Chapter 4

Steel Bridges

455
 8th International Cable Supported Bridge Operators Conference, Edinburgh, June 2013

SEVERN BRIDGE CABLES - CORROSION MODELS, USE OF

INHIBITORS AND THEIR IMPACT ON THE CABLE ASSESSMENT
Jeffrey Fisher1), and Paul Lambert 2)
1
Technical Director, Mott MacDonald, UK
2
Technical Director, Mott MacDonald, UK

Mott MacDonald Ltd,

Mott MacDonald House,
8 10 Sydenham Road,
Croydon, CR0 2EE
United Kingdom

e-mail: jeffrey.fisher@mottmac.com, paul.lambert@mottmac.com , web page: www.mottmac.com

Abstract
In 2006 the UK Highways Agency instigated a series of investigations to determine the effect of
corrosion on the load carrying capacity of the main suspension cables of the Severn Bridge. To
prevent further corrosion, a dry air injection system was installed which was enhanced by the
introduction of corrosion inhibitor into the air stream. Acoustic emission monitoring of the cables was
also installed to allow the location and frequency of wire breaks to be monitored. A second series of
intrusive inspections was undertaken in 2010, enabling the deterioration model to be further refined
and a more accurate assessment of the capacity carried out. These assessments confirmed the
effectiveness of the combined dry air and corrosion inhibitor system at stabilising the condition of the
suspension cables.

Keywords:
Suspension cables, cable assessment, dry air injection, acoustic monitoring and corrosion inhibitors

1. Introduction

The Severn Bridge is a 988m span suspension bridge located in the United Kingdom carrying the M48
to South Wales. Built in 1966 it featured innovations such as inclined hangers and was one of the early
pioneers of streamlined box girder deck construction.
In 2006 the UKs Highways Agency, though the concessionaire Severn River Crossing plc,
commissioned a programme of internal intrusive examination of the main suspension cables of the
Severn Bridge with Faber-Maunsell (now Aecom)/Weildlinger acting as consultant for the inspection.
These investigations provided the data for use in the strength assessment of the cables undertaken by
Mott MacDonald who were the Governments Representative for the Severn Crossings concession up
to mid 2012.
The purpose of the paper will be to explain the corrosion models and show the effectiveness of the
combined dry air injection system and the corrosion inhibitors in stabilizing the corrosion of the main
cables.
2. Background and History

The first intrusive inspection of the main cables, carried out in 2006/7, found the cables to be corroded
and to have reduced structural strength. Following this investigation and assessment, an acoustic
monitoring system and dry air injection system were installed on both cables to control the
deterioration of the cable. Both of these systems were procured through the Severn River Crossing plc
with Aecom as their consultant.

142

456
 8th International Cable Supported Bridge Operators Conference, Edinburgh, June 2013

The purpose of the dry air injection system was to reduce the relative humidity within the cable to less
than 40%, it is accepted within the corrosion industry that the achievement of this figure will prevent
corrosion of metals. As a further measure to enhance the dry air injection system it was decided to
include a corrosion inhibitor.
As recommended in the NCHRP Report 534, a further intrusive inspection and corresponding load
assessment was undertaken in 2010, the results of which are given in the IABSE paper(1). The
assessment showed that the strength of the cables were no worse that the original assessment in 2006.
In addition to the acoustic monitoring system, sensors were installed to monitor vehicle weights and
the air condition within the suspension cables. Using this data Mott MacDonald, developed a overall
monitoring system for cables based on the principles contained in the Highways Agencys standard
BD79/06 The Management of Sub-Standard Structure(2). The data from the monitoring system
were used to supplement results of an assessment of the cable to allow annual certification of the
cable.
The dry air injection system was installed during 2008 and the drying phase had been completed by
September 2009, as demonstrated by the achievement of a relative humidity of 40%.
3. Deterioration Model

To assist in the management of the suspension cables an analytical tool was developed to model the
deterioration of the galvanised steel wires as a result of corrosion. While the present condition can be
quantified, an assessment is required of the mechanism and timescale for this condition to be
established. Details of the development of the model are described elsewhere (1). Using this method it
is also possible to predict the long term effects of various remedial measures.
To assist in the development of the model, it was necessary to measure a number of physical
parameters. In areas where corrosion has not initiated, the as built condition such as wire diameter and
thickness of galvanising can be established. It was also possible to retrieve samples of wires for
detailed inspection and physical testing to identify the factors that govern failure. In areas where
corrosion has taken place it is possible to measure section losses or depths of penetration that have
occurred to date. In addition it is possible to identify failed wires and obtain values for contributory
factors.
For the wires making up the main suspension cables, the following assumptions have been made with
respect to the onset of corrosion:
1) The wires arrive at site adequately protected from corrosion until the cables have been spun.
2) Once in place, the cables are protected by three layers of protection:
a) The zinc galvanising on the individual wires.
b) A layer of red lead oxide paste on the outside of the cable.
c) A protective wrap consisting of wire plus tape plus coating.
3) Initially, the cable is protected from significant corrosion by the cumulative action of the three
protective systems.
4) The first to break down in the outer coating, allowing moisture and more importantly moist air
to enter the bundle. As the cable cools at night, the moisture in the air condenses to liquid.
5) In time, through exposure to water and the atmosphere, the effectiveness of the red lead paste
breaks down allowing the zinc galvanising to start corroding.
6) As patches of the zinc layer become fully consumed, the underlying steel starts to corrode.
7) While the rate of corrosion will be initially fast, the generation of voluminous corrosion
products may eventually occlude the corrosion site, slowing down the rate of metal loss.
8) Under stress, the corrosion of the wires can become concentrated, eventually reducing the cross
section of the wire sufficiently for it to fail by tensile overload.

Each of these stages needs to be modelled individually, based on both published data and site
observations and combined to produce the overall predictive tool. A series of laboratory investigations
were carried out on samples of wire removed from the structure. The failure of the wires was found to
be caused by the formation of narrow V shaped corrosion pits reaching a critical depth. The data on
143

457
 8th International Cable Supported Bridge Operators Conference, Edinburgh, June 2013

the critical defect size was best characterised by a Weibull distribution which confirmed that failure
occurred when the defect reaches approximately one third the thickness of the wire. Based on this
observation, the most appropriate data for modelling section loss relates to depths of penetration with
respect to time and published data was employed for the corrosion loss of zinc and unalloyed steel in a
range of environments(3). The data obtained from site was found to correlate well with the section loss
predicted by the model, see Figure 1, and provided the necessary confidence to use the model to
predict the future performance of the wires and the influence of dehumidification and corrosion
inhibitors.

Probability Density Function of Depth of Penetration 25 years after coating

breakdown, actual and predicted
3000
Depth of Penetration (microns)

2500

2000
Zinc prediction Iron Prediction Actual Penetration Recorded
1500

1000

500

0
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Cumulative Frequency%

Figure 1: Comparison between actual and predicted loss

4. Use of Corrosion Inhibitors

In order to provide additional corrosion protection to the wires during the initial period of moisture
reduction when corrosion rates could increase as oxygen becomes more available, and to provide a
back-up in the event of the dehumidification system being out of action, for example, for maintenance,
a vapour phase corrosion inhibitor system (VpCI) was developed for introduction into the dry air
stream.
VpCIs have been in use for a number of years, initially for military and aerospace applications but
now commonly also for automotive, domestic and construction uses. They are based on volatile
organic compounds that form continuous highly-adherent mono-layers on the surface of metals which
are effective in controlling the processes that lead to corrosion. While the formulation is specific to the
manufacturer, they are generally based on organic compounds such as amino alcohols that have
proved to be effective and durable corrosion inhibitors in addition to being non-toxic and of low
environmental impact.
In industry they are generally used with a carrier system which may be a water or hydrocarbon carrier
solution or a fine starch or talc powder. Due to concerns regarding the potential blocking of the air
voids in the cables, powder-based delivery systems were immediately rejected as a method of
introducing the inhibitor. Water based systems were also considered unsuitable, particularly as
moisture reduction was the principal aim, while solvent based systems were found to be incompatible
with the cable wrap and RH probes that formed integral parts of the dehumidification system.
For these reasons, an approach based on the introduction of the pure inhibitor vapour by employing
the dehumidification air stream as the carrier has been developed. The inhibitor is introduced into the
air stream via permeable emitters with no solid or liquid material being employed. In this manner it

144

458
 8th International Cable Supported Bridge Operators Conference, Edinburgh, June 2013

has been possible to ensure a sufficient level of inhibitor vapour is present within the air voids to
protect exposed metal surfaces while avoiding the risks of blockage by solid or liquid material.

Figure 2: Laboratory test to confirm compatibility of inhibitor with dehumidification system

components.
Because the protection layer is only one or two molecules thick, they should have no influence on
clearances and only a minimal effect on other physical properties. To confirm this, tests were carried
out which found that use of the inhibitor caused a small increase in wire-to-wire friction at low contact
pressure and had not significant frictional effect at higher load. Further tests were carried out to
confirm the inhibitor would not adversely affect the other components in the system, including cable
wraps, sealants and probes. Initially the tests were carried out in the laboratory (see Figure 2), but
exposure tests on site are on-going to confirm no long-term influence on properties. The laboratory
tests confirmed that the inhibitor had no significant effect on the characteristic of the wrap and sealant
materials and did not interfere with the operation of the RH probes.
5. Site Testing of Inhibitors

In order to confirm the effectiveness of the inhibitor delivery system it was considered advantageous
to develop a simple indicator-based test that could be employed on site as part of the on-going
inspection of the main suspension cables. Such a system was developed by the inhibitor manufacturer
and made available for site trials. In the presence of the inhibitor, the colour of the indicator changed
from blue to red and it was possible to confirm the presence of the inhibitor on the surface of the
wires, even deep within the cable, by use of swabs containing the inhibitor solution, see Figure 3.

145

459
 8th International Cable Supported Bridge Operators Conference, Edinburgh, June 2013

Figure 3: Swab test for inhibitor during cable inspection.

Figure 4: Positive inhibitor test on the sensor end of a corrosion probe.

146

460
 8th International Cable Supported Bridge Operators Conference, Edinburgh, June 2013

In addition to the site tests which confirmed the presence of the inhibitor across the full section of the
cable, a failed corrosion probe that had been replaced was also retained for testing. Swab tests with the
indicator confirmed the presence of inhibitor on the sensor section of the probe (see Figure 4) that had
been exposed to exhaust air from the dehumification system, confirming that sufficient inhibitor was
being introduced into the system so as to treat the full length of the suspension cable.

During the design phase of the dry air injection system there were concerns raised as to the ability of
the system to reach all parts of the cable. Site testing for the corrosion inhibitor was found to be
present at all locations tested, not just the external surface of the cable but at all locations

6. Monitoring of Cable

It was recognised early in the assessment process that the future management of the bridge would be
required to adhere to the principles contained in the Highways Agencys standard BD79/06. A
monitoring regime and warning system was developed to allow the Highways Agency sufficient
confidence as to the reliability of the suspension bridge cables.
The plan for the monitoring system was to assess the cable on a monthly basis using the following
methods:-
Visual observations - the operator, Severn River Crossing, were instructed to record any
observations that may be relevant to the condition of the cable. This could include damage to
evidence of entrapment of water under the cable sleeving and broken wrapping or main cable
wire.
Traffic loading traffic is the dominate live load on the bridge and it represents 20% of the
maximum load carried by the cables. Weigh-in-motion sensors have been installed on all road
access to the bridge since 2005; this had enabled a Bridge Specific Assessment Live Loading,
in accordance with BD 50(5), to be undertaken at periodic intervals. To undertake a full BSAL
was not considered necessary and parameter were identified to allow the Government
Representative to look for trends in the traffic flow, i.e. % of HGV and maximum axle
loading.
Acoustic monitoring a monitoring system has been in place on the bridge since November
2006 (partial) and spring 2008 (full). Any acoustic event detected are reviewed by the
specialist system suppliers on the basis of characteristic wire break acoustic signatures to
provide a confirmed wire break; and
Dry air injection system at the air inlet and exhaust locations along the length of the cables,
sensors have been located to provide information on the condition of the voids within the
cable. These sensors included relative and absolute humidity, temperature, flow, pressure and
corrosion.
Hand in hand with the monitoring system was the need to identify a trigger mechanism that would
allow the client and operator to be able to close the crossing should a sequence of events reduce the
cable capacity to a level that was deemed unacceptable for the travelling public.
7. Results of Acoustic Emission Monitoring

The results of the acoustic emissions are plotted in Figure 5 and the main points are:-
emissions before installation of the dry air injection system represented 0.4% of the total
number of wires in the cables;
emissions during the cable drying out period reduced to about 15 breaks per year representing
0.1% per year; and
emissions after drying out were less than 10% of the emissions before installation of the dry
air injection system.
The large increase in wire breaks during 2010 can be attributed to the removal of the wrapping wire
and wedging operations that took place during the 2nd intrusive inspection.

147

461
 8th International Cable Supported Bridge Operators Conference, Edinburgh, June 2013

Figure 5: Results of Acoustic Monitoring of Cables.

8. Conclusion

The installation of a dry air injection system and the introduction of vapour phase corrosion inhibitor
appear to have stabilised the condition of the cable. This has been demonstrated with the reduction in
the relative humidity of the cable and low levels of acoustic emissions since the drying out of the
cable.
The site testing for the corrosion inhibitors during the 2nd intrusive inspection demonstrated the ability
and effectiveness of the dry air injection system to reach all parts of the cable.

References

[1] Fisher, J. & Lambert, P.; Severn Bridge Recent Assessment of Main Suspension Cables, Taller,
Longer, Lighter: IABSE IASS London Symposium Report, UK, 2011.
[2] Highways Agencys standard BD79/06 -The Management of Sub-Standard Structures
[3] Mansfeld, F.; Atmospheric Corrosion Rates, time-of-wetness, and relative humidity,
Werkstoffe und Korrosion, 30, pp 38-42, 1979.
[4]Technical Data Sheet, Cortec PCI Emitters, Lake Chemicals & Minerals Ltd, Birmingham, UK.
[5] Highways Agencys standard BD50/92 Technical Requirements for the Assessment and
Strengthening Programme for Highway Structures Stage 3 Long Span Bridges

148

462
Severn Bridge Cables
- Corrosion Models
- Use of Inhibitors
- Impact on Cable Assessment
Jeffrey Fisher
Paul Lambert

463
Background and History
Corrosion Modelling
Effect of RH on Corrosion Rate
Corrosion Inhibitors
Laboratory & Site Testing of Inhibitors
Monitoring of Cable
Results of Acoustic Emission Monitoring
Conclusions

464
Background & History

Key Facts
Main span - 987.55m
Sag/Span - 1/12
Diameter of Cable - 495mm
No. of wires in each cable
(4.98mm dia) - 8322
Opened September 1966

465
Cable Works - Timeline

2005 Mott MacDonald commence cable

investigation
2006 - 1st intrusive inspection of cable (Faber
Maunsell)
2006 Monitoring of traffic using WIM sensors
2006/2008 Installation of acoustic sensors
2008/2009 Dry air injection system installed
2010 2nd intrusive inspection of cable (Aecom)

466
Corrosion

467
 Corrosion Modelling

Probability Density Function of Depth of Penetration 25 years after coating

breakdown, actual and predicted
3000
Depth of Penetration (microns)

2500

2000
Zinc prediction Iron Prediction Actual Penetration Recorded
1500

1000

500

0
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Cumulative Frequency%

468
Effect of RH on Corrosion Rate

(Vernon, 1935)

469
Effect of RH on Corrosion Rate

Reducing relative humidity has long been

recognised as an effective method of controlling
corrosion
The method is effective as long as the system is
working but what happens if it is off for
servicing or fails?
Also, during the draw-down from saturated to
dry there may be an extended period of damp
when corrosion may accelerate
A method of introducing a secondary protection
system was therefore investigated, based on the
use of mobile, adherent organic corrosion
inhibitors

470
Corrosion Inhibitors

A corrosion inhibitor is essentially any material

that when introduced to an environment where
corrosion is anticipated or already occurring will
either result in a significant reduction in existing
corrosion or prevent low rates of corrosion from
increasing
For practical reasons, this generic definition should
be appended by the requirement for the inhibitor
to be effective at relatively low levels of application

471
Corrosion Inhibitors Three
Types

Anodic Inhibitor - typically oxidising agents such as

nitrites, impede the progress of the anodic
reaction
Cathodic Inhibitor - generally involves limiting the
cathodic reaction by removing oxygen
Mixed Inhibitor - usually volatile film formers that
deposit a thin continuous layer, one or two
molecules thick over the entire surface of the
metal

472
VPCI Testing of Wrap and
Sealants

473
VPCI Effect on Tensile Strength of
Wrap

474
VPCI Effect on Inter-Wire Friction

475
On Site Inhibitor Testing

476
On Site Inhibitor Testing

Inhibitor No Inhibitor

477
Corrosion Monitoring

In order to confirm the effectiveness of the

combined RH/inhibitor system, air from the outlet
ports is monitored for its ability to sustain
corrosion
The corrosion rate should remain within the range:
0.1 1 A/cm2 , which is approximately equivalent to 1
10 microns per year, or
0.04 to 0.4 mils per year
The probes work by measuring small changes in
resistance of a tubular steel specimen corrodes
they do not measure the corrosion rate of the
wires directly
Testing of removed probes confirms the presence
of inhibitor on the surface of the steel tube

478
Corrosion Monitoring Probe

479
Monitoring of Cable

Substandard element of the structure

Reviewed in accordance with BD 79
Monitoring procedure implemented that provides
stakeholders with a reasonable level of confidence

480
Monitoring Plan

Visual Observations
during cable inspections
Traffic Loading
WIM sensors allow a BSALL to be undertaken
Results of BSALL over last 7 years show little change in traffic
loading and mix
Acoustic Monitoring
Provides indication of wire breaks
Dry Air Injection Monitoring
Controls atmosphere within cable
RH less than 40% - no corrosion

481
Acoustic Monitoring (1)

482
Acoustic Monitoring (2)

Emissions before dry air 0.4% of total wires

Emissions during cable drying out period
about 15 breaks per year 0.1% per year
Emissions after drying out less than 10% of
emission before dry air

483
Conclusion

Stabilised condition of cable by installation

of dry air injection and use of vapour phase
inhibitor
Demonstrated by reduction of relative
humidity and low levels of acoustic emissions
Site testing for corrosion inhibitors
demonstrated effectiveness of dry air system

484
485
486
487
488
489
490
491
492
493
494