Sunteți pe pagina 1din 7

Proceedings of ICE

Civil Engineering 162 May 2009


Pages 1824 Paper 800049

doi: 10.1680/cien.2009.162.5.18
Keywords

corrosion; failures;
materials technology

Gareth John

PhD, CSci, CChem, MRSC,


FICorr

is lead consultant in corrosion


management and cathodic
protection at Intertek Capcis in
Manchester, UK
Tony Edwards
BSc, AMIM3

is principal engineer at Intertek


Capcis in Manchester, UK
Adrian Wright
PhD, MICorr, MAE

is business manager consultancy at


Intertek Capcis in Manchester, UK

Mike Broadhurst
BSc, CEng, MIM3, MAE

is principal engineer team manager


consultancy at Intertek Capcis in
Manchester, UK

Learning lessons from


forensic investigations
of corrosion failures
Lessons and solutions learned from the forensic investigation of
corrosion failures need to be applied throughout the lifecycle of
structures. Avoiding misplaced solutions during both design and
operating phases relies upon a disciplined and professional forensic
approach for all corrosion failure investigations. This paper sets
out the elements of the approach and explains how lessons can
be learned to prevent future corrosion failures. In particular it
describes the macroscopic and microscopic methods required
for accurate identification of corrosion failure mechanisms and
successful root-cause analysis.

Chris Newton
DPhil, CEng, FIM3

is business manager
infrastructure at Intertek
Capcis in Oxfordshire, UK

When a component, system or structure fails


this may occur for many reasons, including
changes in materials triggered by corrosion.
Design engineers that consider only the
required strength and fabrication properties of
a construction material overlook the insidious
nature of the impact that service environment
can have on load-bearing or pressure-retaining
capacity over time.
Degrading materials lead to corrosion failures
that, in common with many types of engineering
failure, do not necessarily involve catastrophic
collapse or rupture but usually begin with an
unacceptable difference between expected and
observed performance of a component in service. Eventually this is usually recognised by
n parts becoming inoperable for example,
moving components seizing
n components become incapable of performing their intended function safely or
reliably such as loss of section leading to
reduction in load capacity and hence loss
of factor of safety.
Deficient performance can arise for many
reasons including non-conformity with design
specifications, or simply through sustained
use in the service environment which itself
may have changed over time. The reaction of
materials of construction to their service envi-

C I V I L

ENG I NEER I NG

ronment can lead a well-designed component,


or part of a system, to deteriorate slowly or
suddenly, perhaps degrading to a point that
a structure or system becomes unreliable or
unsafe for continued use.
Full lifecycle engineering of structures, making their performance more predictable and
eliminating unexpected failures, is becoming a
crucial part of asset and integrity management.
Application of lessons learned from corrosion
failures to new and existing structures demands
specialist materials-engineering skills throughout the lifecycle of any structure or component.
At each stage, the skills and technology
needed may change, from materials selection
in design to planning, fabrication, assembly
and construction, through to reliable operational integrity with planned maintenance and
effective repair, ending in demolition with
responsible reuse.
An essential step in understanding how to
eliminate corrosion failures is to run professional and accurate failure investigations
that yield credible analyses of their corrosion
mechanisms and the root causes of problems.

What is a corrosion failure?


The laws of thermodynamics mean that
all metallic materials degrade with time.
However, the form of degradation can vary

Learning lessons from forensic


investigations of corrosion failures

(a)

(b)

Figure 1. Examples of corrosion damage to reinforced concrete structures cracking of cover concrete (a) can eventually lead to spalling and structural failure (b)

widely depending on the material, the environment and the type of component. In fact
materials of all kinds concrete, ceramics,
polymers and metals can degrade due to
interaction with their environment.
In some situations, visible evidence of degradation is clear. Steelwork, for example, may
corrode, leaving voluminous evidence of red
rust, and eventually the steel may become too
thin to withstand the loads applied. However,
in many other cases, only a negligible amount
of visible corrosion can lead to catastrophic failure, such as stress-corrosion cracking (SCC).
Similarly, corrosion of reinforcing steel in
concrete forms a voluminous corrosion product that initially causes cracking of cover concrete (Figure 1(a)), which eventually can lead
to concrete spalling, and eventual structural
failure (Figure 1(b)). However, corrosion of
prestressing steel in concrete can, and has,
led to catastrophic failure by SCC without
any visible deterioration prior to failure as
occurred with the Berlin Congress Hall1 in
1980 (Figure 2).
The susceptibility of materials to corrosion
will depend on the corrosivity of their service
environment. This can be natural, such as the
atmosphere, soils, fresh water or seawater, or
industrial, such as in oil and gas production,
petrochemical, chemical and power generation. Many corrosion-failure mechanisms are
well-documented. Unfortunately, despite a
wealth of published knowledge, surprise catastrophic corrosion failures of components still
occur in engineering structures, pipelines and
process plants of all kinds.
Operators and owners of structures may
become complacent because the combination
of slow rates of degradation and what they
perceive to be high safety factors can induce a
false sense of assurance. Even when structures
are physically observed to be in a poor state of
repair, such as cracked or leaking, genuine surprise is still expressed when a sudden failure

issn 0965 089 X

or collapse occurs which may impact economically and tragically result in loss of life.15
Many lessons can be learned from corrosion failures. Sadly, a major cause of corrosion
failures is a misunderstanding or complete
ignorance by designers and engineers of the
changes in performance of materials as they
interact with their service environment. For
example, many still do not realise that stainless
steels are not corrosion-immune steels. Nor
is it generally appreciated that the numerous
different categories of stainless steels available such as austenitic, martensitic, ferritic
and duplex mean each have markedly different properties that give significantly altered
performance and service life when used in the
same application.

Many consider the concept of environmentally assisted cracking too esoteric, forming
part of the unforeseen or unforeseeable subclause in their insurance policy. This is compounded when designers and engineers do not
seek specialist advice during the design phase
of a project.
By designing out potential material problems, many of the catastrophic structural corrosion failures that have occurred in the recent
past could have been prevented. Therefore it
is important that the lessons from corrosion
failures are understood and that this information is made available to allow intervention,
and used both to predict and prevent similar
failures elsewhere.
In many instances communication of valu-

Figure 2. The Berlin Congress Hall roof collapse in 1980 was caused by stress-corrosion cracking of
pre-stressed, post-tensioned, high-strength-steel tendons1

ProCeedings of the Institution of Civil Engineers CIVIL ENGINEERING, 2009, 162, No. CE5

19

John, Edwards, Wright,


Broadhurst and Newton

Interview relevant
personnel and identify
data sources

Attend site as soon


as possible after failure

Stages of a corrosion failure investigation

Site visit

Conclusions

In common with other forensic activities,


corrosion failure investigations involve a
number of complementary stages to achieve
an accurate diagnosis of the problem. This is
summarised in Figure 3. An important step at
the outset is to recognise quickly the necessary
knowledge and experiences required, and then
form a team with the appropriate specialists.
Corrosion failure investigations can usually
be divided into two elements establishing
the failure mechanism and then the cause. The
first focus of the investigating team is to establish the mechanism of failure, which involves
defining exactly how the component physically came to fail. There are many potential
failure mechanisms that can involve corrosion,
for example, metal loss leading to overload,
fatigue and environmentally assisted cracking.
The second focus is upon defining the most
likely cause as to why the component failed
by the mechanism. To do this accurately it is
essential to understanding the exact operating
conditions that were imposed on the component which led to failure. It is vital to establish
both cause and mechanism if repetition of the
failure is to be avoided.

failure mechanism
causation

Mechanism of failure

in-situ examination
sample selection
collect information

Record and preserve


critical samples

Create a wish-list of
relevant information

Form a skilled team

Modelling

stress analysis
flow erosion
corrosion
thermal

Laboratory
investigation

Research

non-destructive
testing
mechanical testing
micro examination
chemical analysis
macro examination
characterisation
of failure

literature search
review of design data
on operating records
interrogation of
plant history
(process control)
witness statements

Simulated
testing
corrosion
mechanical
flow

Figure 3. Stages of a forensic investigation of corrosion failure the key to success is to get forensic
engineers on site as quickly as possible

Figure 4. Corrosion of this oil-transfer pipeline resulted in a pressure rupture

20

able knowledge gained during a major failure


investigation is not widely circulated, often for
legal reasons, even within the same company,
which sometimes leads to a similar failure
occurring years later.

In many instances, determining the mode or


mechanism of failure is easier when those first
on the scene gather key information and samples in a way that does not confuse or compromise the subsequent investigation.
Determination of the mechanism of failure
generally involves a detailed examination of
the failed component on site and/or in the
laboratory. While there are no rules as to
the exact sequence of steps taken during an
examination, normally the first stage involves
a detailed visual inspection. Key information
can be gathered from physical macroscopic
observation and accurate recording of the evidence. Some mechanistic features may become
self evident from the physical observation,
others may demand more detailed microscopic
investigation to become apparent.
As a general example, corrosion failures
in pipework show that key evidence of the
corrosion failure mechanism could range in
scale, from general uniform corrosion that
led to pressure rupture of an oil-transfer line
(Figure4), to relatively small areas of localised pitting corrosion leading to perforation
(Figure5). Other features may need knowl-

ProCeedings of the Institution of Civil Engineers CIVIL ENGINEERING, 2009, 162, No. CE5 issn 0965 089 X

Learning lessons from forensic


investigations of corrosion failures

edge of the combination of materials involved


like recognising galvanic corrosion caused by
the interaction of different metals or between
a weld and the remainder of the pipe wall
(Figure 6), or crevice corrosion initiated under
insulation, a seal or gasket.
Some mechanisms may be more ambiguous
and difficult to recognise or more hidden.
These can include environmentally-assistedcracking mechanisms such as SCC, which in
steels can be induced by exposure to certain
combinations of load and service environment.
Figure 7 which shows intergranular SCC of a
pipeline steel caused by exposure to dissolved
carbonate and bicarbonate in a high-pH
environment.

Corrosion-failure investigations evolve as they


proceed with a path dictated by the results of
each preceding stage. Investigations to determine
the mechanism of failure do not lend themselves to a tick-box approach. However, they do
include a process of elimination, starting with a
deterministic investigation for example, checking that the component conformed to the design
or legislative code requirements. Many corrosion
failures occur because it was not the correct or
specified material of construction or it was fabricated to the wrong dimensions.
All too often in major catastrophic failures,
those who specialise in this type of investigative work are brought in as an afterthought. It
is essential that if the problem is to be correctly

diagnosed these individuals are brought in as


soon as possible after the failure has occurred to
secure, and more importantly preserve, evidence.
This means getting them to the component and
site of the failure immediately, not several weeks
after the event or, as in some instances, not at
all. It is recognised that in many cases there are
severe operational constraints such as reopening a road or getting plant back online that
will restrict the time available for data collection,
hence the necessity to involve the investigation
team as soon as possible.
Help or hinder
There are many cases when a corrosion
failure investigation is adversely affected by

Cross-section

Surface

Figure 5. Localised corrosion pitting can lead to perforation

issn 0965 089 X

Figure 6. Cross-section of a weld suffering from


galvanic corrosion caused by the interaction of
different metals

ProCeedings of the Institution of Civil Engineers CIVIL ENGINEERING, 2009, 162, No. CE5

21

John, Edwards, Wright,


Broadhurst and Newton

100 microns

Figure 7.Iinter-granular stress-corrosion of a pipeline steel caused by exposure to dissolved carbonate and
bicarbonate in a high-pH environment

(a)

13 mm

(b)

(c)

20 mm

Figure 8. Examples of brittle (a), ductile (b) and fatigue (c) fractures at various macroscopic and microscopic
magnifications

22

poor handling practice or misplaced efforts


to help. It is important to ensure that the
fracture face and the rest of the component is
received as is, meaning with all particles of
rust, debris, dirt, oil and so on, still attached.
There are many instances of the failed part
being cleaned before the investigation begins.
This damaging action is likely to remove evidence of aggressive species that may provide
the information leading to the cause of the
corrosion problem. Poor storage, such as
wrapping failed parts under a tarpaulin outdoors, generally results in additional corrosion of the component that compromises or,
worse, destroys the available evidence.
Close physical or microscopic examination
of the fracture face can distinguish between
key features of the failure mechanism.
Therefore it is important that the fracture
face is not damaged and certainly never deliberately deburred of sharp edges to prevent
people being cut when manually handling the
failure. Sometimes a fractured component is
physically reassembled like pieces of a jigsaw
for transport or to understand where it came
from. In metals this destroys the fine fractographic detail normally observed at magnifications greater than 500 times that an expert
can rely upon positively to identify a fracture
mechanism from its characteristic tell-tale features unique to that mechanism.
Figure 8 shows characteristic examples of
brittle, ductile and fatigue fracture at various
macroscopic and microscopic magnifications.
Caution is necessary before cutting any part
to help make it easier to access because this
can mean non-expert judgement is applied to
remove what is considered important. Often
this attempt at practical assistance uses inappropriate methods, such as flame cutting,
which can degrade the original microstructure
of a metal part or destroy key evidence.
Progressive failure
Careful detailed examination of fracture faces
and microstructures can be essential in accurately diagnosing the mechanism of corrosion
failure. Even with clear macroscopic evidence, it
is possible to arrive at the wrong mechanism.
In a recent investigation, engineers assumed
the failure was a result of metal fatigue because
characteristic marks, sometimes known as beach
marks, were present on the fracture surface.
While beach marks are indicative of a progressive form of failure, showing points of crack
arrest across a fracture face, detailed examination by experienced metallurgists confirmed that
the actual failure mechanism was SCC. While
this is also a progressive failure, it has a different
cause from fatigue. Designing out a perceived
fatigue issue for the replacement part could not
have eliminated the problem and could have led
to a recurrence of the same corrosion failure.

ProCeedings of the Institution of Civil Engineers CIVIL ENGINEERING, 2009, 162, No. CE5 issn 0965 089 X

Learning lessons from forensic


investigations of corrosion failures

SCC can affect metals in specific environments, for example dissolved chloride ions
originating from exposure to marine service or
deicing salts. It is one of several environmentally-assisted crack mechanisms, which if not
identified, can lead to catastrophic failure of
structures and parts (Figure 9).
SCC can be a serious potential problem for
steel structures and other metal components
used in civil engineering because the fine cracking it produces can be difficult to find and is not
necessarily associated with more easily visually
recognised general corrosion or heavy rusting
that engineers are used to looking for during
inspection and maintenance. There is a lack of
awareness about SCC which is compounded by
the fact there is no single answer to all possible
situations that lead to it.
Instances of SCC failures are not confined to
steels. This type of corrosion failure mechanism
can affect some polymers and almost all the
alloys used in civil engineering but only in specific environments that can be different for each
alloy. While there is no single rule which can be
applied to cover all situations where SCC might
occur, in all cases the following combination of
conditions must exist together for it to occur
n susceptible material metal or alloy
n conducive environment chemistry, temperature and so on
n tensile stress sometimes residual stresses
caused by fabrication
n stress raiser affected by design and
damage.
Environmentally assisted cracking mechanisms like SCC can occur in a wide range of

situations, examples being


n collapses of the roof of the Berlin Congress
Hall in 19801 and the Ynys-y-Gwas Bridge
in Wales in 19852 were both due to hydrogen embrittlement of high-strength carbonsteel prestressing tendons
n failure of a dry dock floor in Scaramanga,
Greece, in 19773 was due to SCC of highstrength carbon-steel ground anchors
n SCC of stainless-steel road-bridge roller
bearings in the UK during 2002.4
As well as major failures, SCC can result
in failure of small components such as support road signs and street furniture. While not
dramatic, these corrosion failures still result in
increased maintenance and replacement costs
and increased risk to the public.
Detecting SCC often requires the use of nondestructive testing. The failure location can be
highly localised so it is necessary to know in
advance where to look and what techniques to
use. In some cases SCC can propagate quickly
while in other situations propagation can take
years to manifest itself.
Ideally judgement of the risk of environmentally assisted cracking, for example SCC,
requires a specialist to assess the materials
involved and to decide on the potential danger
areas during the design phase of a project or
new build.

Cause of failure root-cause analysis


Understanding the cause of failure is the most
critical step in learning the lessons from a corrosion failure. This step is essential, not just to

determine what went wrong, and hence who is


responsible, but also to establish how to prevent
it happening again.
The investigative process to determine the
cause of a failure is sometimes referred to as
a root-cause analysis because it considers all
the relevant issues from design and fabrication to construction, operation, inspection and
maintenance.
Understanding the environment
To provide accurate information for rootcause analysis it is always important that exact
details of the actual environment are determined. The components required vary depending on the exposure conditions, which may
include soil, natural waters, atmosphere and
process conditions.
Soils
Testing of soils to determine their corrosivity
with respect to metals requires a more detailed
assessment or site investigation than that commonly undertaken either as part of concrete
mix design, to meet the requirements of standards such as BRE SD-1,6 or in soil tests carried
out for pollution studies for brownfield sites.7
Corrosivity of soils is dependent on their
type, resistivity, chemistry (chloride, sulphate,
sulphide, oxygen, pH, buffer capacity, etc.) and
microbiology; there are detailed procedures
available8, 9 but these are rarely followed.
Natural waters
In natural waters, the presence of contaminants in particular acids, chlorides or bacteria
can generate corrosion-failure mechanisms.
Where contamination is expected, for

Figure 9. Inter-granular stress-corrosion cracking in a stainless-steel bridge roller bearing (a) failed bearing, (b) fracture face showing beach markings originating
from a rivet hole

issn 0965 089 X

ProCeedings of the Institution of Civil Engineers CIVIL ENGINEERING, 2009, 162, No. CE5

23

John, Edwards, Wright,


Broadhurst and Newton

example where microbiological influenced


corrosion is suspected, then samples are
required for analysis. This is a specialist form
of testing and needs to be done in accordance
with industry standards.10
Atmospheric
Atmospheric conditions can be assessed
using known climate data and information
from the operational use of the structure,
for example proximity to sources of pollution such as sulphur oxides, nitrogen oxides
and soot or other aggressive agents such as
cleaning systems.
Process systems
For most industrial systems, the process details should be clearly defined and can
typically be obtained from available process
monitoring data. However, it is often not the
steady-state operating conditions that prove
important in a corrosion failure root-cause analysis but outages, stoppages or excursions away
from the routine where conditions can change.
For example, periodic shut-ins of an oilsupply line can lead to water drop-out that does
not occur when oil flows regularly. Similarly
problems with dehydration facilities can lead to
wet gas entering pipelines, which are designed
and specified for dry-gas service only.
As part of a root-cause investigation, an
owner or operator of a structure may request
a determination of when corrosion started.
However well intentioned, this is usually an
impossible question to answer accurately due to
the localised nature of corrosion mechanisms.
The corrosion rates for any material vary
with the local environment, since it is dependent on temperature, pH, moisture, oxygen content and if the corrosion is localised or general
in nature. Unless it can be measured directly
by in-situ sensors or probes, corrosion rates for
the majority of systems can only be averaged
and estimated by reference to known data.
From these published data, a median corrosion
rate can be calculated from credible upper and
lower limits but these may still yield a range of
approximately a factor of two.
For example, if the amount of metal loss can
be estimated, say 2 mm, and the actual environment is understood to some degree, for example sea water of known temperate conditions
with suggested typical corrosion rates somewhere between 0.10.4 mm/year, then corrosion is likely to have been occurring for a time
between 520 years. Unfortunately, such wide
ranges are at best indicative. However, there are
cases where actual timelines identifying when
corrosion developed these are based on review
of periodic planned inspection data during the
life of the structure.
Two recent failure investigations recently
published by some of the current authors4, 5

24

illustrate how multiple material and environmental factors combine to generate the root
causes of unexpected failure in their operating
environment.
In the first investigation, a significant number
of high-strength martensitic-stainless-steel bearings in a UK-motorway viaduct failed.4 The
roller bearings failed by the development of vertical cracks in a six-year period since they had
been installed in the renovated Thelwall viaduct
over the M6. Root-cause analysis following
metallurgical investigations showed the interaction of atmospheric corrosion with stressing
conditions that contributed to the cause of failure. The failures were considered primarily due
to inadequate resistance of the material used to
corrosion and hydrogen-assisted SCC mechanisms. The material had too-high a hardness
level and did not comply with the specification
requirements.
In the second investigation, a north-African
gas pipeline exploded in December 2005 following a failure due to SCC.5 The line was
used to transfer sweet (low hydrogen sulphide)
natural gas from a field-processing unit through
to major centres of population and industry.
The failure had occurred in a salt marsh area.
It had been in service since 1989 and was fabricated from a carbon steel API 5L grade X52,
nominally 863 mm in diameter with a wall
thickness of 10 mm. The investigation showed
that failure was a result of high pH carbonate
and bicarbonate SCC resulting from ineffective
corrosion protection arising from an incomplete
tape coating and poorly specified impressedcurrent cathodic-corrosion-protection systems.
Generally, causation typically arises from one
of more of the following root causes
n wrong material selected
n material supplied not the material specified
n actual duty different to that accommodated
by the design, for example chemically more
aggressive than expected or applied stresses
higher than anticipated
n corrosion-control procedures adopted to
prevent or limit degradation not suitable or
not properly applied
n insufficient or inappropriate inspection
n inspection data identifying a problem not
properly reviewed and/or acted upon.

Conclusions
Many lessons are learned from corrosion
failures. These include using specialist advice in
designing a structure and throughout its operational life to avoid known issues.
When corrosion-failure investigations
become essential, they involve a number of
complimentary stages to achieve accurate
diagnosis of the mechanism and origin of the
problem.

It is crucial to approach failure investigations


using skilled professionals in a disciplined and
forensic way to maximise the accuracy of rootcause analysis to prevent future failures.

References

1. Isecke B. Failure analysis of the collapse of the


Berlin Congress Hall. Corrosion of Reinforcement
in Concrete Construction (Crane A.P. (ed.)). Ellis
Horwood, Chichester, 1983, 7990.
2. Woodward R. J. and Williams F. W. Collapse
of Ynys-y-Gwas bridge West Glamorgan.
Proceedings of the Institution of Civil Engineers,
1988, Part 1, 84, 635669.
3. Martin G. P. and Arnold A. C. Failure of dry
dock floor in Scaramanga, Greece, resulting
from stress corrosion damage to anchor pile
tendons. Proceedings of the Institution of Civil
Engineers, 1992, 96, 211225.
4. Edwards T., Schofield M. and Burdekin
M. Failure of roller bearings on the Thelwall
Viaduct. 4th International Conference on Forensic
Engineering: From Failure to Understanding (Neale
B. S. (ed.)). Thomas Telford, London, 2009,
423432.
5. Broadhurst M. and John D. G. Failure of a
34 inch gas main. 4th International Conference
on Forensic Engineering: From Failure to
Understanding (Neale B. S. (ed.)). Thomas
Telford, London, 2009, 159168.
6. Building Research Establishment. BRE Special
Digest 1. Concrete in Aggressive Ground. Building
Research Establishment, Watford, 2005.
7. Skinner H., Charles J. A. and Tedd P.
Brownfield SitesIntegrated Ground Engineering
Strategy. Building Research Establishment,
Watford, 2005.
8. Eyre D. and Lewis D.A. Soil Corrosivity
Assessment. Transport Road Research
Laboratory, Crowthorne, 1987, contractor
report 54.
9. Deutsches Institut fr Normung. DIN 50
929, Part 3. Corrosion of Metals, Probability of
Corrosion of Metallic Materials When Subject
to Corrosion from the Outside (Buried and
Underwater Pipelines and Structural Components).
DIN, Berlin,1985.
10. National Association of Corrosion
Engineers. Field Monitoring of Bacterial Growth
in Oilfield Systems. Houston, TX, 2004, NACE
TM0194 (note whilst this standard is aimed
at oilfield applications it is equally applicable to
other areas, including civil engineering).

What do you think?


If you would like to comment on this paper,
please email up to 200 words to the editor at
journals@ice.org.uk.
If you would like to write a paper of 2000 to 3500
words about your own experience in this or any
related area of civil engineering, the editor will be
happy to provide any help or advice you need.

ProCeedings of the Institution of Civil Engineers CIVIL ENGINEERING, 2009, 162, No. CE5 issn 0965 089 X

S-ar putea să vă placă și