Proceedings of OMAE05

th

24 International Conference on Offshore Mechanics and Arctic Engineering

June 12-17, 2005, Halkidiki, Greece

OMAE2005-67579
NON-LINEAR TIME DEPENDENT CORROSION WASTAGE OF DECK
PLATES OF BALLAST AND CARGO TANKS OF TANKERS

Y. Garbatov

C. Guedes Soares

Unit of Marine Technology and Engineering,

Technical University of Lisbon, Instituto Superior Tcnico,
Av. Rovisco Pais, 1049-001 Lisboa, Portugal
Email: yordan.garbatov@mar.ist.utl.pt

Unit of Marine Technology and Engineering,

Technical University of Lisbon, Instituto Superior Tcnico,
Av. Rovisco Pais, 1049-001 Lisboa, Portugal
Email: guedess@mar.ist.utl.pt

G. Wang
American Bureau of Shipping
ABS Plaza, 16855 Northchase Dr.
Houston, Texas 77060-6006, USA
Email: GWang@eagle.org

ABSTRACT
The corrosion wastage of deck plates of ballast and cargo
tanks is analyzed based on a non-linear corrosion model.
This model is able to describe an initial period without
corrosion due to the presence of a corrosion protection
system, a transition period with a nonlinear increase of
wastage up to a steady state of long-term corrosion wastage.
This model is applied to corrosion wastage data of deck
plates collected by the American Bureau of Shipping. The
objective of this work is to fit this corrosion wastage model
to the service measured data, determining the values of the
model parameters that represent the best fit to the data so as
to describe how corrosion wastage varies in time as a result
of generalized corrosion.
1

INTRODUCTION

Corrosion is one of the most critical degrading mechanisms

of the structural strength of ships and offshore structures.
Various theoretical approaches are available to determine
the strength of corroded structures as well as their reliability.
All of these approaches depend on corrosion wastage
models that describe how the structural degradation evolves
with time.
The conventional models of corrosion wastage assume a
constant corrosion rate, leading to a linear relationship
between the material lost and time (e.g. Guedes Soares,
1988).

Southwell et al., (1979) proposed a linear and a bilinear

model for corrosion wastage which were considered
appropriate for design purposes and both are conservative in
the early stages of corrosion, and overestimate the corrosion
depth at the initial phases of corrosion progress. Melchers,
(1998) suggested a steady-state tri linear and another power
for corrosion wastage thickness. In fact, experimental
evidence of corrosion reported by various authors shows that
non-linear models are more appropriate.
Yamamoto and Ikegami, (1998) proposed a corrosion model
based on analyzing collected from plate thickness
measurements. The corrosion wastage process was
considered in three periods: period when the anti-corrosive
paint coating is effective (including the period of generation
of active pitting points), period when pitting points are
progressed and period when corrosion process stops and
corrosion rate becomes zero. In this model corrosion and
wear seen in structural members are assumed to be the
consequence of an extremely large number of pits growing
progressively and individually.
Guedes Soares and Garbatov (1999) proposed one model
that describes the growth of corrosion wastage by a nonlinear function of time in three phases. In the first phase, it is
assumed that there is no corrosion because a corrosion
protection system is effective. Failure of the protection
system will occur at a random point of time and the
corrosion wastage will start a non-linear process of growth
with time, which levels off asymptotically at a long-term
value of corrosion wastage. Since then, several authors have

proposed some variants of this model or have compared it

with other modifications.
Sun and Bai, (2001) suggested a time-variant corrosion rate
model, adapting the model of Guedes Soares and Garbatov,
(1999), to describe the corrosion rate instead of the
corrosion wastage thickness. In their model, the corrosion
rate increases as an exponential function of time and finally
tends to be a constant value in the third phase.
Qin and Cui, (2002) assumed that the coating protection
system (CPS) deteriorates gradually so that corrosion may
start as pitting corrosion before the CPS loses completely its
effectiveness. The corrosion rate was defined by equating
the volume of pitting corrosion to uniform corrosion. This
was regarded as the transition period in which the corrosion
rate increases. After the CPS loses its complete
effectiveness, general corrosion starts and the corrosion rate
decreases due to the increasing thickness of the corroded
product. The whole corrosion process was divided into three
stages: no corrosion when the CPS is fully effective,
corrosion accelerating when the pitting corrosion generates
and progresses, and corrosion decelerating. This model is
flexible and can describe different other models of corrosion
rate using the same format. For a particular value of one of
its parameters ( = 1), it becomes the model of Guedes
Soares and Garbatov, (1999).
The corrosion model of Paik et al., (2003) also categorizes
the corrosion behavior into three phases, durability of
coating, transition to visibly obvious corrosion, and progress
of such corrosion. The coating life is assumed to follow the
log-normal distribution and the transition time is considered
to be an exponentially distributed random variable.
A probabilistic model is currently being developed by
Melchers, (2003), which divides the corrosion process into
four stages: initial corrosion, oxygen diffusion controlled by
corrosion products and micro-organic growth, limitation on
food supply for aerobic activity and anaerobic activity. The
proposed model consists of a number of phases, each
represent a different corrosioncontrolling process. To allow
the data from the various sites to be compared for
preliminary analysis, it was necessary to correct the data for
water temperature.
Wang et al., (2003) collected a corrosion database from
measurements made in ships in service and they performed a
statistical analysis of the data leading to values of corrosion
rate of deck plates in cargo tanks of oil tankers. They
derived regression relations of corrosion wastage as a
function of time which provide a consistent fitting to the
data analyzed.
The model of Ivanov et al., (2003) model assumes three
phases of corrosion wear as the one of Guedes Soares and
Garbatov, (1999). However it substitutes the transition phase
of nonlinear thickness increase with time by a linear
relation.
It is seen that in the literature several authors adopt the basic
ideas of the model proposed by Guedes Soares and
Garbatov, (1999), and modify some details. Therefore it is
worthwhile to consider again the adequacy of the model of
Guedes Soares and Garbatov, (1999), by comparing it with
full scale measured data.

The main objective of this work is to fit this corrosion

wastage model to a corrosion wastage data set of deck plates
collected by American Bureau of Shipping (Wang et al.,
(2003) and determine the values of the model parameters
that represent the best fit to the data so as to describe how
corrosion wastage varies in time as a result of generalized
corrosion.
2

NON-LINEAR CORROSION WASTAGE MODEL

The non-linear corrosion wastage model proposed by

Guedes Soares and Garbatov (1999) describes the growth of
corrosion wastage by a non-linear function of time in three
phases. In the first phase, it is assumed that there is no
corrosion because a corrosion protection system is effective.
Failure of the protection system will occur at a random point
of time and the corrosion wastage will start a non-linear
process of growth with time, which levels off asymptotically
at a long-term value of corrosion wastage.
Figure 1 illustrates the time dependent model of corrosion
degradation, separated into four phases, where in the first
one there is no corrosion ( t [O , O] , in Figure 1). The
second phase corresponds to the initiation of failure of the
corrosion protection system, which leads to corrosion with
the decrease of thickness of the plate (OB-fast growing
corrosion). The third phase, BC, corresponds slowly
growing corrosion and the last one, (t>C), corresponds to a
stop in the corrosion process when the corrosion rate
becomes zero.

d( t)

t
A

Figure 1 Thickness of corrosion wastage as a function of

time
The model is based on the solution of a differential equation
of the corrosion wastage:
t c

d (t ) = d 1 e t

d (t ) = 0
t c

, t > c

(1)

where d is the long- corrosion wastage, d (t ) is the

corrosion wastage at time t, c is the time without corrosion
which corresponds to the start of failure of the corrosion

protection coating (when there is one), and t is the

transition time duration, which may be calculated as:

285
266

t =

247

d
tg

228

(2)

209
190

where is the angle defined by OA and OB in Figure 1.

Since corrosion data has a very large variability, the

approach taken has been to model separately the time
variation of the mean corrosion wastage and of the standard
deviation. This allows the main tendency of the data (mean)
to be described by the above corrosion model and the
uncertainty of the model to be described by the standard
deviation of the errors as a function of time.

152
133
114
95
76
57
38
19
0

<= 140 (160,180] (200,220] (240,260] (280,300] (320,340] (360,380] (400,420]

(140,160] (180,200] (220,240] (260,280] (300,320] (340,360] (380,400] > 420
Ship length between perpendiculars, m

Figure 3 Data of deck plates of ballast tanks, versus

ship length between perpendiculars
350

DATA ANALYSIS

Two sets of corrosion data, deck plates of ballast and cargo

tanks of tankers provided by ABS, are analyzed here (ABS,
2002 and Wang et al., 2003a, b). The first set (see Figure 3)
includes 1168 measurements of deck plates from ballast
tanks with original nominal thicknesses varying from 13.5 to
35 mm on ships with lengths between perpendiculars in the
range of 163.5 to 401 m (see Figure 3).

300

250

No of obs

No of obs

Although this model has been validated with some corrosion

data, it is the purpose of this work to validate it against
measured data supplied by ABS, particularly measured
corrosion wastage for ship deck plates of ballast and cargo
tanks.

171

The second set of data (see Figure 3) includes 4665

measurements of deck plates from cargo tank with original
nominal thicknesses varying from 12.7 to 35 mm on ships
with lengths between perpendiculars in the range of 163.5 to
401 m (see Figure 5).

200

150

100

50

1800

10

12

14

16

18

20

22

24

26

28

30

32

34

36

38

As-built thickness, mm

1600

Figure 4 Data of deck plates of cargo tanks, as-built

thickness

1400

1200

1395

No of obs

1302
1000

1209
1116

800

1023
930

600

No of obs

837

400

200

744
651
558
465
372

279
10

12

14

16

18

20

22

24

26

28

30

32

34

36

38

As-built thickness, mm

Figure 2 Data of deck plates of ballast tanks, as-built

thickness

186
93
0

<= 140 (160,180] (200,220] (240,260] (280,300] (320,340] (360,380] (400,420]

(140,160] (180,200] (220,240] (260,280] (300,320] (340,360] (380,400] > 420
Ship length between perpendiculars, m

Figure 5 Data of deck plates of cargo tanks versus ship

length between perpendiculars

50

297

45

270

40

243
216

No of obs, 19th year

No of obs, 19th year

35
30
25
20
15

189
162
135
108
81

10

54

27

0
0.0

0
0.5

1.0

1.5

2.0

2.5

3.0

3.5

Corrosion Depth, mm

Corrosion Depth, mm

Figure 6 Histograms of corrosion wastage of deck plates of

ballast tanks for 19th year

Figure 9 Histograms of corrosion wastage of deck plates of

cargo tanks for 19th year
396

120

360
100

324
288
No of obs, 21st year

No of obs, 21st year

80

60

40

252
216
180
144
108
72

20

36
0
0.0

0.5

1.0

1.5

2.0

2.5

3.0

0
0.0

3.5

0.5

1.0

1.5

2.0

Figure 7 Histograms of corrosion wastage of deck plates of

ballast tanks for 21st year

100

16

90

3.5

4.0

4.5

5.0

80

14

70

12

No of obs, 25th year

No of obs, 25th year

3.0

Figure 10 Histograms of corrosion wastage of deck plates of

cargo tanks for 21st year

18

10
8
6

60
50
40
30

20

2
0
0.0

2.5

Corrosion Depth, mm

Corrosion Depth, mm

10

0.5

1.0

1.5

2.0

2.5

Corrosion Depth, mm

Figure 8 Histograms of corrosion wastage of deck plates of

ballast tanks for 25th year

0
0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

Corrosion Depth, mm

Figure 11 Histograms of corrosion wastage of deck plates of

cargo tanks for 25thyear

In order to provide a detailed description of the variability of

the data, it has been grouped by year and histograms of the
measurements in each year are presented in Figure 6 to 11.

tanks ( R 2 = 0.9) than for the deck plates of cargo tanks

( R 2 = 0.85). The larger value of R 2 does not necessarily
imply that the model will provide accurate prediction.

The frequency scatter diagrams of corrosion wastage, d, are

shown in Figure 12 for deck plates of ballast tanks and in
Figure 13 for deck plates of cargo tanks respectively. The
scatter plots display the frequencies of overlapping points
between time and corrosion wastage in order to visually
represent the frequencies of the overlapping points and
categorize those frequencies according to the number
specified in the right hand side of figures as for an example
(min=2, max=4). The sizes of the point markers in the plots
represent the frequencies.
The long-term wastage d is defined as the maximum value
in the observed time interval for ballast tanks and cargo
tanks respectively. The period without corrosion, or the time
of initiation of corrosion c , and the transition time, t are
defined based on performing a least squares fit to the data
using a quasi-Newton algorithm, which determines the
direction to search used at each iteration considering the
mean value of corrosion depth taken from the yearly subset
of data histograms (see Figure 6 to 11).!|
The judgment of the adequacy of approximation is done by:

di

i =1

n
n

2.0
1.8

(3)

where n is the total number of measurements, l is the total

number of classified years of observation, i is the mean
value of observations during each year

Deck Plates - Ballast Tanks

d=1.85 mm

1.6
Corrosion Depth, mm

l
n

2
2
2
R = 1 ( di i ) / di
i =1
i =1

Figure 13 Frequency scatter diagram of corrosion wastage of

deck plates of cargo tanks.

t=17.541 years
c =10.541 years

1.4
1.2
1.0
0.8
0.6
0.4
0.2
0.0
0

12

16 20 24
Time, years

28

32

36

Figure 14 Time dependent mean corrosion wastage of deck

plates of ballast tanks
The parameters of the regressed line of corrosion depth as a
function of time were determined under the assumption that
it is approximated by the exponential function given in
equation 1. It can be noted that the long-term corrosion
wastage for deck plates of ballast tanks is d ,ballast = 1.85
mm and d,cargo = 1.91 mm for cargo tanks respectively.
Figure 12 Frequency scatter diagram of corrosion wastage of
deck plates of ballast tanks.
There is some variability of the yearly mean values of the
data around the regressed line, as can be seen in Figure 14
and Figure 15, which is smaller for the deck plates of ballast

The time without corrosion is c,ballast = 10.54 years in deck

plates of ballast tanks and c,cargo = 11.494 years for cargo
tanks, respectively. Finally, the transition period for deck
plates of ballast tanks is t ,ballast = 11.14 years and the one
for deck plates of cargo tanks is t ,cargo = 11.23 years.

In the third zone between 13.13 and 19.73 year the

difference between the corrosion depth varies between 0 and
-29 percent and in the fourth zone between 19.73 and 28.1
year the difference varies between -29 and -24.4 which is its
minimum and in the time interval between 28.1 to 32 years
the difference is reduced up to -23.1 at the 32nd year.

2.0
Deck Plates - Cargo Tanks

1.8

d=1.91mm

Corrosion Depth, mm

1.6

t=11.225 years
c =11.494 years

1.4
1.2

Another important statistical descriptor of the data set is the

standard deviation, which is given in Figure 17 and 18 for
each yearly subset of data. The standard deviation as a
function of time is fit to a logarithmic function:

1.0
0.8
0.6

St Dev (t) = a Ln (t) - b,

(4)

0.4
0.2
0.0
0

12

16 20 24
Time, years

28

32

36

which is shown in Figures 17 and 18. Although the figures

show a regression line, the correlation coefficient is very
low and the regression is not significant. In fact there is a
large variability in the data, which is the dominant feature.

The comparison between the corrosion wastage of deck

plates of ballast and cargo tanks as a function of time is
given in Figure 16. It can be seen that there is a different
starting point of corrosion as a result of the different time
duration without corrosion. There is also a different final
point related to the long-term corrosion wastage already
observed during the analysis.
100%

2.0
Ballast Tanks

1.8

1.4
1.2
1
0.8
0.6
0.4
0.2
0
0

12

60%

1.4
1.2

40%

1.0

16

20

24

28

32

36

Time, years

Figure 17 Standard deviation of the yearly data of corrosion

wastage of deck plates of cargo tanks

20%

0.8
0.6

0%

0.4
-20%

0.2
0.0

-40%
8

12

16

20
24
28
Time, years

32

36

Figure 16 Comparative analysis of corrosion wastage of

deck plates of ballast and cargo tanks
If the service life of ship is split into 5 time zones it can be
observed that in the first one from 0 to 10.54 or 11.494 years
there is no corrosion, which is the zone without corrosion.
In the second zone that is limited in the time interval
between 10.54 or 11.494 year to 13.13 year of the service
life of ship the difference between the corrosion wastage of
deck plates of ballast and cargo tanks varies between 100
and 0 percent.

Standard Deviation of Corrosion Depth, mm

Corrosion Depth, mm

a = 0.834
b= 1.838

1.6

80%

Cargo Tanks

1.6

Standard Deviation of Corrosion Depth, mm

1.8

Figure 15 Time dependent mean corrosion wastage of deck

plates of cargo tanks

0.9

a = 0.384
b= 0.710

0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0

12

16 20
t, years

24

28

32

36

Figure 18 Standard deviation of the yearly data of corrosion

wastage of deck plates of ballast tanks

To define the probability density function of the corrosion

wastage depth, the observed distribution is fit by a
theoretical distribution by comparing the frequencies
observed in the data to the expected frequencies of the
theoretical distribution and for that purpose the
Kolmogorov-Smirnov of fit test is applied here.

Considering that the corrosion wastage depth can be

described by a log-normal distribution function with a mean
value and standard deviation changing yearly as indicated in
Figure 14 to Figure 18, the probability density function as a
function of time of corrosion wastage for cargo and ballast
tank are given in Figure 21 and 22.

Several distributions were evaluated and it was concluded

that corrosion wastage depth is the best fit by the Lognormal distribution.

700
650
600
550

No. of observations

500

5
4.5

Log-normal Probability Density

The mean value and the variance of the long-normal

distribution for the corrosion wastage of deck plates of
ballast tanks are -0.544 and 0.919 and for the corrosion
wastage of deck plates of cargo tanks are -0.369 and 1.046
respectively (see Figure 19 and 20).

4
3.5
3
2.5
2
9.8
8.8
7.8
6.8
5.8
4.8
3.8
2.8
Corrosion
1.8
0.001
Depth, mm

1.5
1

450

0.5

400

0
10

350
300

14

18

24

32

40

Time, years

250

Figure 21 Time dependent probability density function of

corrosion wastage of deck plates of cargo tanks

200
150
100
50
0
0.0

0.5

0.9

1.4

1.8

2.3

2.7

3.2

3.6

4.1

4.5

5.0

Corrosion Depth, mm

2.4

4500

4000

3500

No. of observations

3000

2.2

Log-normal Probability Density

Figure 19 Log-normal probability density function of the

corrosion wastage of deck plates of ballast tanks for the
entire period of 32 years

2
1.8
1.6
1.4
1.2
1
9.8
8.8
7.8
6.8
5.8
4.8
3.8
2.8 Corrosion
1.8
Depth, mm
0.0

0.8
0.6
0.4
0.2

2500

0
2000

10

1500

14

18

24

32

40

Time years

Figure 22 Time dependent probability density function of

corrosion wastage of deck plates of ballast tanks

1000

500

0
0.0

0.9

1.8

2.7

3.6

4.5

5.5

6.4

7.3

8.2

9.1

10.0

Corrosion Depth, mm

Figure 20 Log-normal probability density function of the

corrosion wastage of deck plates of cargo tanks for entire the
period of 32 years

The corrosion rate defined as the first derivative of the

corrosion wastage is also analyzed, and presented in Figure
23. The initial values of corrosion rate are determined when
the corrosion process start, which is earlier for the plates of
cargo tanks in this data set. It can also be observed that the
corrosion rate is higher for cargo than for ballast tanks.

The corrosion rate is less aggressive in ballast tanks than in

cargo tanks for the first 28 years, but after that point the
contrary is true.
It is also apparent from the figure that the rate of decrease of
the corrosion rate is higher for cargo than for ballast tanks.

This work has been funded by American Bureau of Shipping

under the project Condition Assessment of Ageing Ship
Strutures.

Corrosion rate,mm/year

6
0.18

Ballast Tanks

0.16

Cargo Tanks

REFERENCES

ABS (2002), ABS Database of Corrosion Wastage for Oil

Tankers, ABS RD 2002 07.
Guedes Soares, C. 1988, Uncertainty Modeling in Plate
Buckling, Structural Safety, Vol. 5, pp. 17-34.

0.14
0.12

Guedes Soares, C. and Garbatov, Y., 1999, Reliability of

Maintained, Corrosion Protected Plate Subjected to Nonlinear Corrosion and Compressive Loads, Marine
Structures, Vol. 12, pp. 425-445.

0.10
0.08
0.06

Ivanov, L., Spencer J., and Wang G., 2003, Probabilistic

Evaluation of Hull Structure Renewals for Aging Ships,
Proceedings of the 8th International Marine Design
Conference (IMDC), Athens, Greece, pp. 393-406.

0.04
0.02
0.00
8

12

16

20
24
Time, years

28

32

36

Figure 23 Comparative analysis of corrosion rate of deck

plates of ballast and cargo tanks

ACKNOWLEDGEMENTS

CONCLUSION

Melchers, R., 1998, Probabilistic Modeling of Immersion

Marine Corrosion, Structural Safety and Reliability,
Shiraishi, N.; Shinozuka, M., and Wen, Y. K., (Eds), A. A.
Balkema, pp. 1143-1149.
Melchers, R., 2003, Probabilistic Models for Corrosion in
Structural Reliability Assessment, Part 2: Models Based on
Mechanics. Journal of Offshore Mechanics and Arctic
Engineering, Vol. 125, pp. 272-80.

The non-linear corrosion wastage proposed by Guedes

Soares and Garbatov, (1999) was used to describe the
corrosion wastage of deck plates of ballast and cargo tanks
of the ABS database. The corrosion wastage of deck plates
was examined with respect to how the corrosion wastage
varies in time. The parameters of this corrosion wastage
model were obtained.

Paik, J., Jae L., Joon H. and Young P., 2003, A TimeDependent Corrosion Wastage Model for the Structures of
Single and Double Hull Tankers and FSOs and FPSOs,
Marine Technology, Vol. 40, No. 3, pp. 201-17.

A non-linear function of time applied here describes the

growth of corrosion wastage in three different phases. In the
first phase, it was assumed that there is no corrosion. Failure
of the protection system occurs at a random point of time
and the corrosion wastage will start a non-linear growing
process with time.

Southwell, R., Bultman, D., and Hummer, Jr., 1979,

Estimating of Service Life of Steel in Seawater, Seawater
Corrosion Handbook, New Jersey, Noyes Data Corporation,
pp. 374-387.

The formulation presented here used the measurements of

corrosion wastage as input for creating an exponential
function of time that describes the effect of corrosion and
thus can be applied for the reliability assessment of different
plate elements.
The model of corrosion wastage applied to analyze the
corrosion wastage of deck plate is flexible enough to
represent realistic situations.
It was found that there were different corrosion initiation
times for plates of ballast and cargo tanks and that the
corrosion rates were different as well as the rate by which
the corrosion rates decreased with time.

Qin, S., and Cui, W., 2002, Effect of Corrosion Models on

the Time-Dependent Reliability of Steel Plated Elements,
Marine Structures, Vol. 15, pp. 15-34.

Sun, H., and Bai, Y., 2001. Time-Variant Reliability of

FPSO Hulls. Transactions SNAME, Vol. 109, pp. 341-66.
Wang, G., Spencer, J., Sun, H., 2003a, Assessment of
Corrosion Risks to Aging Ships using an Experience
Database, Proceedings of the 22nd International Conference
on Offshore Mechanics and Arctic Engineering, ASME,
Paper OMAE 2003-37299.
Wang, G., Spencer, J., Elsayed, T., 2003b, Estimation of
Corrosion Rates of Oil Tankers, Proceedings of the 22nd
International Conference on Offshore Mechanics and Arctic
Engineering, ASME, Paper OMAE 2003-37361.
Yamamoto, N., and Ikegami, K., 1998, A Study on the
Degradation of Coating and Corrosion of Ships Hull Based
on the Probabilistic Approach, Journal of Offshore
Mechanics and Arctic Engineering, Vol. 120, pp. 121-28.