1

1
Corrosion monitoring whats the point?
Bob Cottis
University of Manchester, School of Materials, P.O. Box 88,
Sackville Street, Manchester M60 1QD, UK
bob.cottis@manchester.ac.uk
1.1 Introduction
The paper presents a series of largely philosophical musings on the value, practise and
processes of corrosion monitoring.
When I teach corrosion monitoring, I start by asking the students four questions:
Does corrosion monitoring reduce the rate of corrosion?
Does corrosion monitoring reduce the cost of corrosion?
Does corrosion monitoring reduce the hazards of corrosion?
Does corrosion monitoring reduce the environmental impact of corrosion?
The response of the students varies a little from year to year, but in general, the
majority answer yes to these questions. You may wish to consider your own answers
so that you are not tempted to read the answers before doing this, this section is
continued at the end of this article.
1.2 Selection of monitoring method
The ideal corrosion monitoring method would give an accurate, real-time, online
indication of many properties of the system, and it would thus have all of the
following capabilities:
Instantaneous, online measure of corrosion rate
Instantaneous, online measure of localised corrosion type(s), location and pene-
tration rate
Online measure of integrated uniform metal loss (i.e. measure of the remaining
thickness)
Online measure of total local corrosion penetration (i.e. pit depth)
Location of points of large corrosion penetration
Detection, location and sizing of SCC and similar defects
Measure behaviour of plant (not a probe).
Of course, there is currently no such method, and corrosion monitoring necessarily
involves a series of compromises. First, we will consider the compromises involved in
the use of probes.
1.3 Probes for corrosion monitoring
Many corrosion monitoring techniques (e.g. all electrochemical and weight loss
methods and most electrical resistance methods) depend on the use of probes (or test
2 Corrosion monitoring in nuclear systems
coupons, for which very similar considerations apply), rather than making measure-
ments of the actual material of the plant. This is convenient for several reasons,
including the easy replacement of probes, the ability to test alternative materials and
the fact that many of the simpler monitoring methods require the use of a probe.
However, it also introduces a number of problems:
The material of the probe is generally different from that of the plant. While it
may be possible to make probes out of the same batch of material as the plant, it
is much more commonly the case that it is from a different batch and probably
supplied in a different form (e.g. rod rather than plate). Thus it is likely to have a
different chemical composition and metallurgical structure. Even if it is possible
to use the same batch of material as that from which the plant is made, it is likely
that the surface presented to the environment will be different from that exposed
in the plant (e.g. the probe will be a machined cylinder, whereas the exposed
surface of the plant will be the as-received surface of the plate).
The stress state of the probe will be different from that of the plant.
The probe will be exposed to only one sample of the range of environments to
which the plant is exposed. This is particularly important in multi-phase systems,
where completely different behaviours can be expected for surfaces exposed in the
different phases or mixtures of phases. It is not always clear that two phases are
present. Thus hydrocarbon pipelines may contain traces of water that will drop
out in low regions of the pipe, or possibly separate out onto hydrophilic surfaces
(such as the oxidised region adjacent to welds). The regions in contact with water
will typically be the sites of corrosion problems, and it may be very difcult to
install probes that will reliably detect these problems.
The probe will typically experience different ow conditions from the plant. Very
often the probe will create a ow disturbance; this may lead to an increased
corrosion rate, but equally it may inhibit pitting corrosion.
The area of the probe is usually much smaller than that of the plant, and many
monitoring methods require that the probe is electrically isolated (and there are
questions about galvanic effects if the probe is not isolated). This has important
implications for localised corrosion. First, initiation of pits and other localised
corrosion phenomena is typically a relatively rare event, and it will therefore be
less likely in any given period on a small probe than on a large area of the plant,
meaning that the plant will suffer the localised corrosion before it is revealed by
the probe (assuming of course that a monitoring method capable of detecting the
initiation of localised corrosion is used). Additionally, stable pitting or crevice
corrosion depends on a relatively large cathode area to supply the current to the
active pit or crevice. The small probe may not have a sufcient area to do this, so
that stable pitting cannot be achieved, whereas it can for pits that are driven by
the much larger area of the plant.
Based on these limitations, we must think carefully about the value of probes for
corrosion monitoring. Probes should be regarded as a method of determining the
corrosivity of the environment, rather than measuring the actual corrosion of the
plant, thereby providing information for the management of inhibitor additions or a
warning of upset conditions that result in an increased corrosivity. Consequently,
there is an argument that corrosion probes should be designed to be slightly less
corrosion-resistant than the plant material, so that there is a reasonable expectation
that they will provide an early warning of problems, rather than responding only
Corrosion monitoring whats the point? 3
after corrosion problems have started on the plant. However, this leads to a delicate
balancing act; as an example, we might consider using Type 304 stainless steel as a
probe for the monitoring of a plant constructed out of Type 316. This then implies
that we are going to operate the plant in conditions where 304 will not normally
suffer corrosion. This should certainly avoid corrosion problems with the plant, but
it will not make full use of the capabilities of the 316. We therefore need a probe
material that is closer to 316 in composition, something like a lean 316, with chro-
mium and/or molybdenum concentrations just below the specication minimum.
However, this then leads to further questions, as the material would have to be a
specialist cast; it would probably be processed rather differently from a commercial
cast, and it would probably have a quite different concentration and size distribution
of MnS inclusions. As the latter are well known to be very important in the initiation
of pitting corrosion, this leads to uncertainty as to whether the probe is more or less
sensitive to corrosion than the plant, and by how much.
It seems unlikely that corrosion monitoring will be able to do without the use
of probes for the foreseeable future, but careful thought needs to be given to the
implications of the factors indicated above for the reliability of the data obtained.
1.4 Accuracy of electrochemical methods
Electrochemical methods, such as the linear polarisation resistance method, are
popular for corrosion monitoring, offering the advantage of an instantaneous, online
measure of corrosion rate. These methods mostly depend on the Stern-Geary
equation, the derivation of which makes a number of assumptions:
There is one anodic metal dissolution reaction and one cathodic reduction reac-
tion (oxygen reduction or hydrogen evolution)
Both anodic and cathodic reactions are far from equilibrium, so that the reverse
reaction can be ignored
Both reactions obey Tafels Law (log i E, hence i di/dE)
The solution resistance is negligible (this can be measured and corrected for using
electrochemical impedance spectroscopy if it presents a problem)
The measurement is made sufciently slowly that capacitive currents can be
neglected.
For many (probably most) real systems, these assumptions are not valid; metal dis-
solution is usually not under pure activation control and there will be surface lms of
oxide, inhibitor or other absorbed species. Indeed, if the anodic reaction is under
pure activation control, then it is almost inevitable that the alloy will be corroding
too fast to be practically useful (except in a few particularly borderline systems such
as steel in de-aerated water at ambient temperature). Similarly, the cathodic reaction
is often under diffusion or mixed control (e.g. oxygen reduction); while complete dif-
fusion control (when the oxygen reduction reaction is at the limiting current density)
can be approximated by an innite Tafel slope, mixed control (which is probably
more common in reality) does not provide the required behaviour, and will lead
to errors. Additionally, there may be other, competing reactions; thus copper will
usually be corroding in conditions where the reverse copper deposition reaction
occurs at a signicant rate. A related effect is found with nickel-base alloys in hydro-
genated high-temperature water (i.e. pressurised water reactor environments) where
the kinetics of the hydrogen/water redox reaction are faster than the kinetics of
4 Corrosion monitoring in nuclear systems
the corrosion reaction, so that electrochemical measurements do not provide any
information about the corrosion rate.
The harmonic analysis and electrochemical frequency modulation methods are
interesting, in that, according to the theory on which they are based, they permit the
independent determination of the B-value for the Stern-Geary equation. However, it
must be appreciated that the theory still depends on the same assumptions; both
methods will produce a B-value for any system, but this does not prove that the
B-value is valid. If one or other of the anodic and cathodic reactions does not obey
Tafels Law, the Stern-Geary equation is not applicable, so the calculated B-value is
meaningless. It may be that the resultant estimate of corrosion rate is not too far from
the true value (too far in corrosion monitoring is an interesting concept within a
factor of two or three appears to be doing quite well, and may well be acceptable,
especially if the response to change in rate is detected reliably).
1.5 Epilogue and answers to the questions
The answer to all of the questions in the Introduction is, of course, no. Simply by
monitoring corrosion, we do not change the corrosion rate, the cost of corrosion, the
hazards of corrosion or the environmental impact of corrosion. This does not mean
that corrosion monitoring is of no value, but it does mean that it is only of value when
it forms an input to an active corrosion management programme that has outputs that
will lead to some or all of the above benets. Corrosion monitoring that does nothing
more than ll ling cabinets or database storage space might as well not have been
measured (except that it may help to explain why that unexpected corrosion failure
occurred).