Effect of Pitting Corrosion On Local Strength of Hold Frames of Bulkcarriers (1st Report)

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 30

ARTICLE IN PRESS

Marine Structures 17 (2004) 403–432


www.elsevier.com/locate/marstruc

Technical note

Effect of pitting corrosion on local strength of


hold frames of bulk carriers (1st report)
Tatsuro Nakaia,, Hisao Matsushitaa, Norio Yamamotoa,
Hironori Araib
a
Research Institute, Nippon Kaiji Kyokai (ClassNK), 1-8-3, Ohnodai, Midori-ku, Chiba, 267-0056, Japan
b
Training Center, Nippon Kaiji Kyokai (ClassNK), 1-8-5, Ohnodai, Midori-ku, Chiba, 267-0056, Japan
Received 29 April 2004; received in revised form 6 October 2004; accepted 6 October 2004

Abstract

Firstly, pitting corrosion observed on hold frames in way of cargo holds of bulk carriers
which exclusively carry coal and iron ore has been investigated in detail. It was shown that the
shape of the corrosion pits observed on them is a circular cone and the ratio of the diameter to
the depth is in the range between 8 to 1 and 10 to 1, which is different from the trend observed
for the bottom shell of the oil tanker where the ratio is in the range between 4 to 1 and 6 to1.
Secondly, a series of tensile tests has been conducted to investigate the effect of pitting
corrosion on tensile strength of members. It was pointed out that the tensile strength decreases
gradually and the total elongation decreases drastically with the increase of the thickness loss
due to pitting corrosion. The reduction of the tensile strength of the members with pitting
corrosion is larger than that of members with uniform thickness loss in terms of average
thickness loss.
Thirdly, a series of compressive buckling tests has been performed to examine the influence
of pitting corrosion on buckling behavior of members. It was found that compressive buckling
strength of pitted members is smaller than or equal to that of members with uniform thickness
loss in terms of average thickness loss.
Finally, an elasto-plastic analysis by FEM has been carried out to simulate the compressive
buckling test in order to validate the method of modeling members with pitting corrosion.

Corresponding author. Tel.: +81 43 294 5894; fax: +81 43 294 5896.
E-mail address: nakai@classnk.or.jp (T. Nakai).

0951-8339/$ - see front matter r 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.marstruc.2004.10.001
ARTICLE IN PRESS
404 T. Nakai et al. / Marine Structures 17 (2004) 403–432

An attempt has been made to simulate the compressive buckling behavior of pitted members
using shell elements of which meshing size is almost the same as the original thickness of the
pitted plate.
r 2004 Elsevier Ltd. All rights reserved.

Keywords: Bulk carrier; Hold frame; Pitting corrosion; Local strength; Static strength; Buckling strength

1. Introduction

Hull structural members are subjected to corrosive environment after com-


missioning and ageing effect such as large diminution of plate thickness due to
corrosion is unavoidable. Hence, corrosion is said to be one of the dominant life-
limiting factors of ships. In order to secure the safety of hull structures, it is very
important to understand the corrosion process and the effect of corrosion wastage
not only on overall strength but also on local strength should be accurately
evaluated.
From such a point of view, one of the authors has developed a probabilistic
corrosion model [1] which could evaluate the generation and progress of corrosion
quantitatively. Furthermore, diminution behavior of hull girder longitudinal
strength, which is the primary strength for ship structures, has been examined
taking into account the effect of corrosion by the probabilistic model [2]. Corrosion
observed in ship structure is divided broadly into two types, namely, general
corrosion and localized corrosion. For example, localized corrosion is observed on
hold frames in way of cargo holds of bulk carriers which have coating such as tar
epoxy paints, while general corrosion is the problem when they have no protective
coating. Generally, pitting corrosion is defined as an extremely localized
corrosive attack and sites of the corrosive attack are relatively small compared to
the overall exposed surface [3]. In the case of localized corrosion observed on hold
frames of bulk carriers, the sites of the corrosive attack, that is, pits are relatively
large (up to about 50 mm in diameter). In the present study, such localized
corrosion is dealt with and the term ‘pitting corrosion’ is also used for this kind of
localized corrosion. When the effect of corrosion on local strength is considered,
pitting corrosion is a great concern. However, the effect of pitting corrosion on
strength is not addressed in the previous studies. In the present study, emphasis is put
on pitting corrosion. As a first step towards evaluating the effect of pitting corrosion,
a series of tensile tests and compression tests has been conducted with actual
corroded members and artificially pitted members following the investigation of
actual pitting corrosion on hold frames of bulk carriers [4]. Furthermore, an attempt
has been also made to simulate the buckling behavior observed in the compression
test by FE-analyses. The main object of the present paper is to examine how
pitting corrosion affects basic mechanical properties of structural members such as
tensile strength and compressive bucking behavior and to explore the method of
simulating deformation behavior of members with pitting corrosion by FE-analysis
using shell elements.
ARTICLE IN PRESS
T. Nakai et al. / Marine Structures 17 (2004) 403–432 405

2. Actual corrosion of hold frames of bulk carrier carrying coal and iron ore

2.1. Corrosion of hold frames

Fig. 1 shows an example of actual corrosion observed on web plates of hold


frames of a 14-year-old bulk carrier which carries iron ore and coal. These hold
frames were coated by tar epoxy paints at the construction. Hold frames of a bulk
carrier are shown schematically in Fig. 2. It is seen that progress of corrosion is
severer at the lower and middle parts of the hold frames than at the upper part.
Many rust blisters were found on the web plates (Figs 1(a) and 1(b)). Fig. 1(c) shows
the member after sandblasted. It can be seen that pitting corrosion prevails all over
the surface of the member. Enlargement of the corrosion pit and the rust blister is
included in Fig. 3. The rust swelled up at the corrosion pit and it seems as if there
were no thickness loss, but when the heavy rust is removed, large thickness loss is
perceived. Furthermore, this type of rust blister is very hard and it is difficult to
remove even with hammers.

Fig. 1. Pitting corrosion on hold frames of a 14 year-old bulk carrier: (a) upper part; (b) middle part; (c)
lower part (after sandblasted).
ARTICLE IN PRESS
406 T. Nakai et al. / Marine Structures 17 (2004) 403–432

Fig. 2. Schematic view of position of hold frames and loaded conditions.

Fig. 3. Corrosion pit and rust blister.

2.2. General corrosion and pitting corrosion

General corrosion and localized corrosion such as pitting corrosion are typical
corrosion observed on hull structural members. Until the rule amendment in 1992,
coating of hold frames was not required. When there was no coating on hold frames,
general corrosion was observed after long-term use. When hold frames have a
protective coating, pitting corrosion is observed. When many corrosion pits are
generated, it is difficult to remove the heavy rust and even if the rust is removed,
correct thickness measurement is difficult due to unevenness of the surface.
Furthermore, the problem arises which indices (average thickness, minimum
thickness, pit intensity and so on) are appropriate for the evaluation of the progress
of corrosion. In the present study, an attempt has been made to evaluate the
ARTICLE IN PRESS
T. Nakai et al. / Marine Structures 17 (2004) 403–432 407

Fig. 4. Cross-sectional view of corrosion pit: (a) pit on hold frame of bulk carrier; (b) pit on bottom shell
of tanker.

experimental data using parameters such as average thickness loss and area ratio of
pitting corrosion as shown later.

2.3. Shape of corrosion pit

Shape of corrosion pits observed on hold frames of a 12 year-old bulk carrier, a 20


year-old bulk carrier and the above-mentioned 14 year-old bulk carrier were
investigated using impression material and a depth gauge. In the case of corrosion
pits whose diameter is approximately larger than 10 mm, the pits are molded with
impression material. Then the shape and depth of pits are investigated from the
observation of the cross section of the molding of the pits. In the case of corrosion
pits whose diameter is approximately smaller than 10 mm, the depth of the pits are
measured using a depth gauge. Furthermore, corrosion pits observed on bottom
shells of a 22 year-old single hull tanker was investigated separately. Cross-sectional
view of the corrosion pit is shown in Fig. 4. The interesting point is that in the case of
the pit on the hold frames of the bulk carriers, the shape of the corrosion pit is a
circular cone and on the other hand in the case of the pit on the bottom shell of the
tanker, the shape of the corrosion pit is a part of a sphere.
The relationship between depth and diameter of the corrosion pits is plotted in
Fig. 5. The trend for the pits observed on the bottom shell of the tanker is also
included in this figure. A trend is seen in this figure that the ratio of the diameter to
the depth is approximately constant (between 10 to 1 and 8 to 1 in the case of
corrosion pits of the hold frames of the bulk barrier and between 4 to 1 and 6 to 1 in
the case of the bottom shells of the tanker). It can be said that the corrosion pits on
the bottom shells of the tanker have larger depth than that of hold frames of the bulk
carrier when they have the same diameter. It is considered that corrosion pits grow in
a different way depending on environment. For example, the difference of the
amount of water existing in each environment would possibly affect the growth of
corrosion pits.

3. Tensile test

3.1. Test specimen (including pitting intensity and average thickness loss)

Hold frames with pitting corrosion have been taken from the above-mentioned 14
year-old bulk carrier as a sample material. The hold frames are made of YP32 steel
ARTICLE IN PRESS
408 T. Nakai et al. / Marine Structures 17 (2004) 403–432

Ratio of diameter to depth


40 Bulk Carrier 10 to 1
A(14years) 8 to 1
B(12years)
Diameter (mm)
30 C(20years)

SH Tanker
D(22 year)
20

10 6 to 1
4 to 1

0 1 2 3 4
Depth (mm)

Fig. 5. Relationship between diameter and depth of corrosion pit.

(NK class grade KA32 steel [5]). Seventeen specimens (sixteen small specimens and
one wide specimen) for tensile tests were cut out from the web plates of the hold
frames. The original thickness of the web plates is 10 mm. In order to examine the
effect of corrosion pit on the strength in detail, specimens with a variety of number
of pits were prepared. Average thickness, width and gauge length of these specimens
are summarized in Table 1. The average thickness of the corroded specimens, which
was calculated from their weight, ranges form 6.3 to 10.0 mm. Variety of number of
corrosion pits on both sides of each specimen is also listed in Table 1. Prior to the
experiment, diameter of corrosion pits observed on the wide specimen has been
investigated. 132 to 178 pits have been observed on one side of the specimen and the
average diameter of the pits is 25 mm.
The original thickness of the above-mentioned specimens is restricted to 10 mm.
For the purpose of investigating the effect of thickness and a variety of pit
distribution in further detail, seventy five specimens were prepared from virgin YP32
steel plates. The thickness of the plates is 10, 14, 16, 19 and 22 mm. Mechanical
properties of each plate are given in Table 2. As shown in Table 2, seven different
heats were used. Yield point ranges from 347 to 457 MPa and tensile strength is in
the range between 504 and 555 MPa and total elongation from 23.3 to 30.5%. Width
and gauge length (GL) of the small specimens are 80 and 200 mm, respectively.
Artificial pits were made on the surface between the gauge marks by drilling to
simulate pitting corrosion. The shape of the artificial pits is determined from the
above-mentioned results of investigating the actual pitting corrosion. The shape
adopted in this experiment is a circular cone and ratio of the diameter to the depth is
8 to 1. And the pit diameter employed is 20, 30 and 40 mm. Thickness of some of the
specimens is reduced by milling. Twenty eight kinds of pit distribution are employed
and among these nine representative kinds are tabulated in Table 3 together with
number of pits and area ratio of pits (pitting intensity). Area ratio of pits is defined as
ARTICLE IN PRESS
T. Nakai et al. / Marine Structures 17 (2004) 403–432 409

Table 1
Tensile test specimens with actual pitting corrosion

No. Average thickness (mm) Width (mm) G.L. (mm) Number of pits

Side-A Side-B

XA1 8.0
XA2 8.3
XA3 8.6 Severala Manyb
XA4 8.7
XA5 7.1
XA6 7.1
XA7 10.0 0 0
XA8 10.0 80 200 1 0
XA9 9.9 1 0
XA10 9.8 1 0
XA11 9.5 10 4
XA12 9.7 9 8
XA13 9.8 6 8
XA14 7.0
XA15 7.6 Manyb Manyb
XA16 7.8
XB 6.3 200 300 132 178
a
More than 40% of surface is corrosion-pitted.
b
More than 80% of surface is corrosion-pitted.

Table 2
Mechanical properties of tested YP32 steel

Heat no. t (mm) Y.P. (MPa) T.S. (MPa) El. (%)

H1 10 381 536 28.9


H2 14 372 523 26.4
H3 16 347 516 30.5
H4 19 366 504 29.0
H5 22 360 527 30.2
H6 10 397 555 27.3
H7 10 457 507 23.3

area covered with pits divided by total area between the gauge marks. Heat nos.,
thickness, pit distribution and average thickness of 27 representative specimens are
represented in Table 4. In addition, five wide specimens are machined from the steel
plate ðt ¼ 10 mmÞ: Width and gauge length (GL) of the wide specimens are 240 and
400 mm, respectively. Artificial pits are made also on these specimens. Central area
of the specimens (240 mm  200 mm) is divided into three equal areas in width
(80 mm  200 mm) and each area has a pit distribution used in the small specimens.
Pit distribution and average thickness loss of the above-mentioned pitted area of the
wide specimens are listed in Table 5.
ARTICLE IN PRESS
410 T. Nakai et al. / Marine Structures 17 (2004) 403–432

Table 3
Type of pit distribution in tensile test

No. Pit dia. (mm) Number of pits Area ratio of pit

000 — 0 0.00
A27 30 6 0.27
A53 30 12 0.53
A75 30 17 0.75
B47 20 24 0.47
C47-1 40 6 0.47
C47-2 40 6 0.47
A04 30 1 0.044
A91 30 27 0.91

3.2. Fracture behavior and deformation behavior

The specimens after the test are shown in Fig. 6. When the specimens have only
one pit (XA8 in Fig. 6(a)), fracture occurred at the cross section including the pit.
And when there are several pits on the specimens, some cracks were generated at the
bottom of the pits and they coalesced with each other to ductile fracture.
Relationship between load and crosshead displacement is shown in Fig. 7. It can
be seen from this figure that maximum load and elongation of the specimens with
pits are small compared with those of the specimen with no pit. It is clear that
reduction of elongation is caused by the effect of pits from the fact that there is no
reduction of elongation in the case of the specimen (F000 in Fig. 7) in which
thickness is reduced by milling.

3.3. Effect of corrosion pit on nominal tensile strength and total elongation

The test results with the specimens from the actual corroded hold frames are
shown in Fig. 8. Fig. 8(a) shows relationship between nominal tensile strength and
average thickness loss. In the present study, nominal tensile strength snominal is
defined by the following equation:
Pmax
snominal ¼ ; (1)
Ao
where Pmax and Ao are the maximum load and the original cross-sectional area,
respectively. The original cross-sectional area Ao is defined as the original thickness
ARTICLE IN PRESS
T. Nakai et al. / Marine Structures 17 (2004) 403–432 411

Table 4
Small tensile test specimens with artificial pits

T.P. Heat t (mm) Pit distribution Average thickness loss (mm)

Side A Side B

A000 H1 9.8 000 000 0.00


AA27 H1 9.7 A27 A27 0.66
AA53 H1 9.7 A53 A53 1.33
AA75 H1 9.7 A75 A75 1.88
AB47 H1 9.7 B47 B47 0.79
AC47 H1 9.6 C47-1 C47-2 1.57
B000 H2 13.1 000 000 0.00
BA27 H2 13.1 A27 A27 0.66
BA53 H2 13.1 A53 A53 1.33
BA75 H2 13.1 A75 A75 1.88
F000 H2 11.0 000 000 0.00
C000 H3 16.0 000 000 0.00
CA27 H3 16.0 A27 A27 0.66
CA53 H3 16.0 A53 A53 1.33
CA75 H3 16.0 A75 A75 1.88
D000 H4 21.6 000 000 0.00
DA27 H4 21.6 A27 A27 0.66
DA53 H4 21.6 A53 A53 1.33
DA75 H4 21.6 A75 A75 1.88
AA04 H1 9.8 A04 A04 0.11
BA04 H2 13.0 A04 A04 0.11
I000 H5 18.6 000 000 0.00
IA27 H5 18.6 A27 000 0.33
IA53 H5 18.6 A53 000 0.66
IA75 H5 18.6 A75 000 0.94
IA04 H5 18.6 A04 000 0.06
AA91 H6 10.0 A91 A91 2.72
BA91 H2 13.1 A91 A91 2.72
IA91 H5 18.7 A91 0 1.36

Note: specimen designation ‘AA27’ A ¼ thickness; (A:10 mm, B:13 mm, C:16 mm, D:22 mm, F:11 mm,
I:19 mm)
A27 ¼ pit distribution as depicted in Table 3;
A ¼ pit diameter (A:30 mm, B:20 mm, C:40 mm)
27 ¼ area ratio of pit in percentage.

multiplied by the width of the specimen. As shown in this figure, nominal tensile
strength decreases with the increase of area ratio of pit. It is clear from Fig. 8(a) the
reduction of nominal tensile strength of the plates with pitting corrosion is 2.5 times
as large as that of the plates with uniform thickness loss in terms of average thickness
loss in worst cases. Fig. 8(b) gives the relationship between total elongation and
average thickness loss. It is seen that total elongation decreases drastically in the case
where the specimens have corrosion pits.
The relationship between maximum net stress (tensile strength at fracture surface)
and average thickness loss is represented in Fig. 9. In the present study maximum net
ARTICLE IN PRESS
412 T. Nakai et al. / Marine Structures 17 (2004) 403–432

Table 5
Wide tensile test specimens with artificial pits

T.P. Heat t (mm) Pit distribution Pit distribution Average thickness


loss (mm)
Side A Side B

Left Center Right Left Center Right

AW1 H6 9.7 A27 A53 A75 A75 A53 A25 1.29


AW2 H6 9.7 B47 A53 C47-2 C47-1 A53 B47 1.23
AW3 H7 9.4 B47 A53 C47-2 B47 A53 C47-2 1.23
AW4 H7 9.4 B47 C47-2 A53 A53 C47-2 B47 1.23
AW5 H7 9.4 B47 C47-2 A53 B47 A53 C47-1 1.23

Note: Type of pit distribution is tabulated in Table 3.

Fig. 6. Small tensile test specimens after test: (a) specimens with actual corrosion pit; (b) specimens with
artificial pit.

stress snmax is defined by the following equation:

Pmax
snmax ¼ ; (2)
Afo

where Afo is the area of the fracture part before loading projected to the plane
perpendicular to the load axis. It is considered that the large data scattering in Fig. 9
is caused by the difficulty to measure the area at the fracture part. Data scatter in the
range between 409 and 589 MPa and the average value is 527 MPa, while the
requirement of the tensile strength for this steel ranges from 440 to 590 Mpa [5].
Although the tensile strength of the steel before use is unknown, it seems that tensile
strength of the material did not change after use. It can be said that the thickness
changes from position to position due to the pitting corrosion and this leads to the
variety of the nominal strength.
ARTICLE IN PRESS
T. Nakai et al. / Marine Structures 17 (2004) 403–432 413

700

600 B000(No Pit)


F000(No Pit)
Load (kN) 500

400 Original
BA53 Thickness (mm)
300 B000 : 13
BA27 : 13
200 BA53 : 13
BA75 : 13
BA75 F000 : 11
100
BA27

0 20 40 60 80 100 120
Displacement (mm)

Fig. 7. Relationship between load and displacement.

An example of the test results with the artificial pitted specimens which have the
same pit distribution on both sides is shown in Fig. 10. Nominal tensile strength
versus area ratio of pit relation is plotted in Fig. 10(a). As shown in this figure,
nominal tensile strength decreases with the increase of area ratio of pit. Fig. 10(b)
gives the relationship between total elongation and area ratio of pit. As shown in
this figure, total elongation decreases drastically in the case where the specimens
have pits.
In cargo holds of bulk carriers, both sides of web plates are subject to
corrosive environment while only one side of side shell plates is exposed to
corrosive environment. From this point of view, the test results with the specimens
with pits on both sides (simulating web plates) and those with pits on one
side (simulating shell plates) are represented in Fig. 11. Nominal tensile strength and
total elongation against area ratio of pit are plotted in Figs. 11(a) and (b),
respectively.
As mentioned above, total elongation of the specimens with pits is small. Cross-
sectional area is different from position to position due to the presence of pits and
strain concentration occurs at the minimum cross section. As a result, total
deformation of the specimens become smaller. In the case where the original
thickness is large, recovery of total elongation with the increase of area ratio of pits is
observed as shown in Figs. 10(b) and 11(b). Number of cross sections with minimum
cross-sectional area increases with the increase of area ratio of pits. This implies that
number of cross sections where strain concentration occurs increases. This leads to
the recovery of total elongation. On the other hand, when the original thickness is
10 mm, the recovery of total elongation is small as shown in 11(b). In the case where
the original thickness is small, strain concentration at the bottom of the pits also
becomes relatively large. In other words, area with plastic deformation becomes
ARTICLE IN PRESS
414 T. Nakai et al. / Marine Structures 17 (2004) 403–432

600 Trend for

Nominal Tensile Strength (MPa)


Uniform Thickness Loss
500
r
400

2.5r
300

200
small specimen
100 wide specimen
original thickness : 10mm

0 1 2 3 4
(a) Average Thickness Loss (mm)

original thickness : 10mm


30
small specimen
Total Elongation (%)

wide specimen

20

10

0 1 2 3 4
(b) Average Thickness Loss (mm)

Fig. 8. Effect of pit on nominal tensile strength and total elongation (specimens with actual pitting
corrosion): (a) nominal tensile strength; (b) total elongation.

small. As a result, the recovery of total elongation with the increase of numbers of
minimum cross sections becomes small.
Figs. 12(a) and (b) compare total elongation and nominal tensile strength of the
actual corroded specimens on which the corrosion pits are located randomly with
those of the artificially pitted specimens on which the artificial pits are located
orderly. As shown in these figures, the trend of the actual corroded specimens
corresponds well with that of the artificially pitted specimens where the pit diameter
is 30 mm. The average pit diameter of each actual corroded specimen ranges from 25
to 30 mm except a few specimens. This suggests that specimens with orderly located
artificial pits which have a fixed diameter equal to the average pit diameter in actual
corroded specimens could simulate the actual corroded specimen where the pit
diameter is random and the pits are located randomly.
ARTICLE IN PRESS
T. Nakai et al. / Marine Structures 17 (2004) 403–432 415

700

600

Maximum Net Stress (MPa) 500

400
Ave. : 527 MPa
300

200 original thickness : 10mm


small specimen
100
wide specimen

0 1 2 3 4
Average Thickness Loss (mm)

Fig. 9. Relationship between maximum net stress and average thickness loss (specimens with actual pitting
corrosion).

3.4. Test result with wide specimen

Wide tensile specimen with actual pitting corrosion after the test is illustrated in
Fig. 13. It is seen that four cracks were initiated at the bottom of the pits and they
coalesced to form one main crack and this led to fracture. One sub-crack was also
observed away from the main crack.
Fig. 14 shows the relationship between load and crosshead displacement for the
wide specimens (see Table 5) with artificial pits where area ratio of pits is the same
for all the specimens. Compared with AW1, maximum load in AW2 is about 20%
small. Diameter of all the pits in AW1 is 30 mm while AW2 has a variety of pit
diameter (20, 30 and 40 mm). Two large pits with a diameter of 40 mm exists on each
side of AW2 in the area shown by a circle in Fig. 15. It is considered that local plastic
deformation developed and local fracture occurred at this area and this leads to the
reduction of maximum load in AW2. In wide specimens, total elongation was
measured at seven points with the space of 30 mm. Distribution of total elongation in
AW1 and AW2 is shown in Fig. 16 in which the horizontal axis indicates the distance
from the edge of the specimens. In the case of AW1 where all the pits have the same
diameter of 30 mm, total elongation is almost the same at all the gauging points. On
the other hand in the case of AW2 where pit diameter is different (20, 30 or 40 mm)
depending on the regions, total elongation is different from position to position and
is relatively small at the region shown by # in Fig. 16. Total elongation of the small
specimens with the pit distribution used in the wide specimens is also plotted in this
figure. Pit distribution in AW1 consists of pit distribution in the three small
specimens (AA27, AA53 and AA75) and total elongation of these specimens is
plotted in corresponding region in Fig. 16 by a dashed line. Among these three small
specimens, total elongation in AA27 is the smallest. In AW1, total elongation in the
region which has the same pit distribution as AA27 is almost equal to that of AA27.
ARTICLE IN PRESS
416 T. Nakai et al. / Marine Structures 17 (2004) 403–432

Average Thickness Loss (mm)


0 0.5 1 1.5 2 2.5
700

Nominal Tensile Strength (MPa) 600

500

400

300 original thickness


22mm
200
16mm
100 13mm
10mm

0 0.25 0.5 0.75 1


(a) Area Ratio of Pit

Average Thickness Loss (mm)


0 0.5 1 1.5 2 2.5
40
original thickness
13mm 22mm
10mm 16mm
30
Total Elongation (%)

20

10

0 0.25 0.5 0.75 1


(b) Area Ratio of Pit

Fig. 10. Effect of pit on nominal tensile strength and total elongation (specimens with artificial pits -1): (a)
nominal tensile strength; (b) total elongation.

Pit distribution in AW2 is the combination of the pit distribution in the three small
specimens (AC47, AA53 and AB47) and total elongation of these three small
specimens is plotted in corresponding region in Fig. 16 by a solid line. Among these
three small specimens, total elongation in AC47 is the smallest. In AW2, total
elongation in the region which has the same pit distribution as AC47 is almost equal
to that of AC47. As shown above, in the wide specimens, total elongation is strongly
affected by pit diameter and distribution and is different from location to location.
As a result, fracture occurs locally at the region where its elongation reaches its
ARTICLE IN PRESS
T. Nakai et al. / Marine Structures 17 (2004) 403–432 417

700
Diameter of Pit : 30 mm

Nominal Tensile Strength (MPa)


600

500

400

300
original thickness
200
10mm (Web : Pit on both sides)
100 13mm (Web : Pit on both sides)
19mm (Shell : Pit on one side)

0 0.25 0.5 0.75 1


(a) Area Ratio of Pit

40
10mm (Pit on both sides)
original
13mm (Pit on both sides)
thickness:
19mm (Pit on one side)
30
Total Elongation (%)

20

10

Diameter of Pit : 30 mm

0 0.25 0.5 0.75 1


(b) Area Ratio of Pit

Fig. 11. Effect of pit on nominal tensile strength and total elongation (specimens with artificial pits-2): (a)
nominal tensile strength; (b) total elongation.

critical value first and this local fracture is the dominant factor which determines the
maximum load.

3.5. Prediction of maximum load

Fig. 17 shows the results of predicting maximum load for the test specimens with
artificial pits. In Fig. 17(a), maximum load is estimated using the following equation:
Ppre1 ¼ Afo  sB ; (3)
where Afo is area of the fracture surface before loading projected to the plane
perpendicular to the load axis and sB is tensile strength. In Fig. 17(b), maximum
ARTICLE IN PRESS
418 T. Nakai et al. / Marine Structures 17 (2004) 403–432

600 original thickness:10mm

Nominal Tensile Strength (MPa)


500 actual corrosion

400

300
diameter of pit
200 10mm
20mm
100 (artificial pit)
30mm
40mm

0 1 2 3 4
(a) Average Thickness Loss (mm)

original thickness:10mm
30
diameter of pit
10mm
Total Elongation (%)

20mm
30mm (artificial pit)
20
40mm

actual corrosion
10

0 1 2 3 4
(b) Average Thickness Loss (mm)

Fig. 12. Effect of pit on nominal tensile strength and total elongation (actual corrosion and artificial
pitting): (a) nominal tensile strength; (b) total elongation.

Fig. 13. Wide tensile specimen after test (XB).


ARTICLE IN PRESS
T. Nakai et al. / Marine Structures 17 (2004) 403–432 419

1000

800
Load (kN) AW1

600
AW3
AW4,5

400

200 AW2

0 10 20 30 40
Displacement (mm)

Fig. 14. Relationship between load and displacement for wide specimens.

Fig. 15. Wide tensile specimen after test (AW2).

15
AW1
AB47 AW2
Total Elongation (%)

AA53
10
AA75
AA27

5
AC47
# #

0 80 160 240
Distance from Edge (mm)

Fig. 16. Total elongation in wide specimens.


ARTICLE IN PRESS
420 T. Nakai et al. / Marine Structures 17 (2004) 403–432

1000

Estimated Maximum Load (kN)

500

Small Specimen
Wide Specimen

0 500 1000
(a) Experimental Maximum Load (kN)

1000
Estimated Maximum Load (kN)

500

Small Specimen
Wide Specimen

0 500 1000
(b) Experimental Maximum Load (kN)

Fig. 17. Estimation of maximum load in tensile test: (a) estimation from area of fracture surface; (b)
estimation from minimum cross sectional area.
ARTICLE IN PRESS
T. Nakai et al. / Marine Structures 17 (2004) 403–432 421

load is calculated using the following equation:


Ppre2 ¼ Am  sB ; (4)
where Am is area of minimum cross section before loading perpendicular to the load
axis. It can be seen from these figures that maximum load in the small specimens is
well predicted. However, maximum load in some of the wide specimens is
overestimated. In the case of the small specimens, fracture occurs after plastic
deformation develops all over the cross section. For that reason, above equations are
appropriate for the estimation of maximum load. On the other hand in the case of
the wide specimens, total elongation is strongly affected by pit diameter and its
distribution and is different from location to location. As a result, fracture occurs
locally at the region where its elongation reaches its critical value first. Therefore,
above equations are not always appropriate for the prediction of maximum load.

4. Compressive buckling test

4.1. Test specimen

Compressive test specimens were fabricated from web plates, shell plates and face
plates by welding. Tested steel is the same as that used in tensile test. One test
specimen is machined from the actual hold frame taken from the above-mentioned
14 year-old bulk carrier. An example of the test specimens is shown in Fig. 18. As
shown in Fig. 18, width and depth of the web plate, which is the test section of the
specimen, are 185 mm and 450 mm, respectively and its thickness is 10 mm or 13 mm.
Web plates with a thickness of 13 mm were prepared by milling the steel with
thickness of 16 mm. Pit distribution, number of pits and area ratio of pit used in
compression test are summarized in Table 6 and pit distribution and average
thickness loss of the web plate of each specimen are listed in Table 7. Prior to the

Fig. 18. Example of test specimen.


ARTICLE IN PRESS
422 T. Nakai et al. / Marine Structures 17 (2004) 403–432

Table 6
Pit distribution in compression test

No. Pit dia. mm Number of Pits Area ratio of pit

00 — 0 0.00
D1 40 4 0.06
D2 40 4 0.06
D3 40 4 0.06
D4 40 20 0.30
D5 40 16 0.24
D6 40 1 0.015
E1 30 36 0.30
E2 30 36 0.30
E3 30 36 0.30
E4 30 36 0.30
E5 30 35 0.30
E6 30 35 0.30
E7 30 6 0.05
E8 30 18 0.15
F1 — 0 See Fig. 19

experiment, corrosion pits observed on the web plate of the specimen from the actual
corroded member was investigated. 80 to 113 pits have been observed on each side of
the web plate and the average diameter of the pits is 29 mm. No pit was made on web
plates of JF11 and KF11 in Table 7 and thickness of the web plates near the shell
plate and the face plate were reduced by milling as shown in Fig. 19. Aspect ratio a of
the web plate is 2.43 and slenderness ratio b of the web plate is 0.81(t ¼ 10 mm) and
0.58(t ¼ 13 mm). a and b are defined by the following equations:
a
a¼ ; (5)
b
rffiffiffiffiffi
b sy
b¼ ; (6)
t E
where a; b and t are depth, width and thickness of the web plate, respectively and sy
and E are yield stress and Young’s modulus, respectively.
ARTICLE IN PRESS
T. Nakai et al. / Marine Structures 17 (2004) 403–432 423

Table 7
Compression test specimens

T.P. Heat T (mm) Pit distribution Average thickness loss (mm)

Side A Side B

JD10 H6 9.9 D1 0 0.10


JD20 H6 9.9 D2 0 0.10
JD30 H6 10.0 D3 0 0.10
JD45 H6 10.0 D4 D5 0.91
KD45 H3 13.2 D4 D5 0.91
J000 H6 9.9 0 0 0.00
JD60 H6 10.2 D6 0 0.025
JE12 H6 9.9 E1 E2 0.76
JF11 H6 9.9 F1 F1 2.2
KF11 H3 13.0 F1 F1 2.2
JE34 H6 10.0 E3 E4 0.76
KE34 H6 13.2 E3 E4 0.76
JE56 H6 9.9 E5 E6 0.74
KE56 H3 13.1 E5 E6 0.74
JE77 H6 9.9 E7 E7 0.13
JE88 H6 10.0 E8 E8 0.38
YA (KA32) 10.0 actual corrosion 2.0

Note: specimen designation ‘JD45’


J ¼ thickness; (J: 10 mm, K: 13 mm)
D45 ¼ pit distribution D4 on side A and D5 on side B (see Table 6)
D ¼ pit diameter (D: 40 mm, E: 30 mm, F : see Fig. 19).

Fig. 19. Web plate of JF11 and KF11: (a) JF11; (b) KF11.
ARTICLE IN PRESS
424 T. Nakai et al. / Marine Structures 17 (2004) 403–432

4.2. Buckling behavior

An example of the test specimens after the test (JE34 and YA) is shown in Fig. 20.
Compressive load acted on the web plate in this experiment and buckling behavior
can be seen as shown in this figure. Load shortening curve and lateral deflection-load
relation in JE34 are shown in Figs. 21 and 22, respectively. In these figures, result of
an elasto-plastic analysis described later is also plotted. It can be seen from these
figures that after the maximum load, load decreases drastically and lateral deflection
develops drastically with the decrease of load.

4.3. Ultimate strength

The test results, that is, ultimate strength and average thickness loss relation is
represented in Fig. 23. Result of an elasto-plastic analysis in the case of general
corrosion (uniform corrosion corresponding to the average thickness loss) described
later is also plotted in this figure. From the above data, it is apparent that in the case
where the web plates have pitting, the ultimate strength is less than or equal to that
of the web plate with uniform thickness loss.

4.4. Effect of pit location

In Fig. 24, the ultimate strength for the specimens where four pits with a diameter
of 40 mm are arranged in a row (JD10-JD30, t ¼ 10 mm) is depicted. That is, this
figure represents the effect of pit location on ultimate strength when the arrangement
of pits is the same. Pit location in each specimen is also shown in this figure. A solid
line shows the ultimate strength when the web plate has no pit (J000, t ¼ 10 mm) and
a dashed line shows the ultimate strength by FE-analysis when the web plate has
uniform corrosion corresponding to the average thickness loss. It can be seen from

Fig. 20. Buckling behavior: (a) JE56; (b) YA.


ARTICLE IN PRESS
T. Nakai et al. / Marine Structures 17 (2004) 403–432 425

600

Original Thickness : 10mm JE34


500 Pit Diameter : 30mm
Area Ratio of Pit : 0.3
400 Average Thickness Loss : 0.76mm
Load (kN)

300

200

100 Experiment
FEM, w0=t/30

0 1 2 3 4 5
Displacement (mm)

Fig. 21. Load shortening curve for JE34.

50
Original Thickness : 10mm JE34
Pit Diameter : 30mm
40
Lateral Deflection (mm)

Area Ratio of Pit : 0.3


Average Thickness Loss : 0.76mm

30

20 Experiment
FEM, w0 = t/30

10

0 100 200 300 400 500


Load (kN)

Fig. 22. Relationship between lateral deflection and load.

this figure that ultimate strength is getting smaller as the location of pits gets closer
to the center of the web plate. In the case of the web plate with no pit, maximum
lateral deflection occurs at the center of the web plate and it can be said that ultimate
strength becomes small as the location of pits getting near to the region where large
lateral deflection occurs. For the case of a both-end clamped plate as in the present
experiment, the ultimate compressive load is in equilibrium with the ultimate
bending moment at both ends as well as at mid-span. It is considered that the pits
ARTICLE IN PRESS
426 T. Nakai et al. / Marine Structures 17 (2004) 403–432

800
t=10mm (Uniform Corrosion, FEM)
700 t=13mm (Uniform Corrosion, FEM)

Ultimate Strength (kN) 600

500

400

300

200
t=10mm (Experiment)
100 t=13mm (Experiment)
t=10mm (Exp.,Actual Corrosion)

0 0.5 1 1.5 2 2.5 3


Average Thickness Loss (mm)

Fig. 23. Relationship between ultimate strength and average thickness loss.

600
Original Thickness : 10mm J000:No Pit
500
Ultimate Strength (kN)

400

300

200

100
uniform thickness loss of 0.1mm (FEM)
0
JD10 JD20 JD30
Test Piece No.

Fig. 24. Effect of pit location on ultimate strength.

near the upper or lower edges and those at mid-span therefore have a similar
effect on the ultimate strength, which contradicts the test results. However, as
shown later in FE-Analyses, in the case where number of pits is small and pits exist
only on one side, the ultimate strength is strongly affected by which side is the
convex of the initial imperfection. Therefore, the pits near upper or lower edges and
those at mid-span don’t always have a similar effect on the ultimate strength as
shown in Fig. 24.
ARTICLE IN PRESS
T. Nakai et al. / Marine Structures 17 (2004) 403–432 427

600
Original Thickness : 10mm J000:No Pit
500
Ultimate Strength (kN)
400

300

200

100
uniform thickness loss of 0.75mm (FEM)
0
JE12 JE34 JE56

Test Piece No.

Fig. 25. Effect of pit distribution on ultimate strength.

4.5. Effect of pit distribution

In Fig. 25, the ultimate strength for the specimens where 35 to 36 pits with a
diameter of 30 mm are made on each side (JE12, JE34, JE56, t ¼ 10 mm) is plotted.
That is, this figure represents the effect of pit distribution on ultimate strength when
pit diameter and number of pits are fixed. Pit distribution in each specimen is also
included in this figure. A solid line shows the ultimate strength when the web plate
has no pit (J000, t ¼ 10 mm) and a dashed line shows the ultimate strength by FE-
analysis when the web plate has uniform corrosion corresponding to the average
thickness loss. It can be seen from this figure that ultimate strength gets smaller when
pits are locally concentrated at the middle part or at the upper and lower part of the
web plate. The average thickness of the web plate of these three specimens is almost
the same and it can be said that average thickness loss is not always a dominant
factor of ultimate strength and it is affected by pit distribution.

5. FE-analysis of compressive buckling test

In this chapter, an elasto-plastic analysis has been conducted in order to simulate


the effect of corrosion pit on compressive buckling behavior. FE-Code MSC. Marc
has been used in the present analysis. MSC.Marc contains an extensive element
library [6]. From the element library, a four-node, thick-shell element (Element Type
75) has been chosen. It is considered that very fine mesh is necessary to model the
pitted members because diameter of the smallest pit in the specimens is 7 mm.
Therefore, the meshing size of the elements used to model the web plates is
ARTICLE IN PRESS
428 T. Nakai et al. / Marine Structures 17 (2004) 403–432

approximately the same as the original thickness of the web plates. The same size of
elements is employed for the thickness, 10 and 13 mm.

5.1. Material constants and initial imperfection

Material constants used in the analysis are; Young’s modulus E ¼ 205:8 GPa;
yield stress sy ¼ 397 MPa (t ¼ 10 mm), sy ¼ 347 MPa (t ¼ 13 mm) for the virgin
YP32 steels and sy ¼ 397 MPa (t ¼ 10 mm) for the actual corroded web plate. Strain
hardening is neglected in the present analysis.
To consider the deflection pattern of the column fixed at both ends, initial
imperfection w is assumed to be expressed by the following equations:
2p
w ¼ w0 cos y; (7)
b
w0 ¼ t=30; (8)
where y direction is the direction of depth and y equals to zero at the upper end of
the web plate and t is the original thickness of the web plate. Maximum value of
initial imperfection w0 is determined so that the estimated ultimate strength
corresponds well with the experimental one in the case where specimen has no pit
(JC1, t ¼ 10 mm). The effect of welding-induced residual stress is not considered in
the present analysis.

5.2. Modeling of corrosion pit

For the sake of modeling members with corrosion pit using shell elements, the
following procedure has been taken:
(1) measure the location and the diameter of the corrosion pit,
(2) assume that the ratio of the pit diameter to the depth is 8 to 1,
(3) assume that the shape of the corrosion pit is a circular cone,
(4) calculate the thickness at each node of the shell elements,
(5) determine the thickness of the shell elements by averaging the thickness at each
node.

In the case of specimens with artificial pits in the present study, the items (1)–(3)
can be omitted because the location and the shape of artificial pit is known.
When pits are made differently on each side of the web plate, eccentricity exits due
to pitting. The eccentricity due to pitting can be taken into account by shifting the
nodes at the center of the thickness. In this case, following two types of modeling are
possible:

(1) eccentricity due to pitting and initial imperfection with convex on side A is
considered (Type A),
(2) eccentricity due to pitting and initial imperfection with convex on side B is
considered (Type B),
ARTICLE IN PRESS
T. Nakai et al. / Marine Structures 17 (2004) 403–432 429

In addition, following two types of modeling are considered in the present


analysis:

(3) only initial imperfection is considered (Type C),


(4) only eccentricity due to pitting is considered (Type D),

In order to examine the effect of eccentricity due to pitting and initial


imperfection, a series of analyses of the test specimens with pits differently on each
side has been made using four types of modeling shown above.
The calculated load shortening curve for JD10 where number of pits is small and
pits exist only on one side and JE34 where number of pits is large and pits exist on

600
Imperfection
+No eccentricity due to pitting (Type C)
500
Imperfection Type A
400
Load (kN)

300 Imperfection Type B

200

100 Experiment
No Imperfection (Type D)
JD10

0 1 2 3
(a) Displacement (mm)
600
No Imperfection (Type D) JE56
500
Imperfection
+ No eccentricity due to pitting
400 (Type C)
Load (kN)

300

200
Imperfection Type B
100 Experiment
Imperfection Type A

0 1 2 3
(b) Displacement (mm)

Fig. 26. Effect of eccentricity and initial imperfection on analysis: (a) JD10; (b) JE56.
ARTICLE IN PRESS
430 T. Nakai et al. / Marine Structures 17 (2004) 403–432

both sides are represented in Fig. 26(a) and (b), respectively. The load shortening
curves obtained from the experiment are also plotted in these figures.
As indicated in Fig. 26(a), in the case where number of pits is small and pits exist
only on one side, the calculated results with Types A and B is quite different. In type
A, the effect of the eccentricity due to pitting is reduced by the initial imperfection
and for this reason the calculated ultimate load is relatively large. In type B, the
effect of the eccentricity is increased by the initial imperfection and for this reason
the calculated ultimate load is relatively small.
As shown in Fig. 26(b), in the case where number of pits is large and pits exist on
both sides, the difference of calculated results with Types A to C are small.
Furthermore, the calculated ultimate strength corresponds well with the experi-
mental one. In this case, it is considered that the eccentricity due to pitting on one
side is reduced by the eccentricity due to pitting on another side and for this reason,
the total effect of eccentricity due to pitting is small.

5.3. Prediction of ultimate strength

When the structural analysis of the actual corroded structural components is


considered, it is very difficult to take into account the eccentricity of every corrosion
pit and when the hold frames of bulk carriers are considered, both sides are exposed
to corrosive environment and the case where there are many corrosion pits is the
problem. According to the analytical results in the previous section, the effect of the
eccentricity due to pitting is small when there are many pits on both sides of the web
plates. From such a practical point of view, an analysis has been made with all the
test results taking into account initial imperfection only (Type C) and estimated
ultimate strength is compared with experimental one.
The results of predicting ultimate strength for each test specimen considering
initial imperfection and neglecting eccentricity due to pitting is plotted in Fig. 27.
Because of the neglect of the eccentricity due to pitting, a little overestimation of
ultimate strength is observed for some testing conditions, but estimated ultimate
strength corresponds well with the experimental results on the whole.

6. Future work

In the present study, the effect of pitting corrosion on tensile strength and
compressive buckling strength as a first step to understand fracture or buckling
behavior of members with pitting corrosion. In actual damages of bulk carriers,
lateral-torsional buckling of a whole hold frame, shear buckling or fracture of web
plates, etc. are observed in some cases. Basic mechanical properties such as tensile
strength or compressive buckling strength investigated in the present study is not
always dominant factor of collapse or fracture of hold frames. Therefore, it is
necessary to examine the effect of pitting corrosion from a view point of structural
strength such as lateral-torsional buckling of a whole hold frame, shear strength or
ARTICLE IN PRESS
T. Nakai et al. / Marine Structures 17 (2004) 403–432 431

800

700
Estimated Ultimate Strength (kN)
600

500

400

300

200
w0 = t/30 (t=10mm)
100 w0 = t/30 (t=10mm,Actual Corrosion)
w0 = t/30 (t=13mm)

0 100 200 300 400 500 600 700 800


Experimental Ultimate Strength (kN)

Fig. 27. Estimation of ultimate strength in compression test.

local buckling of web plates, etc. Further experiment and analysis are ongoing which
will be reported in a subsequent paper.

7. Conclusions

In this research project, in order to reveal the effect of corrosion especially pitting
corrosion on the static strength of hold frames of bulk carriers, ‘investigation of
actual corrosion’, ‘investigation of strength of pitted members’ and ‘simulation of
strength reduction of structural members due to corrosion’ are made. In this first
report, actual pitting corrosion observed on hold frames of bulk carriers were
investigated and a series of tensile tests and compression tests with pitted members
has been conducted. Furthermore, in order to establish the method of modeling
pitted members in FE-analyses, an FE-analysis of the compression test has been
conducted. Following conclusions are derived form this work:
(1) In either case of the three bulk carriers investigated, the shape of the corrosion
pit observed on the hold frames is a circular cone and the ratio of pit diameter to
its depth is in the range between 8 to 1 and 10 to 1.
(2) Nominal tensile strength of pitted members decreases gradually and total
elongation of pitted members decreases drastically with the increase of thickness
loss due to pitting. In worst cases, the reduction of nominal tensile strength is 2.5
ARTICLE IN PRESS
432 T. Nakai et al. / Marine Structures 17 (2004) 403–432

times as large as that of members with uniform thickness loss in terms of average
thickness loss.
(3) Maximum net stress of the actual corroded members is on the level of the
requirement for this steel, that is, tensile strength of the material itself did not
change after use.
(4) In the case of small specimens, maximum tensile load is well predicted by the
parameters expressed by Eqs. (3) and (4).
(5) In the case of the wide specimens, plastic deformation concentrates locally and
local fracture starts at the large pit. For this reason, net fracture stress of the wide
specimens becomes smaller than that of the small specimens.
(6) Compressive buckling strength of pitted members is smaller than or equal to that
with uniform thickness loss in terms of average thickness loss.
(7) The effect of eccentricity due to pitting corrosion is small and can be neglected in
numerical model.

References

[1] Yamamoto N, Ikegami K. A study of the degradation of coating and corrosion of ship’s hull based on
the probabilistic approach. J Offsh Mech and Arc Eng 1998;120:121–8.
[2] Yamamoto N, Yao T. Hull girder strength of a tanker under longitudinal bending considering strength
diminution due to corrosion. Proceedings of ICOSSAR 2001, 2001.
[3] ASM International. Corrosion. ASM Handbook. vol. 13, 2001.
[4] Matsushita H, Nakai T, Yamamoto N, Arai H. Effect of corrosion on static strength of hull structural
members (1st report). J Soc Naval Arch Jpn 2002;192:357–65 [in Japanese].
[5] Nippon K.K. rules for the survey and construction of steel ships, Part K Materials, 2003.
[6] MARC Analysis Research Corporation. MARC vol. B Element Library K7, 1997.

You might also like