure in the head
seas case was assumed to be the 20 year extreme pres-
9 Extreme stress range sure, which adds some uncertainty to the stress range es-

28 Strength of Bulk Carrier Structures


Table 10 Fatigue analysis parameters for side flame frame are estimated from Section C2.2 of the Bulk Car-
rier Guide, now part of the ABS Rules (1997). These are
No. Item Capesize Panamax 0.912 for the Cape, size and 0.931 for the Panamax, so a
Weibull shape parameter of 0.9 was eventually used in
Side Pressure both vessels. The number of load cycles (zero to extreme
1 15.3 13.8
(ton/m:)
amplitude) is nominally taken as 0 5 x 108 in 20 years,
Internal Pressure which should be adequate for comparative purposes al-
2 3.3 down 2.3 down
(ton/m 2) though specific values for pressure processes at specific
3 Weibull Parameter 0.9 0.9 locations may difti:r somewhat from the assumed value.
The various fatigue analysis parameters are shown in Ta-
4 Cycles (20 years) 0.5 x 108 0.5 x 10s
ble 10. The side frame spacing is 880 mm for the
Capesize and 830 mm for the Panamax, the plate thick-
ness being about 22ram in either case. Figure 25 illus-
trates related side shell details.

~ with SCF

Ewith SCF
/ 9.3 Fatigue Damage Estimates
As previously noted, fatigue calculations are made
for the bracket lower end of the frame where it joins the
sloping plate of the hopper tank. The fatigue procedure
used is essentially the same as that suggested in the Safe-

E with SCF
~ wilh SCF
Hull bulk carrier roles, except that three structural cases
are considered, namely the gross structure (i.e., no corro-
sion in 20 years), a 10% corroded structure (i.e., 10%
corrosion in 20 years), and a 20% corroded structure (i.e.,
20% corrosion in 20 years). Tile time variant nominal
stress range is calculated per the SafeHull side frame re-
sponse model. Cak:ulations for the Capesize do not give
(a) Panamax (b) Capesize credit to any bracket flange plate at the lower end, while
the calculations for the Panamax do, because of an inte-
Fig. 25 Side shell details in way of bulkhead No. 1 gral bracket. The UK DEN E curve is used with an SCF
of 2.
The resulting damage estimates (e.g., by Miner's
timates in the sense that the other directions are not spe- rule) are shown in Table 1l. Fatigue lives for a Miner
cifically considered. This could have been avoided, but sum of unity are also shown. Using the same S-N curve
the head seas assumption in severe waves was felt to be and SCF, the sensitivity of fatigue damage estimates to
adequate for comparative purposes. decrease in side .,;hell thickness, decrease in bracket
In the laden condition, the vertical distribution of thickness, and decrease in side frame scantlings were also
side shell pressure at the longitudinal location of interest investigated. The results supported the view that the side
is idealized as uniform, with a value given by the average frame scantlings and the end bracket thickness affect the
of the values occurring at the top and bottom of the frame fatigue damage significantly, and that the damage is rela-
ends (20.4, 10.2 for the Capesize, 18.4, 9.2 for tile tively insensitive to the thickness of the side shell plating.
Panamax). This results in a uniform external pressure of The following conclusions could be drawn from the vari-
15.3 ton/m 2 for the Capesize and 13.8 ton/m 2 for the ous fatigue calculations and related sensitivity studies that
Panamax, compared to the waterline values of 20.4 and we performed:
18.4; i.e., roughly 75% of the waterline values. The ac-
companying net dynamic pressures on the double bottom 1) The side pressures calculated for the Capesize are
structure were also estimated and included in the fatigue about 10% greater and the stiffener spacing is about 6%
calculations, but their effect for these particular vessels is more, meaning that the Capesize pressure force is about
small, meaning that subsequent fatigue estimates may be 15% higher than tile Panamax. This, together with the
considered to be driven mostly by the side shell pressure added effect of the difference in unsupported span (which
alone. is 15% larger in the case of the Capesize than the
The Weibull shape parameter values defining the Panamax) gives rise to fixed end moments at the Capesize
long term stress range distribution for the side shell hold side frame that are nearly 50% greater compared to the

Strength of Bulk Carrier Structures 29


Table 11 Comparative fatigue damage estimates, E curve and SCF=2

Vessel Corrosion Stress Range (kg/mm z) 20 Year Miner Sum Fatigue Life (Years)
Gross 29.69 0.61 20+
Capesize 10% 32.99 1.47 14.3
20% 37.12 2.24 !2
Gross 21.09 0.18 20+
Panamax 10% 23.43 0.46 20+
20% 26.37 0.72 20+

most affecting the difference is the effective flange used


Hold frames of 5 bulk carriers / with the bracket.
(Total measuring points = 622) / 3) Fatigue estimates for the particular Panamax are
acceptable. For the Capesize considered, the fatigue esti-
C = 8 Xl 0"4/T3 fl~ / mates may be unacceptable for corrosion levels exceeding
E NK DATA /O ~ / 10%.
X
oaAxu WEB / 9 / 4) For these two particular vessels, fatigue was the
o only failure mode that indicated lower safety margins for
Xil,~,l'A FACE / x : / the Capesize in relation to the Panamax. Yielding, buck-
e- ling and ultimate strength were also checked in the same
~ study, but in those failure modes this particular Capesize
e- actually fared better than the Panamax.

In summary, in the context of side shell fatigue in


\ X = 0.IT(1 +C'T 3) bulk carriers, the importance of corrosion and fatigue of
the end bracket connections should be emphasized. Lev-
5 10 15 els of corrosion in way of bulk carrier side frame have
Ship age T (Years) occasionally been large, particularly in Capesize vessels
that tend to carry mostly coal and iron ore, see for exam-
Fig. 26 Corrosion damage of bulk carrier side frames ple Fig. 26 which shows corrosion data for bulk carrier
(Akita 1984) side frames as reported by Akita (1984). Also, in the par-
ticular vessels considered, the fatigue loading in the
Capesize is about 1.5 times the Panamax, due to the ac-
Panamax (71.81 ton-m versus 44.79 ton-m). For identical cumulated effect of unsupported span, frame spacing, and
end details, the fatigue stress range is proportional to the local pressure differences between the two vessels, To
fixed end moment, and the fatigue damage varies roughly what extent such an observation would hold for Capesize
as the stress cubed. Hence for identical end details one and Panamax bulk carriers in general needs to be studied.
would expect the Capesize side frame fatigue damage to Another aspect to be highlighted relates to account-
be more than three times that of the Panamax. In these ing for corrosion effects in the fatigue calculation proce-
particular vessels, the difference is even greater because dure itself. Because of the complex nature of local struc-
the end details are obviously not identical. tural behavior, and the sensitivity of fatigue lives to
2) In the two vessels / locations considered, the side stresses / corrosion, it would seem that fatigue damage
frame end connection details are not similar as was previ- calculation procedures need to account for corrosion in a
ously noted. The features of the end bracket are quite time dependent manner (much as we did in the above
important to fatigue performance. In general, an integral analysis). The question of how best to define an equiva-
bracket with an effective flange plate (such as that used in lent "net" structure for design purposes in such a case
the Panamax here) is far superior in terms of fatigue per- requires further study. Also, considering the relatively
formance than the non-integral bracket (which we as high corrosion rates evident in some cases in the past, the
sumed for the Capesize here). The calculation parameter strength of fillet welds at the side frame end connections

30 Strength of Bulk Carrier Structures


needs further study as well. So both control of potential water ingress primarily
through side shell or occasionally the hatches, and
10. DESIGN C O N S I D E R A T I O N S F O R B U L K CAR- strengthening of bulkheads are viable options to improve
R I E R SAFETY vessel safety.
In our example calculations, potential water ingress
Some important design considerations for improving in a Capesize hold loaded with iron ore is found to be
the safety and reliability of bulk carrier structures subject significantly more than in a similarly loaded Panamax,
to age and accidental flooding as obtained from various mainly because of differences in the residual loaded hold
facets of our study are noted below. volume. The resulting increase in shear forces forward
(close to bulkhead No.l) and bending moments aft (in
10.1 Control of Bulk Carrier Losses hold No.4) over the intact values is somewhat greater in
Bulk carriers that start out being clearly adequate to the Capesize than the Panamax (by factor of 1.6 versus
the demands of the sea and rigors of operation do not 1.4 in still water shear force and 1.9 versus 1.7 in the
always remain so with time. The average age of vessels bending moment).
lost each year is generally 15 years or more. The loss rate In our example calculations, stresses were obtained at
for aging Capesize bulk carriers is apparently more than six local locations, namely the deck at side, the bottom
that of Panamax. A significant proportion of the vessels plate in way of the outer double bottom girder, the inner
were lost while carrying heavy iron ore in alternate holds. bottom plating at transverse bulkhead stool, the transverse
According to our study of trade share estimates it is evi- floor at mid-hold, and the side girder one offcenterline, at
dent that Capesize vessels do carry either iron ore or coal bulkhead stool. With regard to the finite element calcu-
during a significant portion of their life, and coal is ad- lated stresses in the hull, the bending moment case was
mittedly a corrosive type of cargo. In addition, cargo the most importanl:, followed by the shear force case. The
holds in vessels built prior to the 1990s were uncoated. pressure load case was less significant.
Hence if one we:re to look for a potential initiator of these The largest stress increases due to cargo hold flood-
bulk carrier losses, one could not blame any one factor, ing occurred at the lower part of side shell in way of
but rather, a combination of them, including the rigors of bulkhead No.1 and were 40% above intact values in case
operation and human factors. of the Capesize and 30% tbr the Panamax. The safety
To reduce bulk carrier losses to acceptable levels, factor against yiekiing was generally less for the Panamax
one must address the problem on many fronts. Neither rather than the Capesize. This reflected a difference in
age, nor alternate hold loading, nor iron ore as a cargo, yield strength between the two vessels. Whereas the
nor the carriage of coal can be blamed in isolation. It is of Capesize as designed is all H36 steel, the Panamax is all
course true that in particular cases, homogeneous hold mild steel except fi~r the H32 steel deck region. Hence the
loading of dense., cargo instead of alternate uneven hold effect of flooding on the safety factor against yield should
loading may reduce the magnitude of extreme hull girder be carefully interpreted. Further, the foregoing observa-
loads. Similarly, a good corrosion protection scheme will tions pertain to a particular set of vessels, and thus no
significantly delay the initiation and progress of corro- claims as to their generality can be made. The effect of
sion. All effects of age on structural performance can be vessel size may however bear further investigation.
monitored and controlled by maintenance and ongoing
inspection. With regard to side shell and hatch covers as 10.3 Corrosion Degradation
potential points of water ingress, structural, operational In a number of damage cases, it has been reported
and maintenance measures may be possible as means of that corrosion and fatigue cracking damages had been
improving safety. It also follows that vessels must be suspected in prima~y member regions. In our calculations,
loaded and operated in a manner that is consistent with it has been obserw.~d that the corrosion damage could re-
their design assumptions. In this regard, improved termi- duce ship hull girder ultimate strength as well as the
nal operations have an important role to play. buckling strength of primary members. The corrosion rate
of hull structures i,; influenced by many factors including
10.2 Load Effects due to Accidental Flooding the corrosion protection system and various operational
In an accidental cargo hold flooding condition, the parameters. The corrosion protection systems usually
hull girder loads are in general amplified. When the for- employed are coatings with or without anodes. The op-
ward hold is flooded in laden condition, the change in erational parameters include the percentage of time in
draft and trim forvcard may initially not be dramatic so as ballast, frequency and method of hold cleaning, mainte-
to be readily detected in heavy weather. But here again, it nance and repair, etc.
does take a progression of circumstances (i.e., water in- To assess corrosion damage tolerance a priori, it is
gress, loss of transverse bulkhead) to cause vessel loss. necessary to have an corrosion rate estimate model for

Strength of Bulk Carrier Structures 31


individual member regions, and our studies have contrib- society rules. And the ultimate strength of the corrugation
uted in this regard. We have studied the strength charac- could be adequately predicted by using simple design
teristics of hull structures considering corrosion using formulae using idealized (simply supported) end condi-
such models. Our calculations for the Capesize bulk car- tions such as those recently adopted by IACS, as our
rier show, for example, a 10% or more reduction in ulti- comparisons of such formulae against experimental re-
mate hull girder strength reduction due to corrosion dur- suits confirm.
ing the vessel life. These calculation models need to be We should note that the safety factors required for
improved, however, with regard to the consideration of design against an accidental loading situation are nor-
spatial extents of corrosion in a more realistic manner. mally much less than that required under normal opera-
Currently, uniform corrosion over large structural extents tion, reflecting the rarity of the accidental loading. In
is assumed in such calculations including the ones de- direct calculations related to the strength of bulkheads
scribed here. under accidental (flooding) loads, the end conditions of
the corrugations should thus be carefully accounted for,
10.4 Fatigue Cracking Damage considering the rigidity of adjoining structures at both
In our example calculations, the fatigue of the side ends. For this purpose, the non-linear finite element
shell connection detail forward was checked. The fatigue method will be a useful tool as we demonstrated.
loads for the Capesize were found to be larger than the Through our studies, we have developed and validated
Panamax, due to a combination of difference in local such tools for potential future use.
(side shell) pressures, frame spacing and the length of the
side frame between supports. The effectiveness of the 10.6 Buckling Strength
flange of the end bracket (which is more with an integral Buckling strength of some primary members subject
flange) also made a considerable difference in the fatigue to accidental flooding together with corrosion may some-
damage estimates. The fatigue damage was found to be times be important. In our example calculations, the shear
quite sensitive to the scantlings of the side frame and end buckling strength of the side shell forward was checked in
brackets used and to levels of corrosion. Explicit fatigue way of bulkhead No.l, using IACS unified formulae,
design of the related hull girder details is thus a necessity. although no details are given in this paper. Again, the
safety factors in the flooded condition were greater for
10.5 Corrugated Transverse Bulkhead Strength the Capesize than the Panamax example vessels. The
In the past, consistent with the International Load safety factors corresponding to the 20% corroded case
Line Convention, transverse bulkhead locations in ships were only marginally adequate as may be expected for
were normally selected to prevent margin line immersion such levels of presumed uniform corrosion. Because of
with a cargo hold accidental flooded. Traditionally, the the potential sensitivity of buckling strength estimates to
design rules given by classification societies have used initial deformations, there may exist a future need to
nominal hydrostatic loads in the intact condition as design study the effects of in-service deformations such as those
loads for the bulkhead. As previously noted, there is the due to mechanical unloading processes on buckling
possibility that the collapse of corrugated transverse strength, in order to help clarify the situation.
bulkheads and subsequent progressive flooding may have
been a factor in some of the recent bulk carrier losses. In 10.7 Hull Girder Ultimate Strength
this regard, the enhanced bulkhead requirements recently Due to age and maintenance related factors, e.g., cor-
adopted by IACS should help improve both levels of rosion, fatigue cracks and other structural deterioration,
safety in design and the long-term maintenance of such the residual strength of hull structures may decrease sig-
bulkheads. nificantly. Even if the vessels may be able to initially sur-
Ideally, the strength of the transverse bulkheads in an vive accidental hold flooding, they can eventually foun-
accidentally flooded cargo hold must be sufficient to der by hull girder collapse resulting from increase of ex-
avoid collapse resulting in possible progressive flooding. treme hull girder loads together with decrease of the re-
For bulk carrier safety, the set of design lateral pressure sidual strength, particularly of parts if the structure (e.g.,
load cases for the corrugated transverse bulkheads should the side shell) are lost.
then include the condition that one adjacent compartment Technology is now emerging to assess the ultimate
has been flooded. If the maximum applied bending mo- bending and shear strengths of the hull girder in as-built
ment along the span of corrugation becomes larger than and corroded conditions against the applied loads, both
the ultimate bending moment the corrugated bulkhead intact and flooded. The same technology may make pos-
will presumably collapse. The maximum (extreme) sible the design of structures with "graceful" post ultimate
bending moment acting on the corrugated bulkheads in a strength characteristics. Such considerations imply that
flooded condition would be calculated by classification ultimate strength may be a better basis for structural de-

32 Strength of Bulk Carrier Structures


sign, particularly for an accidental scenario (Ghose et al. creasingly common type of corrosion that is likely to be
1995, Paik, Thayamballi & Che 1996, Paik, Thayamballi found in ship structures in the future) and the spatial ex-
& Yang 1998). It is worth pointing out that hull girder tent of corrosion needs further investigation.
ultimate strength estimates are sensitive to the level of 3) To effectively evaluate the collapse strength reli-
initial deformations, lateral pressure and also to initial ability of corrugated bulkheads in the damaged and
damage (e.g., corrosion and fatigue degradation). Further, flooded condition:z, similar attention is desirable tbr the
better safety estimates may need more accurate prediction ultimate strength of co~Tugated bulkheads as well, consid-
of the loads involved than we are presently able to. Addi- ering both corrosJton and fatigue effects as appropriate.
tional studies related to the various aspects potentially The related dewdopment of procedures and criteria
affecting strength sensitivity and accidental load predic- should take into ',account the potential influence of ad-
tion are then to be recommended. joining structures at the upper and lower ends of the cor-
rugation.
1 I. CONCLUD][NG REMARKS AND FURTHER R & 4) One of the possible scenarios for water ingress
D NEEDS into the cargo hoi[ds is through the failure of the hatch
covers themselves. Due to the dynamic pressure of green
As a consequence of several bulk carrier losses dur- water in heavy weather, hatch cover failure can poten-
ing the last decade, the safety assessment of bulk carrier tially occur, but not unless the water head so applied sig-
structures has recently been of interest. Extensive investi- nificantly exceeds the head representing hatch cover ulti-
gations have thus been carried out with a view to reduc- mate strength. In view of some recent controversy in this
ing such vessel incidents. As a result, most recently, bulk regard, clarificatory investigation of the loads and the
carrier structure requirements have been enhanced by strength aspects of this particular aspect of bulk carrier
IMO and IACS. design would be desirable in the future.
This paper has reviewed and summarized recent re-
search and development in several important areas related In conclusion., we should note that structurally and
to the strength and reliability of bulk carrier structures in operationally, bulk: carriers are more complex than tank-
the intact, damaged and flooded conditions. Certainly, ers, and the dynamic loads imposed on them are more
there still remain various areas in which new or continu- intricate. Previous SafeHull studies identified five struc-
ing additional work is necessary, some of which are as tural areas and conditions unique to existing bulk carriers
follows: that warrant further investigation, namely transverse cor-
rugated bulkheads in cargo holds, vertical hold frames,
I) Fatigue and corrosion are the fundamental deter- cross deck structures, fore body structures, and cargo
minants of age related ship conditions. In aging vessel overloading (Liu 1994). The various efforts reviewed in
structures, corrosion damage and fatigue cracks can exist this paper clearly illustrate that significant progress has
together. Therefore, studies that consider the simultane- been made in all these fronts in the past few years, and
ous influence of both corrosion and fatigue cracking that the related research continues at a desirable pace.
damage on the ultimate strength of hull structures are
desirable. A simplified direct procedure for predicting the ACKNOWLEGMENTS
ultimate strength of ship hull structures considering the
degradation due to corrosion and fatigue cracking damage The authors wish to thank the American Bureau of
is needed for ewduating the strength reliability of aging Shipping and the Pusan National University for support-
vessel structures. Since ship hulls are normally subject to ing this research. "]'he comments of D. Liu, J.S. Spencer,
combined loads, such a procedure must adequately ac- J.F. Conlon and H.H. Chen on this paper are acknowl-
count for them. Due to progress of corrosion and fatigue edged. The use of SafeHull technology in various parts
damage in the structure, the sequence and potential cou- of the study is gratefully acknowledged. The effort of
pling of failure modes may change, which also need to be Sung Geun Kim, a graduate student of Pusan National
taken into consideration. University, for preparing the tables and figures is appreci-
2) Existing corrosion rate estimation models for hull ated. The views expressed in this paper are those of the
structures are useful for the evaluation of time variant authors and not necessarily those of the institutions the
strength reliability of a vessel considering degradation authors are affiliated with.
due to corrosion. However, such models have so far been
developed based on limited number of corrosion correla- REFERENCES
tion parameters. The corrosion models should then be
continuously updated and expanded. In particular, the ABS (1995). Guide for dynamic based design and
treatment of pitting (which one speculates to be an in- evaluation of bulk carrier structures, American Bureau of

Strength of Bulk Carrier Structures 33


Shipping, SafeHull Project, March. Hattori, K., Sakato, T., Iwahashi, Y. and Mikami, K.
(1984). A consideration of the phase difference between
ABS (1997). Rules for building and classing steel vessels, the wave induced stresses for longitudinal and transverse
American Bureau of Shipping. Strength, J. of the Society of Naval Architects of Japan,
Vol. 156, pp. 273-282 (in Japanese).
Akita, Y. (1984). Reliability analysis in strength of ships
(lst report), J. of the Society of Naval Architects of Ja- HMSO (1995). Lord Donaldson's Assessment
pan, Vol. 155, pp. 207-214 (in Japanese). (Derbyshire), Her Majesty's Stationary Office, London,
December.
Brown, D. (1997). Rapid failure, rapid sinking, Proc. of
International Conference on Design and Operation for Ivanov, L.D. (1993). Bulk carriers- Brief analysis of
Abnormal Conditions, RINA, Glasgow, October. structural failures based on department data files, ABS R
& D Department Internal Report, American Bureau of
BTCE (1993). Structural failure of large bulk ships, Bu- Shipping, New York, November.
reau of Transport and Communication Economics, Report
85, Australian Government Publishing Service, Canberra, Ivanov, L.D. (1994). Bulk carriers- Strength criteria for
December. corrugated bulkheads, ABS Internal Document, American
Bureau of Shipping, New York, July.
Byme, D. and Evans, J. (1998). Hatch covers- Failures on
laden bulk carriers in heavy weather with reference to Kawamura, A., Hashimoto, K., Inoue, S., Kuramoto, Y.
current design practice, Proc. of International Conference and Toki, N. (1986). Ship structure design based on syn-
on Design and Operation of Bulk Carriers, Paper No. 12, thetic evaluation of wave loads (1 st report), J. of the So-
RINA, London, 30 April & 1 May. ciety of Naval Architects of Japan, Vol. 160, pp. 246-258
(in Japanese).
Caldwell, J.B. (1955). The strength of corrugated plating
for ships' bulkheads, Trans. RINA, Vol. 97, pp. 495-522. Kawamura, A., Hashimoto, K., Inoue, S. and Kuramoto,
Y. (1987). Ship structure design based on synthetic
Clarkson Research (1996), Shipping review and outlook, evaluation of wave loads (2nd report), J. of the Society of
Spring. Naval Architects of Japan, Vol. 161, pp. 242-249 (in
Japanese).
Cramer, E.H., Loseth, R., Olaisen, K. and Valsgard, S.
(1995). Fatigue design of ship structures, Proc. of the 6th Kondo, Y. (1987). Prediction method for corrosion fa-
International Symposium on Practical Design of Ships tigue crack initiation life based on a corrosion pit growth
and Mobile Units (PRADS'95), Vol. 2, Seoul, September, mechanism, Trans. of the Japan Society of Mechanical
pp. 2.898-909. Engineers, Vol. 53, No. 495, pp. 1983-1987 (in Japa-
nese).
Emi, H., Yuasa, M., Kumano, A., Yamamoto, N., Arima,
T. and Umino, M. (1993). A study on life assessment of Liu, D. (1994). Advanced technology enhances ship
ships and offshore structures (3rd report): Corrosion con- safety, ABS Annual Report, pp. 18-22.
trol and condition evaluation for a long life service of the
ship, J. of the Society of Naval Architects of Japan, Vol. Liu, D., Jan, H.Y., Chen, H.H., Scotto, F.J. and Akiyama,
174, pp. 735-747 (in Japanese). A. (1995). Recent development of design criteria for hull
structures of bulk carriers, Proc. of the 6th International
Faulkner, D., Corlett, B.J. and Romeling, J.U. (1996). Symposium on Practical Design of Ships and Mobile
Design for hatch covers and coamings for abnormal Units (PRADS'95), Vol. 2, Seoul, pp. 2.883-2.897.
waves, Proc. of International Conference on Watertight
Integrity and Survivability, RINA, London. Liu, D. and Thayamballi, A.K. (1997). The durability of
ships considering fatigue cracking. J. of Ship and Ocean
Frystock, K. and Spencer, J. (1996). Bulk carrier safety, Technology, Vol. 1, No. 1, The Society of Naval Archi-
Marine Structures, Vol. 33, No. 4, October, pp. 309-318. tects of Korea, October, pp. 57-72.

Ghose, D.J., Nappi, N.S. and Wiernicki, C.J. (1995). Re- Loukakis, T.A. and Chryssostomidis, C. (1975).
sidual strength of damaged marine structures, Ship Seakeeping series for cruiser stem ships, Trans. SNAME,
Structure Committee, Report No. SSC-381. Vol. 83, pp. 67-127.

34 Strength of Bulk Carrier Structures


LR (1997). World casualty statistics, Lloyd's Register. of corrugated bulkheads, Proc. of the 7th International
Symposium on Practical Design of Ships and Mobile
Mano, H., Nishihara, T. and Kawabe, H. (1976). A study Units (PRADS'98), The Hague, September.
on the wave loads of bottom structure under the influence
of cargo, J. of the Society of Naval Architects of Japan, Paik, J,K., Thayamballi, A.K. and Chun, M.S. (1997).
Vol. 140, pp. 157-164 (in Japanese). Theoretical and experimental study on the ultimate
strength of corrugated bulkheads, J. of Ship Research,
Nitta, A., Arai, H. and Magaino, A. (1992). Basis of Vol. 41, No. 4, pp. 301-317.
IACS unified longitudinal strength standard, Marine
Structures, Vol. 5, pp. 1-21. Paik, J.K., Kim, S.K., Yang, S.H. and Thayamballi, A.K.
(1998). Ultimate ,;trength reliability of corroded ship
Paik, J.K. (1994). Hull collapse of an aging bulk carrier hulls, Trans. RINA, Vol. 140, pp. 1-18.
under combined longitudinal bending and shearing force,
Trans. RINA, Vol. 136, pp. 217-228. Paik, J.K., Kim, S.K. and Lee, S.K. (1998). A probabilis-
tic corrosion rate estimation model for longitudinal
Paik, J.K. and Kim, C.Y. (1989). A simplified finite ele- strength members .of bulk carriers, Ocean Engineering,
ment method for the ultimate strength analysis of plates Vol. 25, No. 10, pp. 837-860.
with initial imperfections, J. of the Society of Naval Ar-
chitects of Korea, Vol. 26, No. 1, pp. 24-38. Paik, J.K., Thayamballi, A.K., Kim, S.K. and Yang, S.H.
(1998). Ship hull ultimate strength reliability considering
Paik, J.K. and Mansour, A.E. (1995). A simple formula- corrosion, J. of Ship Research, Vol. 42, No. 2, pp. 114-
tion for predicting the ultimate strength of ships, J. of 125.
Marine Science and Technology, The Society of Naval
Architects of Japan, Vol. 1, No. 1, December, pp. 52-62. Paik, J.K., Thayamballi, A.K. and Yang, S.H. (1998).
Residual strength assessment of ships after collision and
Paik, J.K., Thayamballi, A.K. and Che, J.S. (1996). Ulti- grounding, Marine Technology, Vol. 35, No. 1, pp. 38-
mate strength of ship hulls under combined vertical 54.
bending, horizontal bending, and shearing forces, Trans.
SNAME, Vol. 104, pp. 31-59. RINA (1996). Derbyshire- The search, assessment and
survey, The Naval Architect, The Royal Institution of
Paik, J.K. and Thayamballi, A.K. (1997a). An empirical Naval Architects, May, pp. 44-48.
formulation for predicting the ultimate compressive
strength of stiffened panels, Proc. of the Seventh Interna- Shama, M.A. (1995). Ship casualties- Types, causes and
tional Offshore and Polar Engineering Conference environmental impacts, Alexandria Engineering Journal,
(ISOPE'97), Vol. IV, Honolulu, May, pp.328-338. University of Alexandria, Alexandria, Vol. 34, No. 2.

Paik, J.K. arid Thayamballi, A.K. (1997b). Turan, O. and Vassalos, D. (1994). Dynamic stability
CORBHD/FEM: A special purpose nonlinear finite ele- assessment of damaged passenger ships, Trans. RINA,
ment computer program for the ultimate strength analysis Vol. 136, pp. 7%104.
of corrugated bulkheads, Report of Joint PNU-ABS Re-
search Project on Ultimate Strength of Corrugated Bulk- Yamamoto, N. and Ikegami, K. (1996). A study on the
heads, Pusan National University Report, May, degradation of coating and corrosion of ship's hull based
on the probabilistic approach, Proc. of the Offshore Me-
Paik, J.K., Thayamballi, A.K. and Kim, S.G. (1998). The chanics and Arctic Engineering Symposium (OMAE'96),
influence of adjoining structures on the ultimate strength Vol. II, pp. 159-166.

Strength of Bulk Carrier Structures 35


Discussion
Douglas Faulkner, Member It has therefore been suggested recently that additional ex-
treme survival conditions need to be defined and applied for the
The question of bulk carrier safety is so important that it
safe design and operation of all ships (Faulkner and Williams
does seem an opportunity has been lost by the organizers in
1996, Faulkner and Buckley 1997). The need for more refined
arranging for Papers 1 and 3 to be presented simultaneously in
satellite measurements of wind and wave have very recently
different rooms and not sequentially; it is for this reason that
been defined in order to provide better data on wave profiles
I (and no doubt others) am not able to personally read this con-
and crest velocities, and in particular for long-crested wave
tribution and compliment Prof. Paik and Dr. Thayamballi. Their
fronts (Faulkner 1998b).
excellent paper continues the very welcome sequence of their
It is stressed that near-breaking wave conditions occur rea-
recent publications supported by the American Bureau of Ship-
sonably frequently in such conditions, and it may not be widely
ping.
appreciated that crest velocities can exceed the normal wave
My own discussion attempts to redirect attention away from
celerity defined by ),/T by 2 to 4 times. Green Sea impact pres-
what may now be called conventional analyses such as those
sures are then much higher (~2 effect) and this is certainly not
presently employed by class societies toward the response to
considered in the present design of hatch coamings, bridge
the highly nonlinear wind and wave systems which actually sink
fronts, and other vertical structures. Again, casualty data show
most ships, that is, toward the real sea. Leaving aside human
that substantial damage does occur in such structures, often 10
errors in the navigation and operation of ships, there are three
m to 20 m and sometimes up to 30 m above the MWL.
basic hazards arising from rough weather which can sink a ship:
These remarks in no way detract from the value of this ex-
(a) breaking their backs through inadequate strength cellent paper. They are offered to suggest that we are not at pre-
(b) flooding and foundering through poor watertight in- sent designing for conditions which actually do sink ships, and
tegrity that universities and the research departments of ship classifi-
(c) capsize due to insufficient roll stability cation societies as represented by the authors should be exam-
ining survival design as a matter of urgency. Arguing that
These weaknesses are mainly design- and sometimes con-
weather routing ensures that such storms are avoided is naive
struction-related. But losses may also be induced by machin-
in the extreme. With the exception of car carriers, such prac-
ery or equipment failures which often cause the ship to be-
tices are driven by economy and not by ship safety, as any ship-
come beam on to the prevailing weather, thus facilitating loss
master would tell you.
scenarios (b) and/or (c).
Back breaking seldom occurs at sea and yet scenario (a)
still occupies much of our attention but with wave loads equiv-
Additional references
alent to no more than 10.75 m wave height. Once RO/RO fer-
ries lose their watertight integrity they can then be very vul- Faulkner, D. (1998a). An independentassessmentof the sinkingof the
nerable to capsize, as the traveling public now realize. There M.V. Derbyshire, Transactions, SNAME, Vol. 106.
has been a spate of bulk carrier losses since about 1980. These Faulkner, D. (1998b). Definingthe ocean environmentfor the safe de-
ships are generally very stable, so they are mostly lost through sign and operation of ships, Keynote Address, COST Conference, Provi-
flooding scenario (b). The most vulnerable structure to col- sion and Engineering/OperationalApplicationsof Ocean Wave Spectra,
lapse and therefore flood is the side plating and stiffening in Paris, September 21-25.
older single-skin ships and hatch covers and coamings in all Faulkner, D. and Buckley, W.H. (1997). Critical survival conditions
bulk carriers. for ship design, Proceedings, RINA InternationalConference,Design and
Regarding hatch covers, they can occupy 30% or so of the Operation for AbnormalConditions,Glasgow, October 21-22.
upper deck area, and yet their strength to withstand green sea Faulkner, D. and Williams, R.A. (1996). Design for abnormal ocean
actions is usually no more than 10% of the capability of the waves, RINA SpringMeetings (with full discussionin RINA Transactions,
rest of the deck. This anomaly must surely be put right, espe- PartA, Vol. 139, 1997).
cially as casualty data and theory (Faulkner, Corlett and Romel- Williams, R.A. and Torchio, R. (1998). M.V. Derbyshire surveys,
ing, 1996) support the gross inadequacy of the present hatch UK/EC Assessors' Report, DETR EnvironmentTransport Regions, March
cover rules. Estimates show that 30 or more seamen serving 1988 (ISBN 1-85112-286-0).
in bulk carriers lose their lives every year through this cause
alone. And now we have confirmation that this was the fun-
damental weakness that led to the loss without trace of the Cape- John B. Caldwell, Member
sized bulker Derbyshire in 1980 (Williams and Torchio 1998,
Many naval architects have often believed that the design
Faulkner 1998a).
of watertight transverse bulkheads was one of the more straight-
The northeast seaboard of North America and Canada is re-
forward aspects of ship structural design. It is now clear that
ferred to as the graveyard of the Atlantic, and the Northwest
this is not the case; and their correct design has now been rec-
Pacific between Japan and Papua, New Guinea, is referred to
ognized as a crucial factor in ensuring satisfactory standards
as the graveyard of the Pacific; the latter accounts for nearly
of safety for ships. Hence the authors have performed a valu-
30% of all shipping losses;The common denominators for these
able service in giving prominence to this feature of their ex-
regions are the occurrence o~ extreme revolving tropical storms
tensive research into bulk carrier strength and reliability. Also
(hurricanes and typhoons) and the occurrence of opposing cur-
by focusing on the ultimate strength properties of corrugated
rent and wave systems. Extreme temperature changes also oc-
bulkheads, their work will help towards a more rational for-
cur on the Northeast wall of the Gulf stream. All of these con-
mulation of bulkhead design rules and procedures, properly
ditions lead to steep elevated waves whose:
based on limit state, rather than limiting stress design.
9 height H, often exceeds 25 m The authors kindly refer to this writer's work on this topic
9 crest elevations can be 0.6 H to 0.7 H carried out more than 45 years ago, and the tests described in
9 mean crest front slopes approach the wave breaking limit the paper form a valuable extension of that earlier work. For
of 0.6 vertically corrugated bulkheads on which the predominant load-

36 Strength of Bulk Carrier Structures


ing is lateral hydrostatic pressure, there appears to be four limit Bulk carriers are: the workhorses of the seas and are subject
states to be considered in developing a design: to quite rough handling. The day-to-day wear and tear result-
1. Overall failure due to the troughs collapsing as beams ing from rough handling, such as grab unloading etc., will grad-
bending between upper and lower supports. Hinges may form ually take its toll on the vessels. Corrosion and fatigue play
due to plasticity combined with buckling of the compressed important roles in the aging process as well. Hence, with a
parts of the corru~,ation; normal "maintenance by breakdown" philosophy, age becomes
2. Shear failure of corrugations near upper and lower sup- a parameter. This is illustrated by statistical data for the IACS
ports; fleet of vessels as shown in Fig. 27 (Ekenes et al 1996). This
3. Premature failure due to local buckling of the compressed study is based on casualty data relating to heavy weather, heavy
parts of flange and web, leading rapidly to loss of trough bend- cargo and age.
ing strength; and Traditionally, design codes as Classification rules implic-
4. Local failure of flange plating (or web, if that is wider) itly use the concept of Limit States. Buckling and yielding are
under the applied pressure, leading to 3-hinge bending failure examples of extreme loads which are usually considered to have
of these elements of the troughs. a return period of 20 years, and are defined as Ultimate Limit
Mode 2 above is only likely to be of concern in short span State (ULS) conditions. Fatigue evaluation, on the other hand,
bulkheads under very high lateral pressures. Modes 1 and 3 is defined as a Fatigue Limit State (FLS) condition and is as-
are discussed quite fully in the paper. However, rather litlle is sociated with day-to-day loads. These are very useful concepts,
said about mode 4; and, in the writer's view, this can have an but should not always be looked at in isolation as interactions
important influence in the design of corrugated panels. may occur. As an example, the loss of the bow of Neptune
It can be shown that mode 4 is likely to be more important Sapphire off South Africa in the 1970s occurred because of a
than mode 3 if the design ultimate pressure exceeds approxi- combination of buckling of the deck due to excessive bow flare
mately 0.6x yield stress squared divided by Young's modulus; loads leading to tension overloading in the bottom which made
and this will generally be the case for large bulk carriers. This the whole bow fall off.
possible limit state tends to restrict the width of flange and web, It may well happen that a failure starts as a fatigue crack.
and encourages the sue of designs in which these widths are When subjected to an extreme overload a large-scale rupture
roughly equal, leading to trough forms which are more likely may occur. This is one possible initiating event for accidental
to be trapezoidal than rectangular. Indeed, by carrying out an flooding of bulk carrier holds, e.g., the side shell cracks due to
early kind of design optimization using the above considera- fatigue leading to water ingress of the hold. If the transverse
lions the writer showed, in Caldwell (1961), that a design with bulkhead(s) does not stand up to the increased pressure the ad-
a trough angle of .about 58 degrees and web width 1.11 times jacent hold(s) is filled, and the ship either sinks due to lack of
flange width could be weight-optimal. Hence the authors' view floatability and/or most probably breaks her back due to over-
that rectangular troughs should be preferred to trapezoidal loading. Another initiating event may be hatch cover failure due
troughs seems open to question. Certainly the comments in Sec- to shipping of green seas.
tion 6 of the paper about the relative strengths of these alter-
native forms need to be debated. With modern optimization
techniques, the design problem of selecting optimum values 1.0E-01
of trough dimensions and pitch (which is not the number, but
the spacing of troughs) could perhaps be formalized so that more lral
definitive guidance: could be given to designers. Perhaps the au-
thors could clarify their opinions on this matter? S-
1.0E-02
s__
Additional reference
C a l d w e l l , J.B. (1961). N o t e s on the structural d e s i g n o f w e l d e d ships, =
Proceedings, Symposium on Welding in Shipbuilding, Institute of Weld-
ing, London, pp. 14-22. -~, 1.0E-03
O

Sverre Valsg&rd, Member L,, ~_

The authors are', to be commended for a most comprehen- I,d


sive paper covering a lot of ground and containing a wealth of
useful information.
g i 0E-04
The basic structural design of bulk carriers has been kept
~r
unchanged for quite some time. Although some double-side de-
signs have been built, the dominating design is with transverse
framed single-skin construction between the hopper and top
wing tanks. The structure is rather flexible with no longitudi- 1.0E-05
nal bulkheads supporting the double botton~. Tl+.e same basic
configuration is used for all sizes of vessels.
Further, design optimization has in most cases taken away
every second web frame in the top wing tanks leading to spans
of the deck strips of up to 6 m. Keeping in mind that the deck I.OE-06 .
strips act more or less as single, deck panels, the inherent re- I I0 100
dundancy of these parts of the vessel is much less than in the
double bottom structure. Hence, deck strips may have a quite Nuni~r of fatalities, N
marginal compressive strength, making the deck construction
sensitive to sagging overloads as well. Fig. 27 Risk diagram for bulk carrier losses

Strength of Bulk Carrier Structures 37


I support the authors' view that the integrated hull bending
moments and shear force capacities are better suited for over-
all collapse strength assessment than the strength of single struc-
tural elements in the cross section and therefore should be used
for intact as well as accidental strength check scenarios. In
Andreassen et al (1998) ultimate hull girder capacity as well
as local strength of bulk carrier hull girders are examined. For
given coating qualities and maintenance strategies future steel
replacement and associated Net Present Value (NPV) of repair
costs are calculated using a probabilistic analysis process (us-
ing the Nauticus Life Cycle Cost (LCC) program with Monte
Carlo simulation) accounting for the uncertainties in the input
parameters.
The same corrosion model as in the present paper was used
in Kim (1997). It is shown that if damaged coating is properly
maintained, (or a very good coating quality is used) and steel
renewal is carried out as required due to local requirements,
the reduction in hull girder moment capacity up to 30 years of
Fig. 28 Capesize bulk carrier in irregular seas. vessel age will be on the order of 7 to 10%. Moreover, if the
Nonlinear SWAN simulation "missing" deck frames are put back in place the Hull Girder
Capacity in sagging will increase up to about 20%. This is a
quite substantial improvement and provides a better balance be-
The paper discusses the above aspects and offers a quite tween Hull Girder Capacities in sagging and hogging.
interesting case study of the effect of accidental flooding on
two conventional single-hull bulk carrier designs--a Capesize
and a Panamax vessel. The still water effect of forward hold Additional references
flooding and mid-hold flooding is studied showing that the
larger Capesize vessel is more affected by hold flooding than Andreassen, E., Kim, S-K. and Valsg~d, S. (1998). Developmentof
the smaller Panamax vessel. The still water loads increase with ultimatehull girdercapacity,Proceedings, InternationalConferenceon De-
up to 60% for shear and 95% for bending moments for the Cape- sign and Operation of Bulk Carriers, April 30-May 1, London.
size and 40% and 70% for the Panamax, respectively. These Ekenes, M.L., Astrup, O.C., Ronald, K. and Gran, S. (1996). Statisti-
are quite substantial increases which combined with the wave cal data for bulk carrier casualties where structural failure may have been
loads may easily exceed the extreme design load values. a factor, Det Norske Veritas, Report no. 96-2042.
The authors have apparently, like many others, faced the Kim, S-K. (1997). Ultimatestrength reliabilityassessment of corroded
problem that the effect of hold flooding on the wave loads could ship hulls, Departmentof Naval Architectureand Ocean Engineering,Pu-
not be properly calculated with linear hydrodynamic wave load san National University,Pusan, Korea, June.
programs, and have therefore assumed the wave loads in the Kring, D., Sclavounos,P., Vada, T. and Braathen, A. (1996). Nonlin-
flooded condition to be equal to the wave loads in the intact ear ship motionsand wave-inducedloads by a Rankinemethod, Proceed-
condition. This may not necessarily be correct as the flooded ings, 21st Symposiumon Naval Hydrodynamics,Trondheim.
hold(s) will significantly change the ship's mass distribution,
trim and draft. This, however, calls for a new generation of non-
linear hydrodynamics programs capable of taking the relevant Robert J. vom Saal, Member
physical parameters into account. One such possibility is the The authors are to be commended for this significant con-
SWAN program, (Kring et al 1996). Figure 28 shows a snap- tribution to our understanding of the causes and cures for the
shot from a nonlinear computer simulation of a Capesize bulk recent rash of bulk carrier failures. One of the frustrations of
carrier in irregular seas. This type of computer tool holds the ship design profession is the inability to determine the root
promise for future, more precise investigations. cause of the most interesting accidents, and this paper shows
The authors present an interesting experimental and theoret- how many different scenarios must be considered.
ical investigation of the strength of transverse corrugated bulk- There are some areas of the paper where some clarification
heads. In the numerical analysis a limiting value of plastic rup- appears desirable. First, Table 2 on page 1-4 gives the average
ture strain of 15% is included in the nonlinear FEM model. I age of various classes of bulk carriers at loss. This statistic is
strongly support the physical thinking behind such an approach, only meaningful in comparison to the average age of the total
the idea being that plastic "hinges" will develop in the corruga- fleet of each class, i.e., was the average age of lost ships greater
tions as a combined action of short wave buckling in the com- or less than the average age of ships not lost?
pression flange followed by plastic rupture due to high local Second, Figs. 9.b and 9.c on page 1-9 show no apparent re-
strains on the tension side. This is actually the same mechanism lationship between corrugation geometry and ship length. Cor-
(chain of events) as with Neptune Sapphire when she lost her rugation geometry generally seems more closely related to cor-
bow ,~.smentioned earlier. This kind of "hinge" failure will hap- ruga-,'ion length than to ship length. Could the authors rework
pen at the lower bulkhead stool leading to water ingress in the these figures to plot the corrugation geometry against corruga-
adjacent hold. This is probably why the IACS formula only al- tion length rather than ship length?
lows for 50% of the nominal bending capacity at the lower stool. Third, the paragraph on page 1-23 discussing Fig. 19 seems
Due to weld imperfections tension rupture failure invari- to be saying that neglecting the bottom shell pressure loads is
ably start at welds in welded steel structures. For normal bulk- nonconservative because the seawater pressure reduces end fix-
head thicknesses scale effects should probably lead to a lower ity. In fact, Fig. 13 (on page 1-18) and 17 (on page 1-21) im-
plastic rupture strain than the 15% used here. Admittedly, this ply that the inner bottom girder loads have the effect of in-
to a certain extent depends on how the limiting plastic strain creasing the end fixity moment, and could even over-fix the end
criterion is implemented in the FEM code. if the holds are long and depth is small. Could the authors clar-

38 Strength of Bulk Carrier Structures


ify this point with some sample deflection plots from their fi- to consider rupture as a potential failure mode. In fact, the non-
nite-element analysis? linear finite-element calculations of the present paper did in-
Finally, page 1-30 notes that the fatigue life of the side shell clude a rupture criterion. We would agree with Dr. Valsg~rd
brackets for the Capesize bulk carrier is much less than that of that further research is desirable as to the value of the rupture
the Panamax, though all other safety margins are higher for strain to be used in the structures of the type we are dealing
the Capsize than for the Panamax. Page 1-31 notes that the Cap- with. Dr. Valsgfird also notes that we in our calculations as-
size "is all HT36 steel" while the Panamax "is all mild steel," sumed, perhaps unnecessarily, that the wave-induced loads in
implying that the Capsize has proportionately reduced scant- the condition with a hold flooded could be estimated using de-
lings. The relatiw~ fatigue strengths of the two materials are sign load formulae which are indeed based on an intact condi-
not mentioned anywhere in section 9.3, but are usually con- tion. We did this not only to simplify matters, but also because
sidered nearly identical. Could the authors clarify whether in our calculations indicated the draft and trim changes with a
fact HT36 is considered identical to mild steel in fatigue strength single hold flooded were not overly significant. In principle, the
and whether this is the primary reason for the relative lack of new generation of nonlinear ship motion programs of the type
fatigue strength in the Capsize bulker? Dr. Valsgfrd alludeA to may also be used for the same pur-
Again, I would like to thank the authors for their contribu- pose. At the presenl: time, the computational resources needed
tion to furthering our understanding of the intricacies of bulk to do this remain significant, which is a potential drawback to
carrier structural design and the many environmental influences the routine use of such technology. For example, in our expe-
which must be considered. rience, a full nonlinear time domain simulation can require a
computational effort in the order of ten times the real time.
We are grateful for the many incisive and practical com-
Authors' Closure ments of Mr. vom Saal. We agree with him that, in the con-
To start with, we thank all the discussers for their kind re- text of retrospective', analysis, one can never know for certain
marks regarding our paper. We thank them even more for the what caused a particular vessel loss. The difficulty in estab-
time they spent developing their various discussions. lishing the possible ,causes of a vessel loss is even greater when
We ourselves were disappointed at not being able to attend the vessel and crew are lost at sea without warning, and the
the presentation of Prof. Faulkner's own paper on the Der- physical evidence is not recoverable. Regarding the average age
byshire. Prof. Faulkner has been in the forefront of the recent of vessels lost as quoted in Table 2 of the paper, this is sim-
discussion related to hatch cover strength and also the design ply the sum of the ages of all vessels lost divided by the num-
of oceangoing ships for what he terms additional extreme sur- ber of vessels lost. We do agree that this statistic needs to be
vival conditions in his discussion. Upon reflection, it occurs to interpreted with full consideration of the age profile of the fleet
the authors that Prof. Faulkner was also one of the first to com- in general.
prehensively investigate and to champion other important is- Although specific information is not provided in the paper,
sues in the past such as design for compressive strength, re- the world bulk carrier fleet consists of vessels of all ages as
ducing skewness in strength models and the benefits of one might imagine. Further to another suggestion of Mr. yore
reliability-based design, which our profession has now come Saal, we have replctted Figs. 9.b and 9.c to judge the depen-
to generally accept. One can readily agree with Prof. Faulkner dence of corrugation geometry against corrugation length rather
about the need for research into hatch cover strength and also than ship length, see Figs. 29 and 30 of this authors' closure.
structural design for survival conditions. In both cases, the loads It is seen that distribution of the ratio of corrugation flange to
remain the single largest unknown. Additionally, when de- web breadth for conventional bulk carrier corrugated bulkheads
signing for survival conditions, the availability of technology against ship length has a trend that is similar to that against
required to predict structural response well beyond the linear corrugation length, but the distribution of the corrugation sec-
elastic regime may also be of some concern. tion modulus against corrugation length becomes even more
We share Prof Caldwell's hope that the profession will come
to regard our own corrugated bulkhead structural model tests
as a useful complement to his pioneering earlier work. Prof. 20
Caldwell takes us to task for not having considered a particu-
lar failure mode pertinent to corrugated bulkhead design, namely
the local failure of flange plating (or web, if that is wider) un-
der the applied pressure, leading to 3-hinge bending failure of
such elements. He also notes that, had we considered this fail-
ure mode as well, we might have made different observations ge
regarding the optimal shape for bulkhead corrugations. Our
explanation in thi:~ regard is that, to the extent we could as-
certain from our structural model tests and nonlinear finite-
~ ~ 1.0 00 00(110 6) 0 0

O@ 0
element analysis, the failure mode that he draws our attention 0
to was apparently not important. One needs to bear in mind, 0
however, that, we ]nave in this work dealt with corrugated bulk-
head cross sections of certain ranges of structural characteris-
tics typical to the conventional bulk carrier structures that we
considered. Thus our study and the observations resulting from
it, including those related to the optimal corrugation shape are 0.0 I I [ I 9

not completely gelaeral. Prof. Caldwell is undoubtedly correct 80oo 1o~o ~=ooo ~4ooo leOOo lSOOO
in his observations regarding the behavior of corrugated shapes Corrugation Length (mm)
in general, and we do thank him for pointing out a potential
misinterpretation of the results of our study. Fig. 29 The relationship between corrugation length and
It is always a pleasure to have Dr. Valsgg~rd share his wide the ratio of corrugation flange to web breadth for
and varied experience with us. We agree with him on the need conventional bulk carrier corrugated bulkheads

Strength of Sulk Carrier Structures 39


3.0- scattered than the corresponding one against ship length. Mr.
0 vom Saal correctly takes us to task for our not having dis-
0 cussed in detail the effect of differences in material yield
0 strength in the context of fatigue. Regarding fatigue capacity,
0 0
0 there indeed is typically little difference in the fatigue strengths
S~ 2.0
0
of welded structural details in the range of yield strength that
0 a naval architect typically deals with, namely 32 to 55 ksi.
This is because the fatigue performance of such details is dri-
1.5 0
m 0 ven largely by the very local high stress concentration at the
C 0
O weld, the effect of which is significantly greater than the ef-
8 c~
t~ 1.0 fect of yield strength differences themselves. There indeed is
bid
a difference in operating cyclic stresses between the Capsize
o and the Panamax because of material yield strength differences,
O
C.~ 0.5 the effect of which we perhaps should have discussed in greater
detail than we did in the paper. But equally important is the
0.0
fact that even for identical structural details in the two cases,
I I I I I
8o0o ~oooo ~2ooo ~4ooo ~6ooo ,8ooo
the fatigue loading in the Capsize side shell structure is sig-
nificantly greater than the Panamax for the various reasons we
Corrugation Length (mm)
noted in the paper.
Fig. 30 The relationship between corrugation length and In closing, the authors are indeed pleased to have the ben-
the corrugation section modulus for conventional bulk efit of the high quality discussions to their paper, which sig-
carrier corrugated bulkheads nificantly add value to the paper, serve to emphasize various
aspects of the bulk carrier structural safety issue, and also high-
light the need for continued research and development effort
in related areas.

40 Strength of Bulk Carrier Structures

Potrebbero piacerti anche