Corrosion Protection
Corrosion Protection
Corrosion Protection
Corrosion Protection
l
ria
ate
dM
hte
rig
py
Co
l
ria
S. SRINIVASAN
Intercorr International Inc.
Houston, Texas
ate
A. INTRODUCTION
Information and knowledge are the foundations on which all technology has been built. Computers
dM
and computer-based information systems have revolutionized our approach to problem solving,
information access, and knowledge processing in every domain of human endeavor. Corrosion
science and engineering, too, has benefited from the application of numerous computer-based
systems and tools promoting automated data/information access and efficient problem solving.
This chapter provides an overview of the types of computer tools utilized for solving corrosion-
related problems, data storage, and data analysis. An introduction to computer-based corrosion
problem solving is followed by a description of types of computer programs employed in the domain
hte
of corrosion. This includes brief descriptions of different types of computer applications, including
expert systems, neural networks, and object-oriented software systems. A description of currently
available computer tools for modeling corrosion and cracking problems, selection of materials/
equipment as well as for corrosion management, monitoring, and control is also provided. Aspects
relevant to the role of emerging technologies and the internet as a tool for information dissemination
are also presented.
rig
Using computer tools to model and represent corrosion processes is a challenging task since
characterizing corrosion processes requires a fundamental understanding of principles underlying
multiple disciplines, from electrochemistry and fluid mechanics to material science and engineering.
The complexity of characterizing corrosion-related tasks has necessitated use of computer tools in
corrosion science and engineering, from modeling to data acquisition and analysis. Computers, in
Co
the current day environment, are an intrinsic part of both data representation and automated
problem solving. In this context, computer-based corrosion problem solving systems may be
classified as
A large number of early programs in corrosion were billed as expert systems, primarily because
the programs typically attempted to capture human expertise in corrosion [1], and these
l
programs represented research-based development efforts normally lacking rigorous software
ria
engineering foundations necessary for commercial distribution. Most of these programs were
developed using software platforms called shells [2] that supported easy implementation of
heuristic rules (rules of thumb) and representation of common concepts of reasoning. Table 1
below lists a few well-known computer programs in corrosion developed in the late eighties and
early nineties [3-8]. It is interesting to note that none of these early systems were implemented in
ate
commonly used programming languages (such as C, C + +, Fortran, etc.) and many were
implemented by corrosion/materials specialists with little or no formal training in software
development [9].
The single most popular application of computers in corrosion stems from programs built for
modeling corrosion and cracking processes, including programs for diagnosis, failure analysis and
prediction/analysis. Several computer programs have been built to model and predict corrosion,
dM
ranging from early systems such as Auscor [9] for assessment of stress corrosion cracking (SCC) in
austenitic stainless steels, to current day programs for general corrosion prediction, such as LipuCor
[10], Cormed [11] and Predict [12]. A more detailed description of some of these models are
provided in Section C.
Several computer programs and databases are available to assist with material selection and
equipment specifications. The CORIS system [13] is an expert system database that integrates
corrosion problem solving expertise with a comprehensive database on corrosion, materials, and
thermodynamic properties of corrosive media. Other programs for material evaluation and selection,
hte
some of which have had wide commercial application, provide an automated basis for materials
selection, such as Socrates [14] for oil and gas production and Selmatel [15] for selection of
materials exposed to elevated temperatures in refinery furnace tubes.
One of the most common application of computers to corrosion has been in the area of
development of databases integrated with material evaluation heuristics for corrosion and materials
data. The Materials Technology Institute [1] in the late 1980s and early 1990s developed a series of
rig
systems called Chem Cor relevant to evaluation of materials applicable to different segments of the
chemical processing industries. CORSUR [16] is another large database application on corrosion
behavior of metals and nonmetals in over 700 chemical environments.
py
Different types of software technologies have found application in corrosion engineering and
science. These programs can be broadly classified as follows:
l
Object-oriented software systems.
ria
Neural networks.
Hybrid systems that utilize one or more of the above technologies.
Most early computer programs and legacy systems (meaning older systems whose technology is
currently obsolete) developed in the 1970s and 1980s used high-level programming languages such
ate
as Basic, Fortran, Pascal, and so on. Typically, such systems provided a front end for data modeling
and analysis that was computation intensive but provided little or no support for representation of
heuristic concepts and manipulation of symbolic information.
The advent of expert systems drastically changed the direction of computer programs for
corrosion applications. Expert systems or knowledge-based systems, defined as intelligent computer
programs that use expert knowledge to attain high levels of performance in narrow problem domains
dM
[17], became quite common and prominent. The term expert in the expert systems implied the
narrow specialization and competence of a human expert embodied into the system. Expert systems
typically have had a specific structure that distinguishes them from conventional computer programs
in that the domain knowledge (knowledge base) is usually separated from the reasoning process
(inference engine) as shown in Figure 1.
The knowledge base houses the expertise that is embodied in an expert system. The inference
engine controls the manner or the logical path used by the expert system to access the information
hte
(facts and rules) in the knowledge base to make decisions. Knowledge representation in an expert
system refers to the scheme used to represent a given piece of information in the system. Decision
making in an expert system can be represented through interconnected rules (rules of thumb that
correlate facts, data, and expertise) and objects (descriptions of different system components).
Expert systems represent by far the most widely used segment of computer applications in
corrosion. Numerous expert systems have been developed to address different aspects of corrosion
[1], and include systems for applications in cathodic protection [18], assessment of stress cracking in
rig
light water nuclear reactors [19] and prediction of localized corrosion of stainless steels [2O].
With the advent of object oriented systems and the concept of reusable entities that contain both
the data and the procedures relevant to an object, the need for separation of knowledge and reasoning
py
Knowledge
Acquisition
Subsystem
Knowledge
Base
Co
Knowledge
Engineer User End
Interface User
Inference
Engine
Domain Explanation
Expert Subsystem
l
specified for the abstraction.
ria
Here is the biggest reason for using object-oriented methods for software development: Once a
certain class of objects have been created, these objects can be reused to create other objects and
properties, complete with their links to the world (interface and behavior). Such an approach is very
efficient from a software development standpoint, since the self-contained character of objects
provides an efficient basis to create and extend from existing software components. The state of
ate
being self-contained, that is, inherent to good object-based systems, is termed as encapsulation.
Inheritance allows derivation of new objects from an existing class definition, and encapsulation
promotes defining the limits of services an object can provide to other objects. Genera [22] is an
object-oriented framework for development of problem-solving systems in corrosion and
materials and has been used to develop numerous problem-solving systems. The Genera system
uses the following framework for implementation, a schematic representation for which is given
in Figure 2:
dM
Material and corrosion systems can be represented as objects, whose states can be defined in
terms of critical parameters and interrelationships between these parameters.
States of different properties of objects are encapsulated within the objects. The object-based
framework represents a class of problems in terms of abstractions of commonalities between
different domains. For example, an environment could be an environment relevant to
refineries or pipelines or a producing well. An abstract view of the environment is that of an
hte
object that can be characterized in terms of certain critical variables or operating parameters,
such as pressure, temperature, and composition etc. The variables themselves might vary.
However, the environment object can still be represented as a function of the states of certain
variables and their interrelationships.
rig
OBJECT
py
LIBRARY
Normalization
DATABASE Parani2 Paramn
l
electrochemical impedance spectroscopy data analysis [25].
ria
Neural networks perform computation in a manner quite different from that used by conventional
computers, where a single central processing unit sequentially dictates every segment of activity.
Neural networks are built from a large number of very simple processing elements that individually
deal with pieces of a big problem. A processing element (PE) simply multiplies an input by a set of
weights, and a nonlinearity function transforms the result into an output value. The principles of
ate
computation at the PE level are deceptively simple. The power of neural computation comes from
the massive interconnection among the PEs which share the load of the overall processing task, and
from the adaptive nature of the parameters (weights) that interconnect the PEs.
Normally, a neural network can have several layers of PEs. The most basic and commonly used
neural network architecture is the multilayered perceptron (MLP). Figure 3 illustrates a simple MLP.
The circles are the PEs arranged in layers. The left row is the input layer, the middle row is the hidden
layer, and the right row is the output layer. The lines represent weighted connections (i.e., a scaling
dM
factor) between PEs.
The performance of an MLP is measured in terms of a desired signal and an error criterion. The
output of the network is compared with a desired response to produce an error. An algorithm called
back-propagation is used to adjust the weights a small amount at a time in a way that reduces the
error. The network is trained by repeating this process many times. The goal of the training is to
reach an optimal solution based on the performance measurement.
Current day computer programs for corrosion assessment and control utilize one or more of the
hte
computing technologies described above. For instance, the Strategy [26] programs for evaluation of
hydrogen induced cracking in steels combine principles of expert systems and object-oriented
programming, and the work done by Silverman and Rosen [23] integrate expert systems with neural
networks.
rig
py
INPUT
LAYER
HIDDEN
Co
LAYER
(there may be several
hidden layers)
OUTPUT
LAYER
Numerous applications of computers and software technology have been reported over the last two
decades. Just in the area of expert system development, over 57 systems were reported by two
l
surveys [26] conducted by the Materials Technology Institute and the European Federation of
ria
Corrosion (EFC). If databases and other types of applications are included, the number of computer
programs/software systems for corrosion prediction and control extend well into the hundreds.
However, it would be instructive to know about systems that define the evolving trends and those that
have contributed to the continuous growth of computing technology to corrosion applications.
Table 2 provides a listing of different types of representative computer applications in corrosion
ate
currently available in the industry.
In this section, specific examples of current computer systems from different types of computing
applications in corrosion are provided. These include
Conned, a program for corrosion prediction and assessment.
Socrates, a software system for material selection for oil and gas applications.
dM
Programs for online corrosion monitoring and control.
Corsur, A database system for evaluation of metals and non-metals.
An application of neural networks to corrosion data reduction.
A computer-aided learning package for teaching corrosion.
Corsur— Corrosion data for metals and nonmetals in over 700 Relational database with a user
chemical environments friendly front end
CP Diagnostic — Trouble shooting and diagnosis of sacrificial Shell-based expert system with a
anode and impressed current cathodic protection systems database on CP data
Damage predictor — Evaluation of material performance for stress Fortran program modules that calculate
corrosion cracking in boiling water reactors and determination of radiolytic species concentration and
rig
Filter debris analysis (FDA) expert system — condition monitoring Visual basic-based interface and
of aircrafts knowledge base
Genera — Generic problem solving framework for characterizing Object oriented system implemented in
corrosion and materials problems C+ +
LipuCor— Prediction of corrosion in oil and gas systems Implemented as a conventional
structured program
Co
Predict — Prediction of corrosion in oil and gas production and Object oriented program using lab
transmission environments data and numerical relationships
Socrates — Selection of materials for oil and gas production service Object oriented expert system imple-
mented in C + + interfacing with a
database on materials and compositions
Strategy — Programs for evaluation of cracking in steels used in Implemented in C + + and integrated
pipelines and refineries with databases in Paradox
USL Corrosion model — Program for prediction of corrosion in Implemented as an expert system in
gas condensate wells visual basic
l
The Socrates system provides access to the material selection decisions and decision logic of a
ria
domain expert and significant experience on utilization and selection of corrosion-resistant alloy
(CRA) materials and steels. The system embodies information from other sources such as published
literature on lab and field experience related to oil and gas field service as well as proprietary test data
on evaluation of specific CRA material classes.
The flow of data in the SOCRATES system is shown in Figure 4. Even though a specific order of
ate
decision making is indicated by the hierarchy shown in Figure 4, since the system is object-oriented,
data specification does not have to proceed in any specific sequence. At level 1, the initial set of
applicable materials is obtained by determining the application for which the CRA is to be selected.
dM
Application
Requirements
H2S
CO2
hte
Bicarbonates
Temperature Environment
Chlorides Characterization
Sulfur
Gas to Oil Ratio
Water to Gas Ratio
rig
Evaluation for
Stress Corrosion Cracking
Alloys
Database
py
Max. Temp
Env. Severity Evaluation for
Localized Corrosion
Pitting
rules
Co
Literature Cost
Database search Analysts
Recommended
Set of CRAs
l
popular technique for obtaining instantaneous corrosion rate information. However, one limitation of
ria
EIS data is the inability of conventional techniques for validation and extension of data frequency
range. Here, artificial neural networks have been used to extrapolate impedance data to a wider range
of frequencies. Another important application of neural networks is a system that identifies the
problems associated with the inherent variability of corrosion data, and attempts to develop solutions
to these problems [24]. These solutions are considered to be general advances in neural network
ate
technology for use with noisy or poor-quality data. In essence, they depend on the production of two
further output parameters in addition to the corrosion rate. The first of these is the variance, and is
essentially the same as the conventional statistical parameter. The second has been termed the
confidence, and is an indicator of the reliability of the prediction of the neural network for a given set
of input parameters.
The Ecorr (acronym for Engineering Corrosion) [29] computer-aided learning (CAL) package
represents a novel application of computing to corrosion education. This system uses advances in
dM
multimedia technology to promote interactive, computer-aided learning. In the Ecorr approach, the
student is presented with a series of problems to solve by reference to supporting information. The
program itself is built as a string of books wherein information is presented in terms of "objects"
such as text, photographs, video, or sound objects on a page. The system consists of a theory base,
case study modules, a glossary, and control center to coordinate navigation and information flow
between the user and different modules. The theory base contains basic corrosion information that a
student can use while working with a specific case study module. The organization of the system
hte
facilitates learning by users with different levels of corrosion proficiency due to the inherent ease of
navigation. Currently the program is being designed to provide 12 case study modules at two levels
of proficiency (basic/level 1 and advanced/level 2).
Real time computer-based control systems represent an important development in application of
computing to corrosion monitoring and control. Figure 5 shows the structure of an automated
constant rate extension system (ACert) used for conducting corrosion and cracking evaluation tests.
rig
This system facilitates test initiation and conduct, and data collection and analysis from a software
system integrated with a data acquisition control system and data measurement devices. Corrosion
tests can be run with minimal human intervention, and such systems promote corrosion evaluation of
difficult to simulate environments in an automated manner.
One of the problems in conducting conventional laboratory tests stems from the fact that the time
py
required to evaluate cracking can take months of exposure. Slow strain rate (SSR) tests (also referred
to as constant extension rate tests) may be used to mechanically accelerate testing with the
application of a slow constant extension rate. Such constant extension rate tests promote speeding up
the crack initiation process and promote evaluation of material to stress corrosion cracking in a
matter of hours or days. The ACert constant extension rate tester is a computer controlled closed loop
Co
feedback system facilitating testing over a wide range of testing speeds without the need of
cumbersome gear changes. It also has an antibacklash feature for application of cyclic loads and
stresses. An ACert system can be programmed to conduct a variety of tests, including constant
extension rate tests, cyclic slow strain rate tests, fracture mechanics tests and crack growth tests.
Hitherto, computing and its applications in corrosion typically dealt with individual or group
based development efforts. The advent of the internet has fundamentally altered computing,
communication and human interaction on a global basis. Now, we have access to a new medium
where we are not bound by barriers of differing media in sharing information. Given its ability to
combine the best of media such as printed information, sound, video and computer files of different
l
User
ria
Data Data
Acquisition Acquisition Drive
Hardware Panel
ate
Parametric Inputs
(Load, Strain, T, P)
Since 1996, there has been significant growth in the use of the internet in sharing and exchanging
technical information [3O]. This growth has primarily been in the form of web sites or locations on
the internet which have been organized to provide access to different types of materials and corrosion
knowledge. The resources available to the materials/corrosion engineer/scientist fall into different
categories and span a wide range and type of relevant topics, from simple descriptions of
rig
graphics, video, and sound (multimedia) is termed the world wide web (WWW or web). In this
section, different types of internet-based resources for the corrosion engineer are described, since it
is clear now that the internet represents the future domain of computing and communication. The
term Net used here connotes the collective entity represented by the internet and the world wide web.
Co
l
News groups are the equivalent of a worldwide bulletin board (or a class room) where questions
ria
and messages can be posted, comments made, discussions monitored and answers given. There are
currently over 6000 Usenet (the name for such an open net-work entity that supports user interaction)
news groups covering a broad range of corrosion-related subjects, such as metallurgy, materials,
electrochemistry, chemistry, engineering, and nondestructive testing. Discussion threads are
conceptually similar to newsgroups; however, they may be accessed on the Web and may be used
ate
to display graphics, sounds and text, whereas newsgroups, which are hosted on a different server
protocol, do not support the range of media available on the discussion threads. Two examples
of discussion threads (which are hosted on web sites, and are more easily accessible than news-
groups) relevant to corrosion and materials are located at www.intercorr.com/discuss/ (various
topics—coatings, inhibition, cathodic protection, multiphase flow, fracture mechanics, etc.). and
www.corrosion.com (discussion and information on coatings).
Web sites (or information locations on the web) represent the single most important advancement
dM
in promoting information exchange on the net. Web sites have been created for a variety of corrosion
and materials. Web-based resources for corrosion and materials may be classified as
Resources (information/data) for solving specific technical problems.
Resources for sharing information / ideas with peers and co-workers on a world wide basis
through Virtual libraries and online databases.
Resources to support sharing of information within an organization—virtual meeting rooms,
hte
intranets.
A variety of conventional and nonconventional resources that contain technical content relevant
to corrosion are available on the Internet. These range from newsletters to technical abstracts and
technical journals to sites that provide information relevant to specific industry segments. The
following list represents a typical collection of web sites with information relevant to corrosion and
materials:
rig
The internet is poised to alter most conventional methods of information access in corrosion science
and engineering. Already, significant technical resources are available online, including,
Different Types of Technical Databases: This includes corrosion data on general, localized,
galvanic, intergranular, and erosion-corrosion and cracking data relevant to SCC, SSC,
HEC, and so on, as well as materials data on steels, stainless steels, corrosion resistant alloys,
plastics, and composites all available online.
l
as general and localized corrosion, different types of steels and elastomers, and different
ria
application segments, such as oil and gas, power generation, and nuclear industry.
Access to Conference Proceedings: The internet has significantly altered methods used by
people to access archival literature information. It is now possible to organize and provide
access to valuable literature data online and thereby obviate the conventional difficulties
experienced in disseminating current technology and information on a worldwide basis.
ate
A practical example of the virtual conference cited in literature is the first global Internet
Corrosion Conference—InterCorr/96 [33]. The proceedings of this conference are now
available as an archive at http://www.intercorr.com/IC96/. This was the first corrosion
conference where technical papers were presented, published and archived on the net,
without use of any of the conventional publication technologies.
dM
F. COMPUTERS IN CORROSION CONTROL: A CRITICAL EVALUATION
The last 10 years have witnessed explosive growth in the utilization of computers to generate,
analyze and interpret corrosion data and knowledge. This growth in computing tools has facilitated
both efficient data access/dissemination and the ability to perform tasks hitherto impossible for
computer systems, and typically performed only by human experts. For instance, expert systems for
materials selection now are capable of integrating several difficult tasks, including corrosion and
hte
cracking evaluation on a large database of materials, application of intelligent rules to determine the
best choices and providing the end user with the ability to perform on-the-fly sensitivity analyses.
Neural networks for analyzing EIS data provide the ability to learn from the data trends and generate
extrapolated impedance data in frequency ranges never seen heretofore. Object-oriented software
systems have created reusable software components that can be integrated into multiple problem
solving modules for enhanced operational and functional efficacy.
rig
The internet has not only generated the ability to provide access to an interconnected world of
data and information, but has also offered a new paradigm for information/knowledge access,
exchange, and update. Because of the ability of the web to seamlessly integrate different media as
well as create links without the constraint of a specific order (as in the case of conventional media,
such as books), the web offers a new mechanism to share and disseminate information. The impact of
py
this change in paradigm has resulted in the creation of a "virtual" interconnected world wherein we
are able to recreate every efficient aspect of conventional life and use concepts such as the virtual
database (an inter-linked hyper-database that can be seamlessly traversed), the virtual classroom
(interactive computer-based learning systems on the Net), the virtual resource room (where one can
access software programs and resources) and the virtual conference, a global meeting place for
Co
procurement decisions. While such an integration is only a possibility now, the rapidity of advances
dM
over the last two decades makes such a scenario a very probable reality in the near future.
G. SUMMARY
Numerous computer applications have been developed over the last two decades that have
significantly altered our approach to corrosion data acquisition, processing, and analysis. Every
aspect of evolution in computing technology, including databases, expert systems, object oriented
hte
software systems, and neural networks, has been applied to corrosion science and engineering. A
representative sample of such systems has been described and an overview of the relevance of such
technology has been provided.
The growth potential for computer software and hardware technologies and the Internet is
difficult to comprehend at this point in time because the pace of changes provides little time for pause
and thought. However, this technological growth comes at a cost, that we have to strive to continue to
rig
create a higher level of functionality for the displaced human effort. While current technology still
requires human intervention at many significant levels, a day in the not-too-distant future appears
probable when technological progress is accompanied by the need to create a meaningful role for the
human mind, which for all its physical shortcomings vis-a-vis a computer dominated environment, is
still eons ahead in evolution when it comes to creativity and intelligence. After all, did we not
py
H. REFERENCES
Co
1. RR. Roberge, "A critical review of expert systems for corrosion prevention and control," CORROSION/94,
Paper No. 368, NACE, Houston, TX, 1994.
2. J. P. Ignizio, "Introduction to Expert Systems: The Development and Implementation of Rule-Based Expert
Systems," McGraw-Hill, New York, 1991.
3. The Achilles Club Project—Expert system on corrosion and corrosion control, NPL, Teddington, Middlesex,
U.K., October 1986.
4. J. G. Hines and A. Basden, Br. Corros. J., 26(3), 151 (1986).
5. A. Jadot and L. Lancus, "ESCORT: Expert software for corrosion technology," Ph.D. Thesis, K. V. Leuven
University, Belgium, 1985.
l
precocious and obsolete," Paper No. 375, CORROSION/94, NACE, Houston, TX, 1994.
ria
10. Y. M. Gunaltun, "Combining research and field data for corrosion rate prediction," Paper No. 27,
CORROSION/96, NACE, Houston, TX, 1996.
11. J. L. Crolet and M. R. Bonis, "Prediction of the risks of CO2 corrosion in oil and gas wells," SPE Production
Engineering, 6(4), 449 (1991).
ate
12. S. Srinivasan, V. R. Jangama, and R. D. Kane, "Prediction of corrosivity of multiphase CO2/H2S systems,"
EUROCORR/97, The European Corrosion Congress, Trondheim, Norway, Sep. 22-25, 1997, Publ.
Norwegian University of Science and Technology, Trondheim, Norway, 1997, Vol. 1, pp. 33-40.
13. A. Hatzinosios, "CORIS: A computer-based information system for corrosion and corrosion protection,"
CORROSION/94, Paper No. 373, NACE, Houston, TX, 1994.
14. S. Srinivasan, "Role of expert systems in technology transfer of materials for petroleum applications,"
Proceedings of the 12th International Corrosion Congress, Sept. 19-24 1993, NACE International, Houston,
dM
TX, 1993.
15. A. Valdes, A. Gonzalez, R. Techorzewski, and B. Hopkinson, "An expert system for the selection of materials
exposed to elevated temperatures," Paper No. 275, CORROSION/92, NACE, 1992.
16. B. Mashayekhi, C. P. Sturrock, and C. D. Flanigan, "Corrosion data survey: The next generation," Paper No.
604, CORROSION/97, NACE International, Houston, TX, 1997.
17. F. Hayes-Roth, D. A. Waterman, and D. B. Lenat (Eds.), Building Expert Systems, Addison-Wesley, Reading
MA, 1983, pp 1-37.
hte
18. V. L. Van Blaricum, A. Kumar, and Y.-T. Park, "Cathodic protection diagnostic expert system," Paper No.
384, CORROSION/94, NACE, Houston, TX, 1994.
19. M. Urquidi-Macdonald et al., "Object oriented expert system to infer intergranular stress corrosion cracking
in light water nuclear reactors," Paper No. 340, CORROSION/90, NACE, 1990.
20. T. J. Hakkarainen, "A model for prediction of possibility of localized corrosion attack of stainless steels,"
CORROSION/96. Paper No. 363, NACE, Houston, TX, 1996.
21. I. Pohl, Object-Oriented Programming Using C + + , The Benjamin Cummings Publishing Co., Redwood
rig
24. R. A. Cottis, I. Helliwell, and M. Turega, "Neural networks for corrosion data reduction," CORROSION/96,
Paper No. 379, NACE, Houston, TX, 1996.
25. M. Urquidi-MacDonald and P. C. Egan, "Validation and extrapolation of electrochemical impedance
spectroscopy data analysis," In Corrosion Reviews: Special Issue on Applications of computers in corrosion,
Co
l
33. S. Srinivasan and R. D. Kane, "Materials and Corrosion Resources on the Internet," 2nd NACE Latin
ria
American Corrosion Congress, Rio de Janeiro, Brazil, September, 1996.
ate
dM
hte
rig
py
Co