Canadian Metallurgical Quarterly

The Canadian Journal of Metallurgy and Materials Science

ISSN: 0008-4433 (Print) 1879-1395 (Online) Journal homepage: http://www.tandfonline.com/loi/ycmq20

Estimation of Corrosion Rate of Mild Steel in Sea

Water and Application of Genetic Algorithms to
Find Minimum Corrosion Rate

Subir Paul

To cite this article: Subir Paul (2010) Estimation of Corrosion Rate of Mild Steel in Sea
Water and Application of Genetic Algorithms to Find Minimum Corrosion Rate, Canadian
Metallurgical Quarterly, 49:1, 99-106, DOI: 10.1179/cmq.2010.49.1.99

To link to this article: http://dx.doi.org/10.1179/cmq.2010.49.1.99

Published online: 18 Jul 2013.

Submit your article to this journal

Article views: 4

View related articles

Citing articles: 1 View citing articles

Full Terms & Conditions of access and use can be found at

http://www.tandfonline.com/action/journalInformation?journalCode=ycmq20

Download by: [Orta Dogu Teknik Universitesi] Date: 14 April 2016, At: 16:33
 99

Canadian Metallurgical Quarterly, Vol 49, No 1 pp 99-106, 2010

© Canadian Institute of Mining, Metallurgy and Petroleum
Published by Canadian Institute of Mining, Metallurgy and Petroleum
Printed in Canada. All rights reserved

Estimation of Corrosion Rate of Mild Steel in Sea Water and

Application of Genetic Algorithms to Find Minimum Corrosion Rate
Subir Paul

Department of Metallurgical and Material Engineering

Jadavpur University, Kolkata-700032, India
spaulxx@ymail.com

(Received in revised form December, 2009)

Abstract — Life prediction of steel structure in marine environment is a challenging task for
Downloaded by [Orta Dogu Teknik Universitesi] at 16:33 14 April 2016

design and corrosion engineers for existing structures as well as new structures to be constructed.
The rate of marine corrosion is a complex function of many parameters of fluid properties and
chemistry of seawater. The major parameters influencing the rate are salinity, sulphate,
bicarbonates, pH and temperature and dissolved oxygen. These parameters have a wide range of
variation, depending on geographical locations of the structure as well as various depths of
seawater in which parts of the structures are submerged. The inter-relationship among the
parameters make it very difficult to establish a mathematical equation. It is also interesting to find
a geographical location where rate of degradation of structure will be minimal. In the present
investigation, endeavours have been made to estimate the corrosion rate for a given geographical
location in an ocean with any combinations of these variables, by performing few laboratory
experiments. Genetic algorithms have been applied with a combination of five genes: chloride,
sulphate, dissolved oxygen, pH and temperature to obtain the optimum combination of these five
parameters that would bring down the corrosion rate of steel to a minimum.

Résumé — La prédiction de la longévité d’une structure d’acier dans un environnement marin est
une tâche difficile pour les ingénieurs concepteurs et corrosionnistes, pour les structures existantes
aussi bien que pour les nouvelles structures à construire. La vitesse de corrosion marine est une
fonction complexe de plusieurs paramètres de propriétés des fluides et de la chimie de l’eau de
mer. Les paramètres majeurs influençant la vitesse sont la salinité, le sulfate, les bicarbonates, le
pH, la température et l’oxygène dissous. Ces paramètres ont une grande gamme de variation,
dépendant de la localisation géographique de la structure ainsi que des profondeurs variées d’eau
de mer dans laquelle les composantes des structures sont submergées. L’interrelation entre les
paramètres rend très difficile l’établissement d’une équation mathématique. On est également
intéressé à trouver un lieu géographique où la vitesse de dégradation de la structure sera minimale.
Dans la présente étude, on a entrepris d’estimer la vitesse de corrosion d’un lieu géographique
donné dans un océan avec toutes les combinaisons de ces variables, en effectuant quelques
expériences de laboratoire. On a utilisé des algorithmes génétiques avec une combinaison de cinq
gènes: chlorure, sulfate, oxygène dissous, pH et température afin d’obtenir la combinaison
optimale de ces cinq paramètres qui abaisserait au minimum la vitesse de corrosion de l’acier.

Key words: Chloride, sulphate, pH, Dissolved Oxygen, Temperature, Corrosion rate, Genetic
algorithm,

INTRODUCTION Chloride ion (Cl-) has a strong influence on the corrosivity of

structural steel. The chloride (Cl-) concentration of sea water
The geographical variation in the corrosivity of natural varies from about 5.8 g/Kg to about 24 g/Kg, sulphate (SO42-)
seawater results from the variation of salinity, temperature, concentration varies from 0.8 g/Kg to 3.4 g/Kg and the
dissolved oxygen and the presence of various other salts. bicarbonate (HCO3-) concentration varies from 0.01 g/Kg to

Canadian MetallurgiCal Quarterly, Vol 49, no 1

 100 Subir Paul

0.2 g/Kg [1] across different oceans and seas. An increase of surface defects that could serve as pit sites. Following this, the
temperature and dissolved oxygen [O] concentration is known samples were washed with acetone and then left to be air-dried.
to aggravate the degradation rate of the structural material. The samples were observed with a low magnification
Though the pH of seawater is in the range of neutral to slightly microscope to observe whether there were any pit or deep
alkaline, local acidity is developed due to corrosion products as scratches on the polished surface. If any deep scratches or pits
well as crude petroleum products, which results in a lowering were observed, the samples were further polished. This
of pH to around pH 4. It is interesting to find the effect of procedure continued until the samples became pit free.
sulphate (SO42-), which also influences corrosion rate of steel. Fresh solutions were prepared before each experiment,
There have been some laboratory studies [2] and field using KCl, CaCl2, NaHC03 and MgSO4 salts in double
studies [3,5] of corrosion of mild steel in seawater. Melchers distilled water with various combinations of concentration of
and Jeffrey [3] conducted field tests at Swansea Public Wharf chloride, sulphate, dissolved oxygen contents, temperature
and at Pelican Marina [3] and reported 0.13 mm/yr as the rate and pH required for the experiments for factorial design (at
of material loss at the Swansea Public Wharf. Sridhar et al. upper and lower limits, Table I) and further experiments at
[6] developed a model to predict localized corrosion of steel other different combinations for validation of the model
in sea water. They predicted the repassivation potential and (Appendix Table AI).
corrosion potential of stainless steel. Temperature is also A standard corrosion cell was used to perform the
known to have a strong effect on marine corrosion of carbon electrochemical potentiostatic polarization tests on standard
Downloaded by [Orta Dogu Teknik Universitesi] at 16:33 14 April 2016

steel. Melchers [7] built a model on the corrosion rate of mild flat metal specimens. Polarization experiments were carried
steel from the field data of salinity, PH, sulphate concen- out as per ASTM ST72 using Gamry Potentiostat. The
tration and dissolved oxygen at different geographical software used was Gamry Echem Analyst. The potentio-
locations. Free et al. [8.9] developed a mathematical model dynamic experiment was done with the samples; the
based on thermodynamics, kinetics and mass transport corrosion current and corrosion potential were determined by
equations to predict corrosion of steel in aqueous media with Tafel’s extrapolation method as well as linear polarization
the variations of solution compositions method. A software was developed based on estimated
In the present investigation, endeavours have been made equations and 3D mapping of corrosion with variables was
to estimate the corrosion rate of mild steel for a given generated by the software.
geographical location in an ocean with any combination of
those variables by performing a few laboratory experiments
in artificially prepared seawater. The full range of salt FACTORIAL DESIGN OF EXPERIMENT(
constituents, temperature, pH and dissolved oxygen were
selected for a series of experimentations based on 2k factorial Factorial experiment design [10] is a systematic method for
design of experiments [10]. A genetic algorithm [11] has been formulating the steps needed to successfully implement
applied with combinations of five genes: chloride, sulphate, experiments that investigate the effects of two or more factors
dissolved oxygen, pH and temperature to obtain the or input parameters on the output response with a minimal
combination of these five parameters producing the minimum number of observations. A two level factorial design of
corrosion rate of steel. experiment means two levels of each factor will be studied at
once. If there are k factors or input variables influencing the
output, a minimum 2k number of experiments are needed to
EXPERIMENTAL PROCEDURE be run to evaluate the factors affecting the process or output.
Each factor will have two levels, a ‘high’ and ‘low’ level. In
Rectangular samples of 1 ¥ 2 cm2 were cut from low carbon this investigation five factors have been selected with a two
steel. They were ground carefully in order to make the edges level factorial design of experiment as illustrated in Table I.
blunt and to give a near rectangular cross section. The samples
were then further ground using a belt drive and polished up to Let Zmax
j = Upper limit of the variable j and Zmin
j = Lower limit
3/0 emery paper. Polishing provides a uniform surface removing of the variable j

Table I – Composition of different variables in 25 factorial design of experiments

Variables Reagents Limit Upper Limit Lower Average for

Zmax
j Zmin
j Z0j

X1 Chloride 24 gm/L 15 gm/L 19.5 gm/L

X2 Sulphate 3.4 gm/L 0.8 gm/L 2.1 gm/L
X3 PH 8 4 6
X4 Temperature 40°C 10°C 25°C
X5 Dissolved Oxygen 10 ppm 2 ppm 6 ppm

Canadian MetallurgiCal Quarterly, Vol 49, no 1

 Estimation of Corrosion Rate of Mild Steel in Sea Water and Application of Genetic Algorithms to Find Minimum Corrosion Rate 101

Z0j = (Zjmax + Zjmin)/2 (1) performed, offspring is an exact copy of the parents.
-Mutation: With a mutation probability new offspring
DZj = (Zjmax - Zjmin)/2 (2) mutate at each locus (position in chromosome).
-Accepting: Place new offspring in a new population.
where Z0j is the centre point of design or basic level and DZj 4. Use a new generated population for a further run of the
is the interval of variation on the Zj axis. It is usual to change algorithm
the coordinate system of Z1 , Z2 … Zj to a new dimensionless 5. Test: If the end condition is satisfied, stop and return the
system of coordinates of X1, X2, … Xj so that the best solution in the current population or else, (Loop) Go
dimensionless variable Xj is related to Zj by the following to step 2
equation A detailed description of the algorithms can be found in Refs.
[11,12] and online literatures [13-15].
Xj = (Zj + Zj0 )/DZj (3)

where j = 1, 2 ........k. Here k = 5. RESULTS AND DISCUSSIONS

Using this factorial design, the regression equation is The regression equation of the form of Equation 4 is
given as follows generated from the experimental data (Appendix Table AI)
Downloaded by [Orta Dogu Teknik Universitesi] at 16:33 14 April 2016

and subsequent computation is based on the student T test

Y= f (bj,Xj)= b0 + b1X 1 + b2X2 + ..... + b12X 1 X2 with a 95% confidence limit and an adequately fit F distri-
(4) bution test with a 99% confidence limit. The software was
+ ..... + b123X1X2X3 + ..... + b12345X1X2X3X4X5
built using C++ programming to generate a 3D corrosion
where b is the interaction coefficient and X is the mapping, based on a regression equation.
dimensionless variables
The significance of the interaction coefficients is tested Regression Equation for Corrosion Rate of Mild Steel:
by a student t distribution with a 95% confidence limit. The
regression equation is adequately fit by a F distribution test Icorr = 64.225 + 28.05*((Cl  19.5)/4.5)
with 99% confidence limit. + 1.43*((pH  6)/2) + 27.03*((T  25)/15)
+ 20.96*(([O]  6)/4)
– 3.71*((SO4  2.1)/1.3)*((pH  6)/2)
GENETIC ALGORITHMS – 2.09*( (Cl  19.5)/4.5)*((pH  6)/2)
– 4.66*((Cl  19.5)/4.5)*((SO4  2.1)/1.3)* (5)
Genetic algorithms [11] have been widely applied in many ((pH  6)/2) + 20.89*((Cl  19.5)/4.5)*
fields in the engineering world. Not only does GAs provide an ((T  25)/15)
alternative method to solving problem, but they also consis- + 18.98*((T  25)/15)*(([O]  6)/4)
tently outperform other traditional methods in most of the + 11.17*((Cl  19.5)/4.5)*(([O]  6)/4)
problems. It is one of the best ways to solve a problem of + 11.65*((Cl  19.5)/4.5)*
which little is known. An algorithm is started with a set of ((T  25)/15)*(([O]  6)/2)
solutions (represented by chromosomes) called population.
Solutions from one population are taken and used to form a The following figures display a 3D corrosion mapping of
new population to find a better one than the old one. Solutions mild steel with effect of several parameters.
that are selected to form new solutions (offspring) are It is seen in Figure 1 that the corrosion rate increases
selected according to their fitness - the more suitable they are, sharply with increasing chloride (Cl-) concentration, while an
the more chances they have to reproduce. The algorithm increase of pH enhances the corrosion rate marginally. Figure
consists of the following steps: 2 illustrates the effect of chloride (Cl-) and sulphate (SO42-) on
1. Generate a random population of n chromosomes corrosion rate. It was observed that chloride (Cl-) had a very
(suitable solutions for the problem) strong influence in enhancing corrosion while the sulphate
2. Evaluate the fitness f(x) of each chromosome x in the (SO42-) ion decreased the corrosion rate.
population Figure 3 is interesting in that it shows the effect of pH
3. Create a new population by repeating the following steps: and sulphate (SO42-) ion concentration at a fixed chloride (Cl-)
-Selection: Select two parent chromosomes from a concentration. It shows that the effect of the increment of pH
population according to their fitness. In a roulette wheel and sulphate (SO42-) jointly bring down corrosion, whereas
selection, individuals are given a probability of being individually each of them has an effect of enhancing the
selected that is directly proportional to their fitness (the corrosion rate. This is due to the formation of calcareous
better fitness, the bigger the chance to be selected) deposit on the steel surface under favourable conditions of
-Crossover: With a crossover probability, cross over of high sulphate (SO42-) ion concentration in the alkaline region.
the parents to form a new offspring. The most common Figure 4 depicts the effect of chloride (Cl-) ion concen-
type is single point crossover. If no crossover is tration and temperature at a fixed [O] concentration. The

Canadian MetallurgiCal Quarterly, Vol 49, no 1

 102 Subir Paul

diagram shows that while both temperature and Cl- have an It is seen that the concentration of dissolved oxygen O2
influence on the enhancement of corrosion rate, the increase enhances the rate of cathodic reaction. It is known from the
is sharp with a variation of Cl- at higher temperature, but at polarization diagram of low carbon steel in an aerated
lower temperature, the effect is marginal. At higher aqueous environment [16] that the anodic line cuts at the
temperature, the mobility of chloride ion is rapid. So the concentration polarization region of the cathodic polarization
migration of ions from the bulk to the interface of metal- curve at the Ecorr  Icorr point. The rate of cathodic reaction is
electrolyte increases the local concentration of it helping controlled by IL, the limiting current density, which is a strong
degradation at a higher rate. The effect of chloride (Cl-) and function of concentration and diffusion of oxygen from the
dissolved oxygen [O] content together at a fixed temperature bulk solution to the interface. These explain the enhancement
25°C is shown in Figure 5. It is seen that the effect of chloride of corrosion rate with an increase in dissolved oxygen
concentration on increasing corrosion rate is greater than the concentration. The limiting current density is also greatly
effect of dissolved oxygen [O]. influenced by temperature and agitation, both of which
The electrochemical corrosion of plain carbon steel in increased the diffusion rate. This is due to the fact that the
an aqueous environment such as seawater takes place by the corrosion rate of mild steel increases at higher temperature
following anodic and cathodic reactions. (Figure 4). Chloride ion increases the rate of the 1st reaction
(anodic) by forming metal chloride, which hydrolyzes to
2Fe Æ 2Fe2+ + 4e Anodic generate H+ ions, increasing local acidity that further
Downloaded by [Orta Dogu Teknik Universitesi] at 16:33 14 April 2016

02 + 2H20 + 4e Æ 40H- Cathodic aggravates the dissolution process. Finally Figure 6 displays
the effect of dissolved oxygen [O] and temperature together

Corrosion rate microA/cm2

Corrosion rate microA/cm2

pH
pH Chloride ion conc. g/L Sulphate ion conc. g/L

Fig. 1. Corrosion rate of mild steel in artificial sea water as a function of Fig. 3. Corrosion rate of mild steel in artificial sea water as a function of pH
chloride conc. and pH at 1 g/L sulphate ion conc. dissolved oxygen conc. 5 ppm and sulphate ion conc. at 20 g/L chloride conc, dissolved oxygen conc. 5 ppm
and temp. 25°C. and temp. 25°C.
Corrosion rate microA/cm2

Sulphate ion conc. g/L Temperature °C

Chloride ion conc. g/L Chloride ion conc. g/L

Fig. 2. Corrosion rate of mild steel in artificial sea water as a function of Fig. 4. Corrosion rate of mild steel in artificial sea water as a function of
chloride conc. and sulphate conc. at pH 8, dissolved oxygen conc. 5 ppm and chloride conc. and temperature at 1.7 g/L sulphate ion conc., dissolved
temp. 25°C. oxygen conc. 5 ppm and pH 8.

Canadian MetallurgiCal Quarterly, Vol 49, no 1

 Estimation of Corrosion Rate of Mild Steel in Sea Water and Application of Genetic Algorithms to Find Minimum Corrosion Rate 103

at fixed chloride (Cl-) ion concentration. Their combined in the ocean (Swansea Wharf) with an average composition of
effect on increasing corrosion rate is more pronounced than 29.8 to 34.9 gm salinity, 2.4 g/L SO42- and pH 8 is 0.471
their individual effects. mm/year. With this composition of seawater, the corrosion
In the present investigation, the strong effect of chloride rate comes out from the derived equation in the present
(Cl-) ion on corrosion rate of mild steel in artificially made investigation is 0.435 mm/year. It is interesting to find the
seawater has been clearly revealed. It is also interesting to computed corrosion rate is very close to one determined in
note the conjoint effects of pH and sulphate (SO42-) on field test at actual geographical location.
corrosion rate. The effect of temperature and dissolved
oxygen ([O]) content on corrosion rate of mild steel has also
been found to support the electrochemical theory and Genetic Algorithms to Find Minimum Corrosion Rate
polarization of steel in aqueous system ]16]. A genetic algorithm was applied to obtain a best fit
chromosome from the combination of five genes viz chloride,
sulphate, PH, temperature and dissolved oxygen content in
Prediction of Corrosion Rate order to minimize the objective function (Equation 5) which
Attempts have been made to predict the corrosion of real represents the rate of corrosion as complex functions of all
structures submerged in seawater at geographical locations parameters. Each gene length is a six digit coded binary
from the derived equation. number and the chrome length is a 30 digit binary. The
Downloaded by [Orta Dogu Teknik Universitesi] at 16:33 14 April 2016

It is established from the literature [3,7] that the decoded values of genes are subject to the constraints, the
corrosion rate of low carbon structural steel at a specific site minimum and maximum decoded values of gene Cl- 15-24,
gene SO42- 0.8 – 3.4 gene [O] 5-12, gene PH 4-8, gene temp
10-40.
The algorithm consists of the following steps:
1. Each chromosome is tested to see how good it is in
solving the problem at hand and accordingly it is
Corrosion rate microA/cm2

assigned a fitness score. The fitness score is a measure

of how good the chromosome is in minimizing objective
function.
2. There are 600 populations chosen in the program and
each population consists of 2000 generations of
chromosomes. Two members from the current
population are selected. The chance of being selected is
proportional to the chromosome fitness. A roulette
wheel selection [15] was used. A pie chart or roulette
Dissolved oxygen conc. ppm. Chloride ion conc. g/L
wheel represents the total fitness score. A slice of the
wheel is assigned to each member. The size is propor-
Fig. 5. Corrosion rate of mild steel in artificial sea water as a function of tional to the chromosome fitness score, that is the fitter
dissolved oxygen conc. and chloride ion conc. at 1.7 g/L sulphate ion conc., the member the bigger the size it gets.
dissolved oxygen conc. 5 ppm pH 8 and remp. 25°C. 3. Two chromosomes at a time are taken for CROSSOVER,
depending on the crossover rate. The crossover bits are
selected from each chosen chromosome at a randomly
chosen point. Here two chromosomes will swap their
bits by selecting a random gene along the length of the
chromosomes and swapping all the genes after that
Corrosion rate microA/cm2

point. A crossover rate of 0.7 was chosen here.

1000011000101
e.g. Given two chromosomes
0101000110010
Swap at the line
Swap at the line

4. Next in the mutation, bits within a chromosome will be

flipped (0 becomes 1, 1 becomes 0). The mutation rate
Dissolved oxygen conc. ppm.
Temperature °C
has been taken at a very low value of 0.001.
5. Steps 2, 3 and 4 are repeated, until a new population is
Fig. 6. Corrosion rate of mild steel in artificial sea water as a function of created.
dissolved oxygen conc. and temp at 20 g/L chloride ion conc.,1.7 g/L The results of the genetic algorithm are illustrated as
sulphate ion conc., dissolved oxygen conc. 5 ppm and pH 8. follows:

Canadian MetallurgiCal Quarterly, Vol 49, no 1

 104 Subir Paul

Best fit chromosome found = maximum life based on knowledge of the constituents of
010001000111000011001101000101 ocean chemistry, pH and temperature. The chemistry of
No. of generation: 586 cooling water in contact with the condenser pipes and
Population: No. 2 accessories in power plant, near marine environment can be
Gene combinations are manipulated to restrict the corrosion rate at the minimum.
Gene No. 1: (Cl-) Gene (Binary)010001 = (Decoded) 17
gm/L
Gene No. 2: (PH) Gene (Binary)000111 = (Decoded) 7 CONCLUSIONS
Gene No. 3: (SO42) Gene (Binary)000011 = (Decoded)
3gm/L A prediction of corrosion rate and hence life of a marine
Gene No. 4: (Temp) Gene (Binary)001101 = (Decoded) structure can be computed at any geographical location given
13°C the composition of ocean water. The combined effects of Cl-,
Gene No. 5 [O] Gene (Binary)000101 = (Decoded) 5 ppm SO42-, [O], pH and temperature on the rate of degradation of
Minimum corrosion rate Icorr= 16.0225 mA/cm2 mild steel in seawater are better understood. Application of a
It is seen from the above results that the first population genetic algorithm to find minimum corrosion rate produces
of 2000 generations could not produce the required the optimum values of the combinations of five genes:
chromosome. In the second population, after 586 numbers of chloride, sulphate, dissolved oxygen, pH and temperature.
Downloaded by [Orta Dogu Teknik Universitesi] at 16:33 14 April 2016

generations, the best gene combinations are found. Thus an This helps to manipulate the constituents of a marine
aqueous marine environment of 17 gm/L chloride concen- corrosive environment of recirculating cooling water to
tration, 3gm/L sulphate concentration and 5 ppm dissolved maximize the life of the condenser and accessories in a
oxygen content having a pH of 7 at a temperature of 13°C powerplant near seaside.
gives a minimum corrosion rate of 16.0225 mA/cm2 to a mild
steel structure. Experiments were conducted with these
constituents, pH and temperature and the rate of corrosion REFERENCES
obtained was 18.28 mA/cm2 (Table AI Appendix). So the
experimental findings closely matched the minimum value 1. Dechema Handbook of Corrosion, 1992, vol. 11, Weinheim, NY: VCH
Publishers p. 66.
obtained from the genetic algorithm. It is interesting to note 2. U.R. Evans, The Corrosion and Oxidation of Metals: Scientific
that minimum corrosion rate occurs with the sulphate concen- Principles and Practical Applications, 1960, Edward Arnold Ltd.,
tration on the higher side. That is, the sulphate ions helped in London.
reducing corrosion rate. This may be due to the formation of 3. R.E. Melchers and R Jeffrey, “Influence of Water Velocity on Marine
some protective sulphate deposits on the steel surface at pH 7. Corrosion of Mild Steel”, Corrosion (NACE), 2004, vol. 60, pp. 84-94.
4. L.A. Terry and R.G.J Edyvean, “Influences of Microalgae on Corrosion
While dissolved oxygen content comes at a minimum value of Structural Steel”. J R Lewis and A D Mercer, eds, Corrosion and
of the permissible limit of 5 (Table I), the temperature comes Marine Growth on Offshore Structures, 1984, pp. 38-44.
at a slightly higher side of 13°C than its minimum permissible 5. B.D. Craig, Handbook of Corrosion Data, ASM ,international, Ohio,
value of 10°C and chloride content comes out a little higher 1989.
than its minimum value of 15°C. Equation 5, based on which 6. N. Sridhar, C.S. Brossia, D.S. Dunn and A. Anderko, “Predicting
Localized Corrosion in Seawater”, Corrosion (NACE), 2005, p. 915.
genetic algorithm of minima has been computed, has positive 7. R.E. Melchers, “Effect of Temperature on the Marine Immersion of
and negative coefficients of the parameters and interacting Corrosion of Carbon Steels”, Corrosion (NACE)”, 2002, p. 762.
parameters, which suggest that dependence of corrosion rate 8. M.L. Free, “Tri Service”, Corrosion Conference, 2005, p. 1.
on those variables is not straight forward, but complex and 9. M.L. Free, W. Wang, D.Y. Ryu, “Prediction of Corrosion Inhibition
Using Surfactants”, Corrosion, 2004, vol. 60, p. 837.
interrelated. 10. Douglas C. Montgomery, Design and Analysis of Experiments, 1988,
Application of the genetic algorithm in finding 2nd Edition, John Wiley and Sons, New York.
minimum corrosion rate helps to better understand the 11. Melanie Mitchell, Introduction to Genetic Algorithms, Pub: MIT Press
corrosion mechanism why an optimum combination of all 1998, 1st ed.
12. G. Winter, J. Periaux and M. Galan, Genetic Algorithms in Engineering
influencing parameters viz. Cl-, SO42-, [O], pH and and Computer Science, 1995, John Wiley & Son Ltd., 1995.
temperature helps to retard the rate of anodic or cathodic or 13. http://ai-depot.com/knowledge/genetic_algorithm
both the reactions and brings them to a minimum. 14. http://www.obitko.com/tutorials/genetic-algorithms/ga-basic-
Additionally it is also helpful in finding the geographical description.php
location of the site where new marine offshore structure can 15. http://www.ai-junkie.com/ga/intro/gat1.html
16. D.A. Jones, Principles and prevention of Corrosion, 1996, 2nd ed.,
be installed that would give minimum corrosion rate and Prentice Hall, p. 92.

Canadian MetallurgiCal Quarterly, Vol 49, no 1

 Estimation of Corrosion Rate of Mild Steel in Sea Water and Application of Genetic Algorithms to Find Minimum Corrosion Rate 105

APPENDIX

EXPERIMENTAL RESULTS

Table A-I
Experimental data of corrosion rate of mild steel in artificial sea water
with variation of water chemistry, pH and temperature
X1 = Cl conc. g/L, X2 = pH., , X3 = SO4 conc. g/L X4 = Temperature °C , X5 = Dissolved [O] conc. ppm
Z = corr rate microA/cm2

Expt. No. X1 X2 X3 X4 X5 Z

1 15 4 0.8 25 6 22.2725
2 15 4.8 0.8 25 6 22.0325
3 15 5.6 0.8 25 6 21.7925
4 15 6.4 0.8 25 6 21.5525
5 15 7.2 0.8 25 6 21.3125
Downloaded by [Orta Dogu Teknik Universitesi] at 16:33 14 April 2016

6 15 8 0.8 25 6 21.0725
7 15 4 0.8 25 6 22.2725
8 15 4 1.32 25 6 21.5135
9 15 4 1.84 25 6 20.7545
10 15 4 2.36 25 6 19.9955
11 15 4 2.88 25 6 19.2365
12 15 4 3.4 25 6 18.4775
13 15 4 3.92 25 6 17.7185
14 15 4 0.8 25 6 22.2725
15 16.8 4 0.8 25 6 28.1405
16 18.6 4 0.8 25 6 34.0085
17 20.4 4 0.8 25 6 39.8765
18 22.2 4 0.8 25 6 45.7445
19 24 4 0.8 25 6 51.6125
20 25.8 4 0.8 25 6 57.4805
21 24 8 3.4 25 6 37.5275
22 24 7.2 3.4 25 6 47.0395
23 24 6.4 3.4 25 6 56.5515
24 24 5.6 3.4 25 6 66.0635
25 24 4.8 3.4 25 6 75.5755
26 24 4 3.4 25 6 85.0875
27 24 8 3.4 25 6 37.5275
28 24 8 2.88 25 6 44.2225
29 24 8 2.36 25 6 50.9175
30 24 8 1.84 25 6 57.6125
31 24 8 1.32 25 6 64.3075
32 24 8 0.8 25 6 71.0025
33 24 8 0.28 25 6 77.6975
34 24 8 3.4 25 6 37.5275
35 22.2 8 3.4 25 6 34.9955
36 20.4 8 3.4 25 6 32.4635
37 18.6 8 3.4 25 6 29.9315
38 16.8 8 3.4 25 6 27.3995
39 15 8 3.4 25 6 24.8675
40 13.2 8 3.4 25 6 22.3355
41 15 8 2.1 10 2 33.32
42 15 8 2.1 10 3.6 35.328
43 15 8 2.1 10 5.2 37.336
44 15 8 2.1 10 6.8 39.344
45 15 8 2.1 10 8.4 41.352

Canadian MetallurgiCal Quarterly, Vol 49, no 1

 106 Subir Paul

Table continued from previous page

Expt. No. X1 X2 X3 X4 X5 Z

46 15 8 2.1 10 10 43.36
47 15 8 2.1 10 11.6 45.368
48 15 8 2.1 10 2 33.32
49 15 8 2.1 16 2 32.404
50 15 8 2.1 22 2 31.488
51 15 8 2.1 28 2 30.572
52 15 8 2.1 34 2 29.656
53 15 8 2.1 40 2 28.74
54 15 8 2.1 40 2 27.824
55 15 8 2.1 10 2 33.32
56 16.8 8 2.1 10 2 31.532
57 18.6 8 2.1 10 2 29.744
58 20.4 8 2.1 10 2 27.956
Downloaded by [Orta Dogu Teknik Universitesi] at 16:33 14 April 2016

59 22.2 8 2.1 10 2 26.168

60 24 8 2.1 10 2 24.38
61 25.8 8 2.1 10 2 22.592
62 24 8 2.1 40 10 344.6
63 24 8 2.1 40 8.4 294.426
64 24 8 2.1 40 6.8 244.252
65 24 8 2.1 40 5.2 194.078
66 24 8 2.1 40 3.6 143.904
67 24 8 2.1 40 2 93.73
68 24 8 2.1 40 10 344.6
69 24 8 2.1 34 10 281.79
70 24 8 2.1 28 10 218.98
71 24 8 2.1 22 10 156.17
72 24 8 2.1 16 10 93.36
73 24 8 2.1 10 10 30.55
74 24 8 2.1 40 10 344.6
75 22.2 8 2.1 40 10 295.084
76 20.4 8 2.1 40 10 245.568
77 18.6 8 2.1 40 10 196.052
78 16.8 8 2.1 40 10 146.536
79 15 8 2.1 40 10 97.0201
80 13.2 8 2.1 40 10 47.5041
81 19.5 6 3.1 25 6 24.10
82 19.5 6 3.1 25 6 25.35
83 19.5 6 3.1 25 6 29.85
84 19.5 6 2.1 25 6 34.81
85 19.5 6 2.1 25 6 31.39
86 19.5 6 2.1 25 6 31.03
87 17 7 3 13 5 18.28

Canadian MetallurgiCal Quarterly, Vol 49, no 1