Professional Documents
Culture Documents
composite materials
by
Doctor of Philosophy
Abstract
The aged offshore platforms will be forced to undergo repair and strengthening when either the structural
components have degraded critically due to the corrosive marine environment or the structure needs to carry extra
loads than they were originally designed for due to work over demands, heavier equipment or increased
environmental loads. The conventional methods to rehabilitate these structures are by welding or bolting steel
sleeves to the degraded member or replacing them with new members to account for the additional strength
requirements. Any hot work, including but not limited to welding process, above water to implement the
conventional repair in oil and gas platforms is risky and hence the normal operations will have to be interrupted to
carry out the repair thus incurring a huge monetary loss. Underwater welding is a complex job with less quality
and the welding process becomes very expensive. Bolted sleeves are sometimes used to overcome the complexity
of welding but these solutions also bring complications such as reliance on friction and loss of pretension over
time. Moreover, these kinds of repair add extra dead weight on the structure which is disadvantageous for the
structural performance.
Using Fibre Reinforced Polymers (FRPs) is established as a cost effective and efficient method for
rehabilitation and strengthening of onshore structures. However, the use of FRPs and composite material in
repairing or strengthening of offshore facilities is not fully developed and is limited to few basic applications yet.
Lack of extensive knowledge about the structural performance of FRP retrofitted steel members under complex
loading in the offshore environment is one of the main reasons why the technology is not practised widely.
This research project has tried to address this gap to a certain extent. Tubular steel structural members were
experimentally and numerically investigated in this study. Structural load tests of the intact and corroded tubular
members wrapped with FRP layers under pure axial compression, axial compression combined with bending load
and pure bending loads were investigated to understand how these retrofitted structural members behave in
different loading scenarios. Corrosion in these members was introduced by artificially machining out the wall
thickness in the mid-span. Repairing with Carbon Fibre Reinforced Polymer (CFRP) was carried out with the
objective of regaining the lost structural properties due to corrosion. Strengthening was performed using CFRP to
Repairing and strengthening were executed out using conventional in air retrofitting method as well as a
novel underwater retrofitting process. The comparison between underwater and in-air retrofits revealed that
i
The University of Western Australia Abstract
underwater retrofitting is performing almost as good as conventional in air retrofit. A parametric study on the
effect of the number of CFRP layers in repairing and strengthening was also conducted. Repairing of tubular steel
members was found successful in regaining the lost strength and stiffness due to corrosion. A simplified Finite
Element (FE) model was developed to predict the behaviour of steel tubular specimens repaired with CFRP under
concentric and eccentric compressive loading and a close match was obtained with the corresponding experimental
specimens. Strengthening the tubular members produced effective improvement in the structural performance only
when full length strengthening was utilised. The set of full length strengthening specimens also investigated the
difference between two underwater consolidation techniques for composite laminates namely, stricture and
vacuum consolidation. Underwater vacuum consolidation which is a very innovative method found to be the better
way to consolidate underwater FRP laminates on steel structures. An analytical check for design of retrofitted
offshore structural members was also formulated. The interaction behaviour of repaired tubular offshore structural
members was also studied. The durability of retrofitted members due to exposure in actual underwater or marine
conditions was also investigated through ‘real corrosion’ tests. The tests revealed that there is no loss in the
In conclusion, retrofitting of offshore structural members, above or below sea surface, using composite
FRPs is a potentially cost-effective and structurally safe alternative to conventional retrofitting solutions. However,
more detailed studies on long term performance and cyclic behaviour of such retrofitted members are essential at
ii
The University of Western Australia Acknowledgements
Acknowledgements
I would like to express my deep sense of gratitude to my supervising team: Dr. Mehrdad Kimiaei for his
constant encouragement, support, time for unplanned meetings and detailed technical reviews of my work
throughout this research, Dr. Mohamed Elchalakani for his guidance, sharing his experimental and technical
expertise and immense support during the publication process, Professor Mike Efthymiou for his keen insights and
mentorship during this research. I would also like to thank Dr. Sabrina Fawzia, from the Queensland University
of Technology for her review and guidance of the numerical components of the research.
I would like to express my sincere thanks to ICR.IAS Joint Venture for providing composite materials and
expert labour for the preparation of the experimental specimens. I would like to acknowledge the Structures
Laboratory team at the University of Western Australia, especially Jim Waters and Stephen Naulls for their
assistance and support during the experimental tests in the research. I would also like to acknowledge the efforts
of Master of Professional Engineering students, Anjali Gayathri Konara, Michael Mason, Liam McGurdy, Kit
Strickland, Vignesh Sairam Giridhar, Dong Hyun Kim and Bowei Sun who helped with the conduct of the
experiments.
I would like to acknowledge the financial support provided by Australian Government Research Training
Program (RTP) Scholarship during my study. I would like to thank Oceans Graduate School for all the academic
I would like to thank all my colleagues, friends and family for their support and encouragement extended
to me during the last four years. Last but not the least, I would like to thank my loving wife for understanding,
iii
The University of Western Australia Acknowledgements
iv
The University of Western Australia Thesis Declaration
Thesis Declaration
The work presented in this thesis is original work that has been performed by the candidate as a part of
attaining the Doctorate degree at the University of Western Australia. No part of this work has been submitted
elsewhere to any other university or tertiary institution. Appropriate acknowledgement and citations have been
made in text while referring to work that have been previously published elsewhere.
Necessary permission has been obtained from all the co-authors regarding the papers already published or
Signature:
April 2022.
v
The University of Western Australia Thesis Declaration
vi
The University of Western Australia Thesis format and authorship
The main chapters in the thesis (from Chapter 2 to Chapter 6) have been presented as a series of papers,
each of which have been published or submitted for potential publication as journal or conference papers. The
Paper 1: The first paper is compiled into Chapter 2 and is authored by the candidate, Dr. Mehrdad Kimiaei
and Dr. Mohamed Elchalakani. This paper has been published as a conference paper in the 5th International
Conference on Smart Monitoring, Assessment and Rehabilitation of Civil Structures (SMAR 2019), Potsdam,
Germany titled as “Simplified FE model predicting the bending behaviour of corroded tubular steel members
rehabilitated using CFRP”. The numerical model presented in the paper was developed and parametrically
investigated by the candidate. The candidate prepared the first draft of the paper and all authors reviewed and
Paper 2: The second paper is compiled into Chapter 3 and is authored by the candidate, Dr. Mehrdad
Kimiaei, Dr. Mohamed Elchalakani and Dr. Sabrina Fawzia. This paper has been published in Engineering
Structures, Elsevier Journal, titled as “Experimental and numerical investigation of underwater composite repair
with fibre reinforced polymers in corroded tubular offshore structural members under concentric and eccentric
axial loads”. The candidate performed the experimental testing under the guidance of supervisors and analysed the
experimental test results. The numerical simulations reported in this paper were also performed the candidate. The
candidate prepared the first draft of the paper and all authors reviewed and contributed to the final version.
Paper 3: The third paper is compiled into Chapter 4 and is authored by the candidate, Dr. Mehrdad Kimiaei,
Dr. Mohamed Elchalakani and Dr. Sabrina Fawzia. This paper has been published in Thin-walled Structures,
Elsevier Journal titled as “Underwater strengthening and repairing of tubular offshore structural members using
Carbon Fibre Reinforced Polymers with different consolidation methods”. The candidate performed the
experimental testing under the guidance of supervisors and analysed the experimental test results. The analytical
checks presented in this paper were also developed by the candidate. The candidate prepared the first draft of the
paper and all authors reviewed and contributed to the final version.
vii
The University of Western Australia Thesis format and authorship
Paper 4: The fourth paper is compiled into Chapter 5 and is authored by the candidate, Dr. Mehrdad
Kimiaei, Dr. Mohamed Elchalakani and Professor Mike Efthymiou. This paper has been submitted for publication
in Structures, Elsevier Journal titled as “Flexural response of underwater offshore structural members retrofitted
with CFRP wraps and their performance after exposure to real marine conditions”. The candidate performed the
experimental testing under the guidance of supervisors and analysed the experimental test results. The interaction
behaviour presented in this paper was also developed by the candidate. The candidate prepared the first draft of
the paper and all authors reviewed and contributed to the final version.
Paper 5: The fifth paper is compiled into Chapter 6 and is authored by the candidate, Dr. Mehrdad Kimiaei
and Dr. Mohamed Elchalakani. This paper has been published as a conference paper in the ASME 2021, 40th
International Conference on Ocean, Offshore and Arctic Engineering (OMAE2021 Virtual), Online titled as
“Experimental Study on Structural Rehabilitation of Severely Damaged I-Beams Using Fibre Reinforced
Polymers”. The candidate performed the experimental testing under the guidance of supervisors and analysed the
experimental test results. The candidate prepared the first draft of the paper and all authors reviewed and
viii
The University af Westem Australia Thesis format and authorship
r
Sign u1"t 2S.1O.ZO2!
Signature: Date:02.11.2021
ix
The University of Western Australia Thesis format and authorship
x
The University of Western Australia Table of Contents
Table of Contents
Abstract .................................................................................................................................................................... i
1 Introduction ....................................................................................................................................... 1
2 Simplified FE model predicting the bending behaviour of corroded tubular steel members
rehabilitated using CFRP ....................................................................................................................................... 17
xi
The University of Western Australia Table of Contents
3 Experimental and numerical investigation of underwater composite repair with Fibre Reinforced
Polymers in corroded tubular offshore structural members under concentric and eccentric axial loads ............... 27
3.2.1 General............................................................................................................................... 27
xii
The University of Western Australia Table of Contents
4 Underwater strengthening and repairing of tubular offshore structural members using Carbon Fibre
Reinforced Polymers with different consolidation methods.................................................................................. 65
4.4.1 Strengthened leg specimens in air vs. underwater under combined loading ...................... 74
5 Flexural response of underwater offshore structural members retrofitted with CFRP wraps and their
performance after exposure to real marine conditions......................................................................................... 109
xiii
The University of Western Australia Table of Contents
6 Experimental study on structural rehabilitation of severely damaged I-beams using Fibre Reinforced
Polymers ....................................................................................................................................................... 145
xiv
The University of Western Australia Table of Contents
7.1.4 Numerical and analytical models for retrofitted tubular steel members ........................... 163
xv
The University of Western Australia List of Tables
List of Tables
Table 3.6 Elastic properties used for the composite repair laminas in the FE model ........................................... 64
Table 4.2 Results of Strengthened leg specimens in air vs. underwater under combined loading....................... 103
Table 4.3 Strain values of Strengthened leg specimens in air vs. underwater under combined loading .............. 104
Table 4.4 Results of specimens with different number of retrofit layers ............................................................. 105
Table 4.5 Strain values of specimens with different number of retrofit layers .................................................... 106
xvi
The University of Western Australia List of Figures
List of Figures
Figure 1.1 Repaired Risers [27]............................................................................................................................... 8
Figure 1.3 The geometric model used for ABAQUS modelling [31].................................................................... 10
Figure 2.2 Comparison of the experimental specimen [left] with the FE model [right] ........................................ 20
Figure 2.3 Comparison of the FE model with the experimental results for typical bare steel specimens ............. 21
Figure 2.4 Comparison of the FE model with the experimental results of 1T1L series ........................................ 22
Figure 2.5 Investigation of the over prediction in the FE model of T1L2C2 1T1L .............................................. 24
Figure 2.6 The trend for the ultimate load with the change in the percentage of corrosion .................................. 24
Figure 3.2 Tri axial GFRP cloth (left) & Quad axial CFRP cloth (right) ............................................................. 32
Figure 3.3 Underwater repairing: (a) Repair tank filled with water from Swan River, (b) Wrapping of fibre layers,
(c) Stricture consolidation and (d) Curing ............................................................................................................ 34
Figure 3.4 Gridding (left) & gauging (right) of repaired specimens .................................................................... 35
Figure 3.5 Experimental setup: (a) Specimen setup in the Amsler just before testing, (b) Buckled specimen during
testing and (c) Digital data monitoring of the experiment .................................................................................... 36
Figure 3.6 Repaired specimens after testing: In air repair (left) & Underwater repair (right) .............................. 37
Figure 3.16 Different types of FRP failure in the repaired specimens ................................................................. 42
xvii
The University of Western Australia List of Figures
Figure 3.21 Geometry for the FE model of the tubular member with composite repair ....................................... 50
Figure 3.22 Validation of FE models for bare steel control specimens ................................................................ 52
Figure 3.23 Visual comparison of FE models with experimental bare steel specimens ....................................... 53
Figure 3.25 Visual comparison of the FE model (CAE0F2_FE) with the experimental repaired specimen
(CAE0F2A) .......................................................................................................................................................... 55
Figure 4.3 Steps in underwater vacuum consolidation: (a) Application of flow medium, (b) Bagging with
polyethene sheets, (c) Bagged specimen containing air and water inside, (d) Bag connected with a hose, (e) Hose
connected to the vacuum pump via a bypass tank and........................................................................................... 72
Figure 4.4 Testing set up: (a) Specimen fitted in the loading equipment before testing and (b) Specimen buckling
during the test ........................................................................................................................................................ 73
Figure 4.5 Method for calculation of Ductility Index (DI) [1] .............................................................................. 74
Figure 4.6 Strengthened leg specimens after testing: (a) in air & (b) underwater ................................................. 75
Figure 4.7 Unsymmetrical buckling in partial length strengthened leg specimens (IAE20F2W).......................... 75
Figure 4.8 Load displacement graphs of strengthened leg specimens under combined loading............................ 76
Figure 4.10 Load displacement behaviour of leg specimens strengthened underwater ......................................... 80
Figure 4.12 Load displacement behaviour of brace specimens strengthened underwater ..................................... 82
Figure 4.14 Load displacement behaviour of leg specimens repaired underwater ................................................ 84
Figure 4.17 Full length strengthened underwater (stricture) leg specimens after testing....................................... 87
Figure 4.18 Load displacement behaviour of full length strengthened underwater (stricture) leg specimens ....... 88
Figure 4.19 Strain values of full length strengthened underwater (stricture) leg specimens ................................. 89
Figure 4.20 Full length strengthened underwater (vacuum) leg specimens after testing ....................................... 90
xviii
The University of Western Australia List of Figures
Figure 4.21 Load displacement behaviour of full length strengthened underwater (vacuum) leg specimens ....... 91
Figure 4.22 Comparison of load displacement behaviour of leg specimens strengthened full length using stricture
and vacuum consolidation ..................................................................................................................................... 92
Figure 4.23 Strain values of full length strengthened underwater (vacuum) leg specimens ................................. 94
Figure 4.24 Typical cross section of a strengthened or repaired tubular specimen ............................................... 95
Figure 5.1 Real Corrosion specimens: (a) prepared with endcaps to be placed in the river and (b) lowered into river
hung with two ropes ........................................................................................................................................... 115
Figure 5.2 Retrieved Real Corrosion specimens: (a) covered with full marine growth, (b) manual removal of marine
growth, (c) water jet washing and (d) fully cleaned for structural testing .......................................................... 115
Figure 5.4 Method for calculating Ductility Index (DI) ..................................................................................... 117
Figure 5.6 Load displacement behaviour of repaired in air vs. underwater specimens ...................................... 119
Figure 5.7 Load displacement behaviour of repaired underwater specimens ..................................................... 120
Figure 5.10 Load displacement behaviour of in air and underwater strengthened specimens............................ 124
Figure 5.13 Real Corrosion - Corroded & Repaired specimens after testing ..................................................... 127
Figure 5.14 Load displacement behaviour of Real Corrosion - Corroded & Repaired specimens ..................... 128
Figure 5.15 Strain values of Real Corrosion - corroded & repaired specimens ................................................. 129
Figure 5.16 Real Corrosion - strengthened specimens after testing ................................................................... 130
Figure 5.17 Load displacement behaviour of Real Corrosion - strengthened specimens ................................... 131
Figure 5.20 Normalized N-M interaction behaviour of repaired specimens ...................................................... 134
xix
The University of Western Australia List of Figures
Figure 6.7 Local yielding of top flange of DIBC0 (damaged I-beam) ................................................................ 151
Figure 6.8 lateral torsional buckling of IIBC0 (intact I-beam) ............................................................................ 151
Figure 6.11 Strain observed on flanges (top and bottom) .................................................................................... 155
xx
The University of Western Australia Chapter 1
1 Introduction
Fibre Reinforced Polymer (FRP) composites are widely used to strengthen, retrofit and repair structures.
Composite is a general word used for a compound composed of at least two distinct materials with different
properties. These materials are often designed to work together to deliver better performance in terms of strength,
stiffness, formability and durability [1]. FRPs are called fibrous composites, as they contain fibrous reinforcements
embedded in a polymeric matrix. Carbon, glass and aramid are the typical fibres used for structural purposes, in
combination with thermosetting resins such as epoxy or vinylestre. FRPs similarly to other composites, offer
density. FRPs are well known for their high specific strength/modulus [1]. Moreover, they are very light weight,
durable and flexible for application on any surface. Of all the FRPs available, CFRP (Carbon Fibre Reinforced
Polymer) is the strongest and the most durable [2]. Moreover, CFRP can potentially possess a modulus of elasticity
comparable to that of steel which makes this one of the best suited materials for rehabilitating steel structures [3].
FRPs have been successfully used in Reinforced Concrete (RC) structures for their strengthening and repair
[4]. This technology is very well established and various design methods are developed to standardize their usage
in RC structures [5]. However, the usage of FRPs to retrofit, repair and strengthen steel structures is still being
studied by many researchers. The studies of FRPs on steel structures have been mainly focusing on onshore
structures which are not usually subjected to complex and extreme loading conditions as in offshore fields. There
is no proper analysis or design procedure developed so far to facilitate their application as structural components
in offshore platforms.
The current offshore applications for FRPs to repair or strengthen the structural members are very limited.
Most of the application of FRPs in marine conditions is for corrosion protection of non-structural elements in the
topside, leakage protection of piping systems and repair of pipelines or risers or conductors. The study on checking
the applicability of FRP for repair of subsea pipelines, where the critical load is the internal pressure of the line,
have shown positive results encouraging further research in the area [1, 3, 6]. However, for subsea pipelines, the
FRPs are mainly used to increase the strength of the degraded or leaking sections in hoop direction only to assist
in handling the internal pressure. Unlike the subsea pipelines, structural components in fixed or floating platforms
are under combinations of significant axial forces and bending moments. Hence, the potential application of FRPs
1
The University of Western Australia Chapter 1
for strengthening and repair of structural components in offshore facilities (fixed, floating or subsea ones) needed
further research.
Repair of the offshore structural members will become necessary due to deterioration of structural
characteristics due to corrosion from the marine environment, fatigue strength loss due to cyclic loads on the
structure or physical damages due vessel impact or dropped items and so on. Traditionally, steel pieces or sleeves
are usually either welded or bolted into the degraded or damaged member to regain the lost strength. If the repair
is above water, the repair works can interrupt the normal operation of the platform and if the repair is underwater,
special tools and skills will become necessary. In both cases, an economic loss is inevitable. The use of FRP for
Offshore or marine steel structures will require strengthening due to the work over demands or change in
usage during the lifetime of the structure. The strengthening may be required only for the critical members which
are usually identified through detail structural integrity analysis. Strengthening using FRP wraps around these
The use of FRP to rehabilitate reinforced concrete structures is a successfully practised technique. On the
contrary, the use of FRP to strengthen or repair steel structures is a comparatively less established technique.
Researchers all around the world have been working on this idea with different types of cross sections and sizes
with the focus on onshore structures. Research looking into steel sections retrofitted with FRP is gaining more and
more interest but and their explored applications in offshore conditions are mostly limited to pipelines and risers
with no complex loading scenarios. There is a huge potential for this technology for the rehabilitation of structural
members in offshore platforms. These types of members are subject to a wide range of combined loads where
usually both bending loads and axial forces are coming into play simultaneously. Most of the existing studies are
concentrating on either the behaviour of FRP retrofitted steel members under pure bending or pure axial loads. To
utilise this technique into steel structural systems, particularly offshore platforms, the effect of combined loading
DNVGL provides a recommended practice for ‘Design, fabrication, operation and qualification of bonded
repair of steel structures’ as DNVGL-RP-C301 [7]. This document gives some basis for designing and qualifying
bonded CFRP patch repairs for steel structures. DNV also provides an offshore standard, DNV-OS-C501 defining
minimum requirements for design, fabrication and installation of load carrying FRP laminates, sandwich
2
The University of Western Australia Chapter 1
structures and components [8]. However, these documents are only providing limited guidance on analysing the
performance and designing FRP composite retrofits for critical offshore structural members. A standardised
analysis and design framework for these types of modifications in existing structures should be formulated.
There are thousands of offshore platforms around the world which are approaching or have already
exceeded their original service life. These aged platforms are still profitable and are producing substantial levels
of hydrocarbons. Every year millions of dollars are being spent on structural rehabilitation and repair of existing
aged platforms using traditional strengthening methods. The project is trying to provide a better understanding of
using FRPs as a cost effective alternative method for structural retrofit of aged or damaged existing offshore
platforms. The results of this research would ultimately help with improving the integrity and extend the life of
existing offshore steel structures in a less risky and more efficient way than the conventional methods. The
monetary benefits foreseen from the success of this project are twofold. Firstly, the loss of money from long
disruption of the operations on a platform due to conventional repair can be averted. Secondly, the construction
and implementation costs can be significantly reduced with this technology generating huge amounts of profits.
The results of this research can be extended to the maintenance and repair of not only the fixed platforms but also
Much of the strength of FRPs is due to the type, amount and arrangement of the fibre reinforcement. E-
glass is the most used fibre reinforcement. It is strong and has good heat resistance. For applications that are more
critical, S-Glass offers higher heat resistance and about one-third higher tensile strength (at a higher cost) than of
E-glass. Carbon Fibres (graphite) are available in a broad range of properties and costs. These fibres combine
lightweight with very high strength and modulus of elasticity. For high stiffness applications, these reinforcements
perform extremely well, with a modulus of elasticity that can equal steel. FRP with carbon fibre reinforcement
also has excellent fatigue properties. Aramid fibres provide high strength and low density (40% lower than glass)
as well as high modulus. These fibres can be incorporated in many polymers and are extensively used in high
Both carbon and glass fibres absorb water but elevated temperatures are more critical for glass fibres. Glass
fibres do not perform well in alkaline conditions too [4]. Aramid fibres also absorb water and degrade in moisture
rich conditions. Long term performance of carbon fibre was found to be the most satisfactory among all the fibres.
3
The University of Western Australia Chapter 1
The strength modulus of carbon fibres was reduced by only 16% and they elongated by 12% after a period of 8
years of water immersion compared to fresh fibres [9]. Hence, we can conclude carbon fibre is the most appropriate
On the contrary, CFRP is known to cause galvanic corrosion in the steel members and it could become a
very serious concern [10]. In order to avoid this issue, a layer of GFRP (Glass Fibre Reinforced Polymer) is
wrapped as the bottom layer before wrapping the CFRP layers in order to electrically insulate the system [11]. A
hybrid composite system with E glass circumferentially arranged on carbon fibres is also available in the market
to avoid the issue of galvanic corrosion while using on steel structures [10].
The matrix or resin is the other major component of an FRP composite. Thermosetting polymers are usually
liquid or low melting point solids that can easily combine with fibres prior to curing. Thermosets feature cross-
linked polymer chains that become solid during a chemical reaction or cure with the application of a catalyst and
heat. The thermoset reaction is essentially irreversible. The thermoset resins for FRP are unsaturated polyesters,
The bond created by the resin must have good durability in extreme climates. The adhesive that has
demonstrated good durability under such environmental conditions is the epoxy resin and bonding is enhanced
when the surfaces are treated with silane [11]. It is ideal to have failure in the fibres rather than having delamination
or debonding failure in the FRP repaired system. The adhesive thickness should be also minimized in order to
avoid any residual stresses from developing in between the FRP layers [11]. Also, epoxy resin is suspected to
degrade under UV rays [10]. The risk of UV degradation can be mitigated using epoxy resins containing UV
resistant components or painting the FRP repair with UV protective coatings [12]. This factor will be of concern
if the repair is done on members in the splash zone of an offshore structure. A proper choice of the resin is equally
In order to obtain composite behaviour between steel and CFRP composites, it is essential to have a good
bond. Existing literature indicates that debonding and delamination are the main failure modes of the experiments
done with FRP and steel. Since the forces are transferred from the steel to the FRP through the adhesive, to take
advantage of the full capacity of the composite the adhesive has to be able to transmit the forces efficiently and it
is preferred that the fibres rupture before a failure of the adhesive occurs. Some investigators have suggested that,
when adhesives are not fully capable of transferring these forces, additional fasteners may be employed [11].
4
The University of Western Australia Chapter 1
Researchers have looked into the onshore applications of CFRP with different types of steel cross sections.
However, this review is concentrated on work done with Circular Hollow Sections (CHS) only as the structural
members used in the offshore industry are abundantly having this cross section. In the following sections the
experimental work done on CFRP on CHS steel members are presented. The experimental studies presented here
are mostly looking into simple loads like bending, axial tensile and axial compressive loads. Many parameters like
the number of CFRP layers, length of the CFRP wrap, the fibre orientation, etc. were usually studied by the
researchers.
Bending tests are a very common type of tests conducted by researchers to find the effectiveness of CFRP
retrofitting on CHS steel members. Kabir et al. (2015) studied the effect of CFRP on strengthened steel circular
hollow sections by subjecting them to four point bending test. The strengthened tubular members showed a
maximum increase of 43% in terms of the service load and 33% in terms of the ultimate load. The strengthened
and non-strengthened members showed similar stiffness for smaller loads, but at higher loads the strengthened
Elchalakani (2016 & 2017) did three point bending test on CHS to study the effect of corrosion penetration,
length of corrosion and type and number of CFRP wraps. The above study revealed an average increase in the
strength from the bare specimens to the CFRP repaired specimens to be 97%. The maximum strength gain was
obtained for the most severe (80%) corrosion. The strength and stiffness were only restored for 20% corrosion
specimen to that of the control specimen. All other levels of corrosion failed to be rehabilitated to their required
strength or stiffness due to inadequate number of CFRP layers [14, 15]. The necessity for a proper design procedure
for repair using CFRP can be understood from here. The inability of CFRP wraps to rehabilitate the original
strength of high corrosion percentages will not be a severe roadblock for our study. Unlike onshore structures, the
offshore structures are subjected to frequent inspections, which may lead to the identification of corrosion in its
members before reaching severe corrosion levels. Hence, the study revealing the full rehabilitation of members
In another study, Haedir et al. (2009) conducted experimental bending tests on 18 specimens that are
retrofitted with CFRP. The experiments revealed an increase in strength and stiffness of reinforced CHS
specimens. The composite effect of CFRP bonded CHS beams was more pronounced for thinner (more slender)
steel sections in pure bending [16]. The necessity for using proper dimensions to differentiate the offshore
applications from the onshore uses can be inferred from the above work. Moreover, a categorization of different
5
The University of Western Australia Chapter 1
structural elements in an offshore structure may become a necessity. A broad classification of offshore structural
members based on the dimensions, namely legs (less slender or compact elements) and braces (highly slender
elements) is to be implemented in the study to differentiate the difference in the behaviour of these types of
members.
In both above mentioned experimental studies, the wrapping of the CFRP was done in air which may not
be an accurate representation for the offshore applications. Seica and Parker (2007) conducted bending tests on
2.4 m long steel circular hollow sections strengthened with three layers of CFRP (2 longitudinal directions
followed by one hoop direction). The parameter studied was the in air curing and underwater curing of the
specimens. Two different varieties of fibres were utilized for the study. The flexural tests revealed better strength
and stiffness for strengthened specimens than bare steel members. Moreover, underwater curing showed a decrease
in this strengthening than in air curing [4]. The study reveals the adverse conditions in a marine environment will
be decreasing the performance of the CFRP repair system to a certain extent. However, these products were not
Axial loading is another typical experimental loading that is carried out to assess the performance of CFRP
retrofitting on steel CHS members. Gao et al. (2013) studied the buckling behaviour of steel tubular members
strengthened with CFRP sheets by employing axial compression test on them. To avoid galvanic corrosion a layer
of GFRP was used on the steel surface before applying CFRP in 2, 4, 6 and 8 layers. The strength improvement
was found to be increasing from 26% to 84% for the strengthened members and a linear relationship from the
results was developed between axial load and number of CFRP layers. The load deflection in the axial and lateral
directions indicate that the CFRP layers have increased the axial stiffness and stability against lateral deflection
[17]. Fawzia et al. (2006) studied the effect of high modulus CFRP strengthening on CHS members by testing
them in axial tension. The results revealed that the axial capacity has increased and the effective bond length was
found to be 75 mm for the used CFRP. The CFRP bond length, beyond which no significant increase in axial
capacity occurs, is known as effective bond length. In the same set of experiments, the strain was found to be
decreasing from the bottom layer to the top layer [18]. The research of CFRP on CHS steel sections is not limited
to these. The studies presented here have promising results to extend this research to offshore structural members.
The use of FRP composites for retrofitting steel non-tubular structural elements had been also investigated
by other researchers. Harries et al. (2009) experimentally demonstrated the feasibility of using FRP to provide
6
The University of Western Australia Chapter 1
cross-sectional stability through the bonding of FRP strips to flange elements and thereby improving the overall
capacity of the member [19]. Elchalakani and Fernando (2012) showed undamaged steel I-beams under 3-point
bending loads can be strengthened using CFRP by up to 32% when strengthened on both flanges [20]. Hosseini et
al. (2021) demonstrated prestressed CFRP composites could result in a fatigue crack arrest in steel I-beams even
under large external loads [21]. However, these studies are not looking into retrofitting I-beams or non-tubular
members which are heavily damaged or lost most of their cross section. Non-tubular structural members in offshore
facilities often present with heavy corrosion leading to loss of their web and flanges completely or partially.
Therefore, it is very useful to understand whether CFRP retrofitting can be utilized to retrofit such damaged non-
Pipelines are commonly retrofitted with FRP materials. Standard practices for repair of pipelines using FRP
wraps are provided by ASME in their code ASME PCC2 for ‘Repair of Pressure Equipment and Piping’ and ISO
in their code ISO 24817 for ‘Composite repairs for pipework - Qualification and design, installation, testing and
inspection’ [22, 23]. These code provisions cannot be used for the proposed application in the study as the external
loads acting on a pipeline is completely different from that of an offshore structural element. However, for the
subsea pipelines, the environmental conditions become very similar to that of offshore structures which brings the
The critical difference of repair of subsea pipelines to that of onshore pipelines lies in the fact that the
wrapping and curing of the CFRP will have to take place underwater. Recently, manufacturers are developing
CFRP and resins which are specifically tailored to perform underwater. If the proposed application of CFRP is to
be used for the rehabilitation of offshore structures, such special material will have to be employed.
Saeed (2015) studied the effect of GFRP and epoxy when it is cured underwater for a long period. The
study indicated a reduction in the properties of the material due to long term immersion in water. A pipe bursting
test on a specimen manufactured based on the code requirement and repaired with CFRP was also conducted. The
maximum depth of the defect was 80% of the original wall thickness, so the minimum remaining thickness was
1.28 mm. The specimen was gradually pressurized internally up to the test pressure of 32.6 MPa within about 100
seconds and sustained at that pressure over 60 seconds. Displaying no visible damage or drop in pressure, the
specimen was considered to have passed the test. On further increasing the pressure, the maximum load carrying
capacity of the repaired pipe was found to be 37.6 MPa and the failure was observed outside the repair zone [1].
7
The University of Western Australia Chapter 1
Green and Liew (2015) presented an experimental study which successfully demonstrated the effectiveness of
composite wraps for the repair of 70-80 % loss of wall thickness (internally) in subsea pipelines [24].
Alexander et al. (2005, 2007 & 2008) conducted experimental tests on repaired pipelines to find the
effectiveness of different composite repair systems developed by different manufacturers. The axial length of the
repair was fixed considering the variability of each product. Four products as named below were utilized for the
study.
The test program consisted of three tests per product and the details of the tests are as given below.
Pressure only test – sample destructively tested by increasing internal pressure to failure.
Pressure-tension test – sample destructively tested by increasing axial tension to failure while
holding internal pressure constant.
Pressure-tension test – sample destructively tested by increasing bending load to induce gross
plastic deformation while holding internal pressure and axial tension constant.
All the different tests revealed superior performance from product C which is a carbon fibre product [3, 25,
26]. The conclusion from this work justifies the decision of choosing CFRP as the repair material for the current
study.
However, the durability of CFRP in marine environments are not properly studied which might be an area
of concern for practical application. On the other hand, durability of GFRP in marine conditions is tested. Salt
spray test on GFRP overwrapped tubes were conducted by Leong et al. (2011) and the repaired pipe survived 4500
8
The University of Western Australia Chapter 1
hours of testing without corrosion under the GFRP overwrap [28]. A field demonstration was conducted by the
same overwrap system on a platform repairing corroded risers at splash zone (R1 & R2) and topside (R3) as shown
in Figure 1. The risers do not showed any degradation in visual inspection until 9 months from the repair work
[28]. Moreover, the galvanic corrosion of the steel caused by CFRP is also a concern. In order to mitigate all the
above drawbacks, a single layer of GFRP will have to be used before the CFRP overwraps. The additional layer
of GFRP will be bringing in some uncertainties in the structural performance of the member. The effect of this
additional GFRP layer will also have to be investigated to gain a more practical design procedure for rehabilitating
offshore structures.
In ABAQUS the FRP is commonly modelled using the inbuilt composite layup option [29]. However,
researchers have employed different techniques in ABAQUS to model the FRP behaviour more effectively. Mostly
the researchers try to reproduce their experimental results just to verify the results they have obtained.
Saeed (2015) did an FE analysis of the experiment conducted to verify the results using ABAQUS. Three-
dimensional reduced integration, eight-node linear solid element (C3D8R) was selected to produce the mesh. The
pipe, the filler and the FRP were modelled with the C3D8R element and the geometric nonlinearity was activated.
The FRP repair was modelled in three layers of elements through the FRP thickness. Tie constrains were used to
connect the interfaces in this model, hence creating a perfect bond between FRP and steel. The tested specimen
was capped with two solid steel plates at both ends, so both ends could expand or contract axially and rotate about
the longitudinal axis. Loading, including the internal pressure and axial load induced by the internal pressure, were
gradually applied to the model. The pressure was then gradually increased to the point that the analysis could not
proceed and was terminated due to the high level of deformation [1].
Shouman and Taheri (2009) modelled FRP repair in pipelines using ABAQUS. An 80% damage was
modelled and the steel was acting according to the Ramberg-Osgood expression. Combined loading due to internal
9
The University of Western Australia Chapter 1
pressure and bending were investigated in the FE study. FRP repaired model was compared with the undamaged
and unrepaired cases too. The wrinkling shapes in the pipe at different operating pressures were also studied [30].
A typical wrinkling pattern obtained from the above mentioned FE modelling is presented in Figure 2.
Figure 1.3 The geometric model used for ABAQUS modelling [31]
Kabir et al. (2015) conducted an FE study to find the effect on layer orientation and number of layers in a
strengthened pipe in bending using ABAQUS. Firstly, the FE models were validated by experimental results and
later the effects of parameters such as the number of layers and layer orientations were studied. Figure 3 shows
the geometry of the model utilized [32]. Alam et al. (2017) modelled the lateral impact of tubular members
strengthened with FRP materials. Even when the number of FRP layers were multiple, it had been designed as a
single composite plate. A thin adhesive layer was developed between the steel surface and cured FRP composite
plate. The cohesive elements were used to model the thin adhesive layer. The material failure behaviour was
defined by traction-separation law. Peak lateral displacements and average residual displacements between FE and
tests displacement responses were well matched [33]. All the Finite Element (FE) studies presented here show
good agreement with their corresponding experimental results and hence it is encouraging to pursue FE modelling
The research aims to study the feasibility of using FRP composite materials for the structural rehabilitation
of offshore steel structures. These structures may require structural repairing due to degradation in load bearing
capacities because of metal loss via corrosion in the marine environments, dents or damages due to accidents or
10
The University of Western Australia Chapter 1
extreme environmental loads. Strengthening of existing steel structural members required due to harsher
environmental conditions or increased functional loadings is another potential application. This study will try to
lay the groundwork for a framework for structural analysis and design of FRP retrofitted offshore structures.
The overarching aim of this research thesis is to qualify and quantify the performance of Fibre Reinforced
Polymer (FRP) retrofitted offshore steel tubular structural members under different loading conditions. This aim
3. To understand the effect of exposure of FRP retrofitted steel structure to actual marine conditions
4. To develop numerical models and analytical tools to help design FRP retrofitting for offshore
structural members
This thesis is presented in ‘thesis-by-papers’ format and comprises seven chapters altogether. The body of
the thesis contains 5 chapters which are papers published or submitted for publication. Each of these body chapters
separately contains an abstract and a relevant literature review at the beginning to introduce the context and
problem definition. Among them, chapters 3 to 5 are technical papers that have been either published or submitted
for publication in scientific journals and chapters 2 and 6 are published conference papers. The specific objectives
and scope of each of these chapters and how they contribute to the overarching aims of the thesis are described
below.
Chapter 1 or this chapter introduces the entire thesis with the research background, problem statement,
Chapter 2 presents the development of a simplified Finite Element (FE) model to simulate the behaviour
of corroded steel tubular members under bending and bearing loads. The chapter is discussing the FE modelling
technique in detail and how well the simulations matched with the existing experimental results by Elchalakani et
al. [34, 35]. This chapter is addressing 4th objective of the thesis and formed the basis for FE modelling presented
in the next chapter. This chapter is a conference paper presented and published in the 5th International Conference
on Smart Monitoring, Assessment and Rehabilitation of Civil Structures (SMAR 2019) in Potsdam, Germany
11
The University of Western Australia Chapter 1
titled ‘Simplified FE model predicting the bending behaviour of corroded tubular steel members rehabilitated
using CFRP’.
Chapter 3 presents a qualitative and quantitative assessment of underwater repair with FRP of corroded
tubular offshore structural members undergoing concentric and eccentric axial compressive loads through
experimental and numerical works. A comparison of the steps involved in the onshore or in air repair and the
subsea or underwater repair of corroded steel members is done. Experimental results of corroded steel tubular
members repaired with FRP under concentric and eccentric axial compressive loads focussing on their ultimate
strength, load displacement behaviour, energy absorption, ductility and strain values are presented in detail. The
development of a simplified FE model which can replicate the behaviour of such structural members with
composite repairs so that future repair designs can be made more effectively is also included in this chapter. This
chapter is addressing 1st, 2nd and 4th objectives of the thesis and it is published as a journal paper in ‘Engineering
Structures’ titled ‘Experimental and numerical investigation of underwater composite repair with Fibre
Reinforced Polymers in corroded tubular offshore structural members under concentric and eccentric axial
loads’.
Chapter 4 presents experimental results of strengthened and repaired tubular offshore structural members
undergoing concentric and eccentric compressive loads along with an analytical check to see whether an FRP
retrofit would reach a certain target level. Strengthening of tubular steel structures in air and underwater is
compared using structural testing of partial length strengthened specimens under concentric and eccentric
compressive loads. A parametric study on the effect of the number of retrofit layers and type of cross section is
presented for underwater repairing and strengthening of offshore tubular steel members. Underwater full length
strengthening of steel tubular structural members undergoing axial compressive loads is also showcased along
with the introduction of a novel underwater vacuum consolidation technique for FRP composites on steel
structures. Finally, an analytical methodology to check whether retrofitted tubular members under axial
compressive loads would reach their target section properties is described. This chapter is addressing 1st, 2nd and
4th objectives of the thesis and it is published as a journal paper in ‘Thin-Walled Structures’ titled ‘Underwater
strengthening and repairing of tubular offshore structural members using Carbon Fibre Reinforced Polymers
Chapter 5 presents experimental results of strengthened and repaired tubular offshore structural members
undergoing bending loads and demonstration of durability of FRP retrofitted steel tubular members after exposure
12
The University of Western Australia Chapter 1
to actual marine or underwater conditions. Comparison of in air and underwater retrofitting using FRP composites
is studied using four point bending test on retrofitted tubular members. The effect of increasing the number of
retrofit layers in repaired and strengthened specimens is also investigated. The effect of retrofitted specimens when
they are exposed to underwater or marine conditions through ‘real corrosion’ tests is also presented in this chapter.
Finally, the interaction behaviour of tubular members repaired with CFRP under a combination of axial
compressive and bending loads is also presented. This chapter is addressing 1st, 2nd, 3rd and 4th objectives of the
thesis and it is submitted for publication as a journal paper titled ‘Flexural response of underwater offshore
structural members retrofitted with CFRP wraps and their performance after exposure to real marine
conditions.
Chapter 6 demonstrates the rehabilitation of a severely damaged I-beam typically used in the offshore
industry as structural components using FRP composites. Experimental investigation of structural rehabilitation
of heavily damaged I-beams using Carbon and Glass Fibre Reinforced Polymers (CFRPs and GFRPs) is presented.
This chapter is a conference paper virtually presented and published in the 40th International Conference on Ocean,
Offshore and Arctic Engineering (OMAE2021) titled ‘Experimental study on Structural Rehabilitation of
Chapter 7 summarises the main conclusions of the research and identifies key areas for future research.
1.6 References
[1] Saeed N. Composite Overwrap Repair System for Pipelines - Onshore and Offshore Application Queensland
[3] Alexander C, Cercone L, Lockwood J. Development of a Carbon-Fiber Composite Repair System for Offshore
Risers. 27th International Conference on Offshore Mechanics and Artic Engineering. 2008.
[4] Seica MV, Packer JA. FRP materials for the rehabilitation of tubular steel structures. Composite Structures.
2007;80:440 - 50.
[5] ACI. Guide for the Design and Construction of Externally Bonded FRP Systems for Strengthening Concrete
13
The University of Western Australia Chapter 1
[6] Shouman A, Taheri F. Compressive strain limits of composite repaired pipelines under combined loading
[7] DNV-GL. Design, fabrication, operation and qualification of bonded repair of steel structures. DNVGL-RP-
C301 2015.
[9] Sleeper B, Arnold SF, Pridmore AB, Carr H. Carbon Fiber Reinforced Polymer (CFRP) As a Long Term
[10] Shamsuddoha M, Islam MM, Aravinthan T, Manalo A, Lau K-t. Effectiveness of using fibre-reinforced
[11] Setvati MR, Mustaffa Z, Shafiq N, Syed ZI. A Review on Composite Materials for Offshore Structures 33th
International Conference on Ocean, Offshore and Arctic Engineering. San Francisco, California, USA OMAE;
2014.
[12] Chennareddy R, Tuwair H, Kandil UF, ElGawady M, Reda Taha MM. UV-resistant GFRP composite using
[13] Kabir MH, Fawzia S, Chan THT. Effects of layer orientation of CFRP strengthened hollow steel members.
[14] Elchalakani M. Rehabilitation of corroded steel CHS under combined bending and bearing using CFRP.
[15] Elchalakani M, Karrech A, Basarir H, Hassanein MF, Fawzia S. CFRP strengthening and rehabilitation of
corroded steel pipelines under direct indentation. Thin-Walled Structures. 2017;119:510 - 21.
[16] Haedir J, Bambach MR, Zhao X-L, Grzebieta RH. Strength of circular hollow sections (CHS) tubular beams
externally reinforced by carbon FRP sheets in pure beding. Thin Walled Structures. 2009;47:1136 - 47.
[17] Gao XY, Balendra T, Koh CG. Buckling strength of slender circular tubular steel braces strengthened by
[18] Fawzia S, Al-Mahaidi R, Zhao XL, Rizkalla S. Strengthening of circular hollow steel tubular sections using
high modulus CFRP sheets. Journal of Construction and Building Materials. 2006;21:839 - 45.
14
The University of Western Australia Chapter 1
[19] Harries KA, Peck AJ, Abraham EJ. Enhancing stability of structural steel sections using FRP. Thin-Walled
Structures. 2009;47:1092-101.
[20] Elchalakani M, Fernando D. Plastic mechanism analysis of unstiffened steel I-section beams strengthened
[21] Hosseini SM, Melchior J, Izadi M, Ghafoori E. Fatigue crack arrest in steel beams using FRP composites.
[22] ASME. Repair of Pressure Equipment and Piping. ASME PCC2 USA2015.
[23] ISO. Petroleum, petrochemical and natural gas industries - Composite repairs for pipework - Qualification
[24] Green M, Liew SC. Testing and history of composite repair systems for subsea pipeline repair. Australaisn
[25] Alexander CR. Development of a composite repair system for reinforcing offshore risers Texas: Texas A&M
University 2007.
[26] Alexander C. Evaluation of the Aqua Wrap system in repairing mechanically- damaged pipes California,
[28] Leong AYL, Leong KH, Tan YC, Liew PFM, Wood CD, Tian W et al. Overwrap composite repairs of
offshore risers at topside and splash zone 18th International Conference on Composite Materials Jeju Island,
Korea2011.
[30] Shouman A, Taheri F. An investigation into piples repaired under combined bending and internal pressure.
In: OMAE, editor. 28th International Conference on Ocean, Offshore and Artic Engineering. Hawai, USA2009.
[31] Bambach MR, Elchalakani M, Zhao XL. Composite steel–CFRP SHS tubes under axial impact. Composite
Structures. 2009;87:282-92.
15
The University of Western Australia Chapter 1
[32] Kabir MH, Fawzia S, Chan THT, Batuwitage C. Effect of layer variations on CFRP strengthened steel circular
hollow members under bending: Numerical studies. PLSE 2015 - Second International Conference on
[33] Alam MI, Fawzia S, Tafsirojjaman, Zhao X-L. FE modeling of FRP strengthened CHS members subjected
[34] Elchalakani M. Rehabilitation of corroded steel CHS under combined bending and bearing using CFRP.
[35] Elchalakani M, Karrech A, Basarir H, Hassanein MF, Fawzia S. CFRP strengthening and rehabilitation of
16
The University of Western Australia Chapter 2
2.1 Abstract
Offshore structural members will be forced to undergo repair and strengthening when either the structural
components have degraded critically due to the corrosive marine environment or the structure needs to carry extra
loads than they were designed for due to work over demands, heavier equipment or increased environmental loads.
Using FRP composites is established as a cost-effective and efficient method for rehabilitation and non-structural
repair of these type of structures. However, the use of FRP composite material in strengthening or structural repair
of offshore facilities is still limited to a few basic applications due to the lack of a proper design framework.
Tubular steel members are widely used in offshore platforms and in this paper, existing experimental test data for
the ultimate strength of tubular sections retrofitted with CFRP under bending was used to develop an FE model.
The accuracy of the model was examined against the experimental data and a reasonable match was found. The
simplified FE modelling technique presented in the paper is expected to aid in the development of a design
2.2 Introduction
Circular Hollow Sections (CHS) are widely used as structural members in the offshore industry. The highly
corrosive marine environmental conditions in which these members are utilized makes it necessary to rehabilitate
them. Due to the high specific strength, Carbon Fibre Reinforced Polymer (CFRP) composites are accepted as a
suitable repair material for steel CHS [1]. However, the rehabilitation of steel structural members with CFRP
composites are practiced limitedly in the offshore industry due to the lack of a design framework for this kind of
retrofitting [2]. The FE modelling technique presented in the paper will be a first step in addressing this issue for
Elchalakani (2016) & Elchalakani et al. (2017) presented experiments looking into the rehabilitation of
corroded Circular Hollow Sections (CHS) using CFRP [3, 4]. The corrosion was artificially induced in the steel
tubular members by machining them. The corrosion thickness was varied in order to understand the effect of the
level of corrosion and the length of the corrosion was also parametrically investigated. The Structural Technology
(ST) V-wrap, a unidirectional CFRP was utilized to retrofit the corroded specimens. The retrofitting showed a
significant increase in the ultimate strength of the specimens under three-point bending. The experimental
17
The University of Western Australia Chapter 2
specimens which used only two layers of CFRP were selected from Elchalakani (2016) to develop the FE model
The paper aims to present the development of a simplified Finite Element (FE) modelling technique to
predict the three-point bending in steel CHS retrofitted with CFRP. The potential of the modelling technique in
predicting the ultimate capacities FRP retrofitted steel tubular members is showcased in the paper.
2.3 Methodology
The modelling of bare steel specimens under three-point bending was the first step carried out in the FE
modelling. The experimental set up as shown in Figure 2.1 was replicated in Abaqus software [5] by defining the
saddle supports and loading rod as rigid bodies in contact with the steel CHS. The saddle supports were fixed and
the loading rod had a degree of freedom in translation to facilitate the loading motion. The bare steel CHS was
modelled as a deformable shell element in Abaqus. Elchalakani (2016) used mild steel CHS of grade B335JR
confirming to ASTM A53 Schedule 30 in the experimental study [3]. An elastic-plastic material model was
employed to define the steel for the bare specimens in the FE model. The average yield stress (482.63 MPa) found
out from the tensile coupon tests conducted by Elchalakani (2016) was used in the material definition [3].
The experimental specimens were cut from CHS of outside diameter, Do = 101.6 mm and thickness, to = 5
mm. The clear span of specimens between the saddle supports was 900 mm. The corrosion was artificially induced
by machining 360o around the pipe for various depth. T5 is the control specimen representing a 0% corrosion in
the wall (i.e., 5 mm thickness). The same logic was applied to T4, T3, T2 and T1 representing a 20%, 40%, 60%
and 80% corrosion in the wall respectively (i.e., 4 mm, 3 mm, 2 mm and 1mm thickness). L1 and L2 represented
the length of corrosion (Lc) equal to 100 mm and 200 mm respectively. C0 or C2 was indicative of 0 and 2 CFRP
18
The University of Western Australia Chapter 2
layers respectively used to retrofit the specimen. 1T1L was conveying the orientation of the CFRP layers, one
layer in the transverse (T) direction and one layer in the longitudinal (L) direction [3].
The second step in FE modelling was the development of the retrofitted model. The same setup utilized
earlier was used for this as well. The corroded region was filled with an elastic filler material of negligible structural
stiffness and the retrofitted zone was wrapped with CFRP layers similar to the experimental specimens. In the
experimental study by Elchalakani (2016), parameters like length of corrosion (Lc), length of CFRP wrapping
(LCFRP) and corrosion percentage were investigated. The effects of these parameters were studied using the FE
The CFRP properties were defined using an elastic-plastic orthotropic lamina with a stronger elastic
modulus in the predominant direction. The elastic moduli in a lamina were defined by strong and weak directions
which simulated the properties along and across (longitudinal and transverse for the pipe) the fibre orientation.
The elastic properties along the thickness were considered negligible by default in a lamina definition. The
Poisson’s ratio and the shear moduli in all 3 planes were required to complete the elastic lamina definition. The
Hashin damage model was included to define the failure of the composite layer [6]. A Hashin coefficient (α) equal
to 1 was employed. The damage initiation was modelled by defining tensile, compressive and shear strengths in
the longitudinal and transverse directions. Damage evolution was defined with tensile and compressive fracture
energies in both directions. Damage stabilization was defined with the corresponding viscose regularization for
each fracture energy. The material data provided by the CFRP manufacturer was not sufficient to define all the
above properties. Hence, the microscopic material definition for CFRP developed by Rentsch et al. (2011) was
utilized as the basis for the CFRP material definition in Abaqus for fibre strengths in the transverse directions,
fracture energies and viscose regularization [7]. The CFRP layers were defined with the properties as presented in
Table 2.1 to have the best representation of the fibres used in the experiments by Elchalakani (2016) [3].
The FE model of the CFRP retrofitting was developed using the composite layup technique in Abaqus. The
composite layup is an inbuilt feature in Abaqus where multiple layers of different materials can be modelled with
ease [5]. Instead of taking the thickness of the entire composite which is the sum of resin and fibre thickness, the
thickness of the fibre (0.23 mm) was only used in the model. This was done as only the fibre will be actively
participating in the load sharing with steel. In the layup shell element, the transverse (T) and the longitudinal (L)
layers of the CFRP are distinguished based on their orientation angles. The disadvantage of the model is that
debonding and delamination of the fibres due to adhesive failures will not be simulated. In an ideal case scenario,
19
The University of Western Australia Chapter 2
debonding and delamination failures should be avoided until the retrofitted fitted specimen reaches the ultimate
strength. If this condition is achieved in the experimental specimens, the FE models would be capable of capturing
the bending behaviour of the CFRP retrofitted steel specimens. This capability was demonstrated through the
visual comparison between the experiment and the FE model of a CFRP retrofitted steel CHS under bending
Figure 2.2 Comparison of the experimental specimen [left] with the FE model [right]
The comparison between the FE models with the experimental results for typical bare steel specimens is
presented in Figure 2.3. It is seen that the bare steel models are showing a very good agreement with the
experimental results. A high level of agreement in the ultimate strength, as well as the load-displacement behaviour
of bare steel specimens, was observed. However, the agreement decreased as the level of corrosion increased. The
reason behind this was the fact that the stress concentration at the sudden drop in thickness cannot be properly
20
The University of Western Australia Chapter 2
simulated in the FE model. Due to this, the FE model was predicting a slightly lower capacity than what was
Figure 2.3 Comparison of the FE model with the experimental results for typical bare steel specimens
The FE models of CFRP retrofitted specimens showed a good agreement with the experimental results as
presented in Figure 2.4. In the elastic region, the FE model was slightly higher than the experimental load-
displacement behaviour. The ultimate load and post-failure behaviour were matching very well for most of the
cases. The ratio of the ultimate load from the FE model to the experimental value was calculated and presented in
Table 2.2. The ratio from all the specimens in the 1T1L series showed an average of 1.04 and a standard deviation
of 0.11 indicating the good reliability of the FE model in predicting the bending behaviour of CFRP retrofitted
CHS.
21
The University of Western Australia Chapter 2
Figure 2.4 Comparison of the FE model with the experimental results of 1T1L series
22
The University of Western Australia Chapter 2
The FE model for T4L1C2 1T1L was under predicting the experimental results by 9% which was the worst-
case scenario of the FE model failing to reach the ultimate capacity as in the experiment. Even in this case, the
load-displacement behaviour was reasonably matching well until the yielding started.
The comparison of the FE model and the experimental results for the specimen T1L2C2 1T1L showed
considerable disagreement. This was the only case where the FE model significantly over predicted the
experimental results. The FE model predicted the ultimate load to be 29% higher than the observed experimental
ultimate load. The occurrence of premature debonding in the CFRP layers could be a probable explanation for the
low ultimate load in experiments. In order to investigate this issue, the exterior layer of the CFRP retrofitting
system was removed and analysis was carried out to examine its effect. The comparison of the FE models with
that of the experimental results is presented in Figure 2.5. The FE model with only the transverse (T) layer of
CFRP showed a closer match with the experimental results confirming the probable cause of the low values from
this experiment. The difference in the ultimate load had dropped from 29% to 6% due to this assumption made in
the FE model. Hence it would be reasonable to presume that the exterior longitudinal (L) layer of CFRP in the
experiment debonded prematurely and did not contribute to the load carrying abilities of the T1L2C2 1T1L.
Elchalakani (2016) had reported debonding for this specimen [3], but whether this debonding was initiated
23
The University of Western Australia Chapter 2
Figure 2.5 Investigation of the over prediction in the FE model of T1L2C2 1T1L
Moreover, when the ultimate strength was studied with the corrosion percentage, an overall linear
decreasing trend was observed as shown in Figure 2.6. The data points were grouped based on the two different
lengths of corrosion, L1 = 100 mm and L2 = 200 mm. All the FE results have agreed to the general trend line
presented and a good fit was found for most of the experimental results. The two experimental results showing an
out of trend results suggest the possibility of non-standardized CFRP retrofitting process for them. Even though
the exact reason for such discrepancy cannot be explained, the mismatch between the FE results and the
experiments for these cases can be attributed to the application procedure of the CFRP repair system.
Figure 2.6 The trend for the ultimate load with the change in the percentage of corrosion
The simplified FE model presented appears promising to predict the bending behaviour of CFRP retrofitted
steel tubular members. The work presented in this paper was limited to only retrofitting of 2 layers of CFRP. The
model needs to be checked for its accuracy in predicting the behaviour of more than 2 layers of CFRP. Fawzia et
24
The University of Western Australia Chapter 2
al. (2006) had experimentally found the decreasing strain distribution towards the exterior layers of CFRP [8].
This consideration will have to be accommodated into the FE model to get the accurate behaviour of the CFRP
retrofitting as the number of layers increases. This could be a potential avenue for further development in this FE
model.
2.5 Conclusion
The paper presented a simplified FE modelling technique for CFRP retrofitted steel tubular members under
bending loads. The accuracy of the model in predicting the load-displacement behaviour of the CFRP retrofitted
CHS under three-point bending was demonstrated successfully. On average, there was only a variation of 4% in
the ultimate load simulated by the FE model to that of the experimental results. The paper also attempted to address
certain unexpected behaviour in an experimental program. The accuracy of the model could be further improved
by the inclusion of cohesive elements between the CFRP layers and the steel surface. Overall, an FE model which
can be easily developed for typical offshore tubular sections was demonstrated and the effectiveness of the model
2.6 References
[1] Saeed N. Composite Overwrap Repair System for Pipelines - Onshore and Offshore Application Queensland:
[2] Alexander C, Cercone L, Lockwood J. Development of a Carbon-Fiber Composite Repair System for Offshore
Risers. ASME 2008 27th International Conference on Offshore Mechanics and Arctic Engineering 2008. p. 389-
405.
[3] Elchalakani M. Rehabilitation of corroded steel CHS under combined bending and bearing using CFRP.
[4] Elchalakani M, Karrech A, Basarir H, Zhao X-L, Fawzia S, Hassanein MF. Strengthening of mild steel struts
using CFRP sheets subjected to uniform axial compression. Thin-Walled Structures. 2017;116:96-112.
[6] Hashin Z. Failure Criteria for Unidirectional Fiber Composites. Journal of Applied Mechanics. 1980;47:329-
34.
25
The University of Western Australia Chapter 2
[7] Rentsch, R., Pecat, O., and Brinksmeier, E. Macro and micro process modelling of the cutting of carbon fibre
[8] Fawzia S, Al-Mahaidi R, Zhao XL, Rizkalla S. Strengthening of circular hollow steel tubular sections using
26
The University of Western Australia Chapter 3
3.1 Abstract
Repair of offshore steel structural members become essential due to the corrosive marine environment in
which they exist. Structural repair using composite material has significant advantages in these cases than the
conventional repair by welding or bolting steel sleeves. However, the composite repair technology with fibre
reinforced polymers (FRP) is not very well established for their usage on load bearing offshore steel structural
members either in underwater or above water conditions. The study had attempted to do a pilot experimental study
on steel tubular members under the combination of axial compression and bending loads to address this gap in
knowledge. The specimens were grouped into four categories: intact, corroded, repaired in air and repaired
underwater. The repair scheme remained the same for in air and underwater repairs with one layer of glass and
two layers of carbon fibre reinforced polymers, but with resin and repair methodology being different for both.
The repaired specimens showed notable improvement in the ultimate strength from the corroded members to match
the intact strength. Underwater repairs proved to be adequate to replicate the ultimate strength achieved by the
conventional in air repairs. Investigations into parameters like load displacement behaviour, energy absorption,
ductility and strain values also revealed that the underwater repair performed very similar to the conventional in
air repair. A simplified Finite Element model simulating the repaired specimens was also presented in the paper
which had predicted the experimental behaviour with great accuracy. In conclusion, the underwater and above
water repair of structural steel tubular members using FRP composite materials seem a very viable solution for the
corroded load bearing members in the offshore industry, however more detailed study on long term performance
and cyclic behaviour of the composite repaired members is yet to be carried out.
3.2 Introduction
3.2.1 General
Structural retrofitting using Fibre Reinforced Polymers (FRP) is quite well established. However, this
technology is more practised on Reinforced Concrete (RC) structures than steel structures. While looking into the
repair of steel structures with FRPs, it is evident that it is only utilised in a limited capacity. This is because there
27
The University of Western Australia Chapter 3
is a clear knowledge gap in the performance of steel structures repaired with FRPs. The retrofitting of steel
structures is again mostly practised among the onshore steel structures and the repair of offshore and subsea
structures using FRP is very negligible. The point of concern here is on the adaptability of FRP repairs from an
onshore scenario to offshore and subsea environmental conditions. This paper has attempted to address these in
detail.
Repair of the offshore structural members will usually become necessary due to the marine environment
and its effects on corrosion induced deterioration or fatigue strength loss due to cyclic wave loads of the structure
or impact damage from vessels or dropped objects. Usually, steel pieces or sleeves are either welded or bolted into
the degraded member to regain the lost strength. However, the additional dead load created by these are a huge
structural disadvantage. Also, any welding or other repair activities will interrupt the operation of the structure
causing enormous monetary loss. Use of FRP will be a more efficient and economical alternative to address this
problem.
FRP composites are widely used to strengthen, retrofit and repair structures. Composite is a general word
used for a compound composed of at least two distinct materials with different properties. These materials are
often designed to work together to deliver better performance in terms of strength, stiffness, durability and so on.
FRPs are called fibrous composites, as they contain fibrous reinforcements embedded in a polymeric matrix.
Carbon, glass and aramid are the typical fibres used for structural purposes, in combination with thermosetting
resins such as epoxy or vinylestre [1]. Moreover, they are very lightweight, durable and flexible for application on
any surface.
FRPs had been successfully used in Reinforced Concrete (RC) structures for their strengthening and repair.
This technology is very well established and various design methods are developed to standardize their usage in
RC structures [2]. However, the usage of FRPs to retrofit, repair and strengthen steel structures is still being studied
by many researchers [3-7]. The studies of FRPs on steel structures were mainly focusing on onshore structures
which are not subjected to extreme conditions as in offshore structures. There is no proper analysis or design
The current offshore applications for FRPs to strengthen the structural members are very limited. Most of
the application of FRPs in marine conditions is for corrosion protection of non-structural elements in the topside,
leakage protection of piping systems and repair of pipelines or risers or conductors. The study on checking the
applicability of FRP for the repair of subsea pipelines, where the critical load is the internal pressure of the line,
28
The University of Western Australia Chapter 3
have shown positive results encouraging further researches in the area [1, 8-10]. However, for subsea pipelines,
the FRPs are mainly used to increase the strength of the degraded or leaking sections in hoop direction only to
assist in handling the internal pressure. Unlike the subsea pipelines, structural components in fixed or floating
platforms are usually under the combination of significant axial forces and bending moments. Hence, the potential
application of FRPs for strengthening and repair of structural components in offshore facilities (fixed, floating or
Experimental studies in the past had shown some capabilities of FRPs in structural retrofitting of steel
structures. Main parameters like the number of layers, length of the carbon fibre reinforced polymer (CFRP) wrap,
the orientation of the fibres, etc. have been studied by the researchers in the last two decades. This section presents
the key researches which had formed a basis for formulating the experimental work presented in the paper.
Bending and axial loading tests are the typical tests utilised by the researchers. Gao et al. studied the
buckling behaviour of steel tubular members strengthened with CFRP sheets by employing axial compression test
on them. The strength improvement was found to be increasing from 26% to 84% for the strengthened members
and a linear relationship from the results was developed between axial load and number of CFRP layers [11]. Short
Square Hollow Section (SHS) and Circular Hollow Section (CHS) columns strengthened with CFRP also showed
significant improvement in their strength and stiffness [3-5, 12]. Kabir et al. studied the effect of CFRP on
strengthened steel circular hollow sections by subjecting them to four point bending test. The strengthened and
unstrengthened members showed similar stiffness for smaller loads, but at higher loads, the strengthened member
showed a significant improvement in stiffness [13]. Elchalakani et al. did three point bending test on circular
hollow sections (CHSs) to study the effect of corrosion penetration, length of corrosion and type and number of
CFRP wraps. The strength and stiffness was only restored for 20% corrosion specimen to that of the control
specimen. All other levels of corrosion failed to be rehabilitated to their required strength or stiffness due to
inadequate number of CFRP layers [14, 13]. The necessity for a proper design procedure for repair using CFRP
The above mentioned studies revealed the capability of FRPs to improve the structural performance of steel
structures. However, the adaptability of this technology for structural load bearing components under offshore or
subsea environmental conditions is not investigated in detail yet. Seica and Parker were able to address this issue
partially through their bending tests on CHS retrofitted with CFRP with the parameter being in air and underwater
29
The University of Western Australia Chapter 3
curing of the specimens. The underwater curing showed a decrease in this strengthening than in air curing [16].
The study revealed the adverse conditions in a marine environment will be decreasing the performance of the
CFRP repair system considerably. The limitation of this work is that the retrofitting was still carried out in air with
The main aim of the paper was to distinguish between the onshore repair and the susbsea repair of corroded
steel members. The first objective was to have a qualitative assessment of the steps involved in both these repair
procedures. The second objective was to quantitatively assess both kinds of repairs in terms of ultimate strength,
load displacement behaviour, energy absorption, ductility and strain values in tubular members under axial
concentric and eccentric loads. The final objective was to develop a simplified finite element model which can
replicate the behaviour of such structural members with composite repairs so that future designs can be made more
effectively.
3.3.1 Specimens
In order to investigate the structural behaviour of FRP strengthened offshore structural members, in this
study series of experimental load tests were carried out on tubular specimens under concentric and eccentric axial
loads. Specimens presented in this paper can be categorised into four groups namely intact, corroded, repaired in
air and repaired underwater. Intact members were bare steel members without any damage or corrosion which
would act as a control group. Corroded members were bare steel specimens which were machined at the mid span
to reduce the thickness of the pipe. This was done to simulate the effect of wall thinning in offshore structural
members due to corrosive environments. Specimens identical to the corroded members were used to prepare the
next two groups of repaired specimens. Repaired in air was the group of specimens which were prepared by using
normal repair consumables and techniques employed for onshore steel repair using FRP. On the other hand,
repaired underwater was the group of specimens which were prepared using consumables and techniques specific
for the subsea repair of steel. Every group had 5 specimens each. The first specimen in every group was a
concentrically loaded specimen under pure axial compression. The remaining 4 specimens in a group were under
an eccentric axial compressive load at different eccentricities as 20 mm, 40 mm, 60 mm and 80 mm representing
30
The University of Western Australia Chapter 3
The convention utilised to name the specimens was quite straight forward. The intact specimens were
named with ‘I’ in the beginning whereas a corroded or corroded and repaired specimens’ name started with ‘C’.
The axial loading scenario was described in the second part of the specimen name. A pure axial specimen was
named with ‘AE0’ to describe the loading scenario as axial loaded with 0 mm eccentricity. Similarly the axial
specimens with other eccentricities were named. When the specimen was bare steel, the repair scheme was
represented as ‘F0’ describing the usage of 0 layers of fibre. The repaired specimens had 2 layers of CFRP wrapped
on them and hence they were named ‘F2’ following the same logic as above. The repair in air and the repair
underwater specimens were distinguished with ‘A’ and ‘W’ at the end of the specimen names respectively.
All the specimens had tubular cross-section with outside diameter, Do = 114.3 mm and wall thickness, t =
6 mm and were made of API 5L Grade X52 steel material. The Young’s modulus of the pipe material is 200 GPa.
The length of all specimens, L was equal to 1800 mm and a length of corrosion, Lc = 300 mm was machined in
the mid span of the corroded and repaired specimens to remove the corrosion thickness, tc = 1.2 mm (i.e., 20%
thickness reduction due to corrosion). For the ease setting up the specimens in the test equipment, they were welded
on to 10 mm thick steel end plates with a 10 mm thick continuous fillet weld on both ends as shown in Figure 3.1.
Table 3.1 summarises these geometrical properties as well as the mechanical properties of the steel cross-section
at the mid span such as moment of inertia (I), elastic section modulus (Zs), plastic section modulus (Zp),
compressive yield strength (Pyield), elastic moment capacity (Mys) and plastic moment capacity (Mps).
31
The University of Western Australia Chapter 3
The repair materials used for this study were all TechnowrapTM products [17]. TechnowrapTM is an accepted
and approved product by the industry for composite patch repair works. CFRP is known to cause galvanic
corrosion in the steel members when they come in direct contact with steel substrate and it could become a very
serious concern [18]. Hence, an E-glass fibre, Technowrap 2KTM was used as the first layer of repair to prevent
galvanic corrosion between steel and carbon fibre. The TechnowrapTM 2K was a triaxial woven cloth with fibres
running in 0o and ±45o. This cloth would give an average thickness of 0.8 mm as a lamina. TechnowrapTM SRS is
the type of CFRP used to provide structural strength to the repair. These quad axial woven cloths with carbon
fibres running in 0o, 90o and ±45o were used as the subsequent layers after the glass layer. On an average, a single
layer of the carbon fibre would show a thickness of 1 mm as a lamina. The fibre cloths used for this study are
shown in Figure 3.2. The normal TechnowrapTM resin system was used for the repair in air specimens whereas the
Subsea TechnowrapTM resin system was utilised for the underwater repair.
Figure 3.2 Tri axial GFRP cloth (left) & Quad axial CFRP cloth (right)
The same repair scheme was used for both the conventional repair and the subsea repair. The steel pipe’s
nominal diameter, D = 100 mm was used as the basis for the repair scheme. The machined corroded area of length
300 mm (= 3D) and was considered as the defect region with 1D landing on either end. Appropriate tapering was
given on the edges of the repair to ensure a smooth transitioning from the repair to the bare steel, thus negating
any possibilities of issues associated with stress concentration. The repair length, LR came to approximately 6D or
600 mm which is a sum of defect length and landing and taper on both sides. The glass fibres were extended
slightly beyond this repair length to ensure there was no chances of galvanic corrosion due to the interaction of
carbon fibres with steel. The repair details are tabulated in Table 3.1.
32
The University of Western Australia Chapter 3
In order to obtain a reliable composite behaviour, it is firstly essential to have a good bond between steel
and the first CFRP lamina and between CFRP laminas. Existing literature indicated that debonding (i.e., breakage
of bond between the substrate and the composite material) and delamination (i.e., separation of layers of the
composite material) were the main failure modes of the experiments done with FRP and steel [13-15]. Since the
forces would be transferred from the steel to the FRP through the adhesive, to take advantage of the full capacity
of the composite the adhesive should be able to transmit the forces efficiently and it was preferred that the fibres
ruptured before a failure of the adhesive occurred. Some investigators have suggested that, when adhesives were
not fully capable of transferring these forces, additional fasteners may be employed [19]. The experimental study
presented in the paper did not employ additional fasteners as it is very challenging to apply them on circular
profiles. The effectiveness of the subsea resins in comparison to the conventional onshore resin in this regard was
Two different repair processes were employed in the study. The first repair process was carried out in air
using the consumables conventionally utilised for onshore or above water repair of steel structures. The second
repair process was carried out underwater using techniques and consumable specific for subsea repair.
The first and foremost step in the repair process was the steel surface preparation. Bond between the FRP
lamina and the steel substrate is due to chemical and mechanical adhesion. The chemical adhesion was derived
from the resin system used for the wrapping of the fibres and the surface preparation ensured the mechanical
adhesion. The surface profile was created by employing an air blast equipment and later cleaned using acetone. In
case of underwater specimens, foam blasting could be employed to achieve the same standards of surface
preparation. However, the underwater repair specimens presented in the paper were prepared using air blasting,
the same technique used for in air specimens due to practical difficulties. This would not have a serious impact on
the results of the specimens repaired underwater. A surface cleanliness of Sa2.5 and a roughness profile with a
minimum of 60 µm was achieved for both in air and underwater specimens. The surface roughness was measured
by using a Testex tape to ensure the required profile. The surface preparation was done in accordance with ISO
The second step was the wrapping of the FRP layers. The glass fibre reinforced polymer (GFRP) was
impregnated with resin and wrapped around the prepared steel surface. The wrapping was carried out in two stages.
In the first stage, the GFRP was wrapped around the steel surface. A tether cloth was placed on top of the wrapped
33
The University of Western Australia Chapter 3
GFRP to develop a rebondable surface for the subsequent repair layers. Stricture bands were employed to push out
the extra resin and trapped in air from the repair. This stricture banding will be ensuring a level of consolidation
of the wrapped repair and was removed only after the soft curing period for the GFRP wrap. In the second stage,
the 2 layers of impregnated CFRP layers were wrapped and stricture banded in one go. The stricture bands were
removed after the soft curing period and the repaired specimens were left idle for a day for the curing process to
complete.
The wrapping process in the above paragraphs was describing the in air repair which was completely carried
out in air. The underwater wrapping was carried out almost identical to the in air repair wrapping with major point
of difference is the application of the entire process under the water as shown in Figure 3.3. The water was
collected from Swan River in Perth, from a location not very far from the Indian Ocean, and the collected water
was pooled in a tank. The repair process and the curing of the underwater specimens were carried out underwater
in the tank. Development of a rebondable surface on the GFRP layer using a tether cloth for underwater specimens
is not an easy task and therefore this step was eliminated from the wrapping procedure for underwater specimens.
Another point of difference was the use of specific resin for underwater or subsea applications. Overall, the
underwater repair process is harder to execute with a similar level of quality of the in air repairs. Skilled technicians
Figure 3.3 Underwater repairing: (a) Repair tank filled with water from Swan River, (b) Wrapping of
fibre layers, (c) Stricture consolidation and (d) Curing
34
The University of Western Australia Chapter 3
Specimen preparation was a key step in the testing setup. The black exterior surface of the CFRP repair
was gridded out using white markers to better visualise the local deformation and damage location in the repair
while they were being tested to failure. All specimens were gauged with four foil strain gauges on the steel surface
in case of bare steel specimens and on the exterior surface of CFRP of the repaired specimens. On both the tension
and compression faces of buckling, two strain gauges each were placed at the mid span, one in the axial direction
(along the length of the pipe) and one in the hoop direction (around the circumference of the pipe). These
The testing equipment used for the experiments was an Amsler loading machine in the structural laboratory
at the University of Western Australia. The specimens were connected to roller plates at the top and bottom to
create hinge support conditions for the specimen. The specimen was later lifted and fitted into the testing
equipment and the axial load was applied and then gradually increased until the overall failure of the specimen.
Once the ultimate load was reached the loading continued until the axial displacement has reached at least 5% of
length of the specimen, (L = 1800 mm). A general overview of the experimental setup and the data monitoring is
presented in Figure 3.5. In a practical scenario, such repairs would be done on a member which is already under
structural loads. The experimental setup presented here has some limitations in this regard.
35
The University of Western Australia Chapter 3
Figure 3.5 Experimental setup: (a) Specimen setup in the Amsler just before testing, (b) Buckled specimen
during testing and (c) Digital data monitoring of the experiment
The compilation of all the ultimate loads and relevant comparisons from the 20 specimens are presented in
Table 3.2. The most important finding from the tests is that no indication of any premature failure is observed
before the attainment of the ultimate load. This reflects good surface preparation and an adequate bond between
the FRP and steel as well as a good application procedure for the repair system. The intact specimens showed the
base capacity of a steel member under a combination of axial compression and bending moment. The corroded
specimens showed a reduction in capacity due to the machined corrosion equal to 20% of the wall thickness for Lc
= 300 mm. The maximum reduction of 28.5% was observed for the pure axial loading and the other specimens
showed descending reduction percentages as the eccentricities increased. The repaired specimens showed
improvement in their ultimate strength when compared with the corroded bare steel specimens. Figure 3.6 shows
the repaired specimens after testing. The repairs showed an average improvement of 28.3% in their ultimate loads.
CAE80F2W (i.e., corroded sample that was wrapped with 2 layers of CFRP under axial load with 80mm
eccentricity) is the only specimen which showed out of pattern results. The ultimate load observed for CAE80F2W
36
The University of Western Australia Chapter 3
was having an improvement of only 4.9% whereas the rest of the repaired specimens showed an improvement of
around 30%. Further investigations showed that this could be attributed to slight initial damage in the GFRP lamina
used in this specimen and then a premature debonding of the GFRP during early steps of the loading. Therefore,
the test results from this specimen were eliminated to make any conclusions and interpretations from this study.
Figure 3.6 Repaired specimens after testing: In air repair (left) & Underwater repair (right)
The Strength Regain Ratio (SRR) was calculated as per equation (3.1) which gave an insight into how the
repair had been able to bring the ultimate load from the corroded level to the intact capacity and were tabulated in
Table 3.2.
_
SRR (3.1)
_
where, Pu_repair is the ultimate load of the repaired specimen and Pu_control is the ultimate load of the
corresponding control bare steel intact specimen. The SRR determined for the repaired specimens were very close
to 1.0 showing the adequacy of the repair with two CFRP layers to regain the capacity loss due to 20% of wall
thinning for all of the tested structural members. The SRR from the underwater repairs and the in air repairs were
examined against each other to understand the difference between both repair technologies. The examination
revealed that the subsea repair technology was performing equally well to that of the conventional in air repair
technology.
The axial load versus axial displacement responses for the specimens with different eccentricities of 0, 20,
40, 60 and 80 mm for intact, corroded, repaired in air, and repaired underwater specimens are presented in Figures
3.7, 3.8, 3.9 and 3.10 respectively. The axial displacements are measured between the top and bottom of the
specimen. It was seen that the load displacement response in overall was very similar to typical buckling behaviour
such that an elastic phase was followed by a peak at the ultimate buckling load then followed by a structural
collapse of the column associated with large bending deformation resulted from the instability of the column. The
37
The University of Western Australia Chapter 3
peak was sharp for the concentric loading regime whereas a smoother peak was found for the eccentric loading
scenario. The ultimate load in all specimens was associated with a combination of overall buckling and yielding
of the member and the structural collapse (failure) was associated with local buckling and excessive local
deformations at the mid-span. The load displacement behaviour provided a greater insight into the behaviour of
the specimens than the ultimate strength. The specimens under pure axial compression or with no eccentricity
showed the highest ultimate strength and stiffness in each group. As the eccentricities were increased the ultimate
strength and the slope in the elastic portion of the load displacement curve decreased within a group of specimens.
The general trend was consistent among the four different sets of load displacement behaviour.
38
The University of Western Australia Chapter 3
The comparison plots for the load displacement behaviour for the five different loading conditions namely
eccentricity equal to 0 mm, 20 mm, 40 mm, 60 mm and 80 mm were presented as Figures 3.11, 3.12, 3.13, 3.14
and 3.15 respectively. This comparison showed that how the repaired specimens for all different cases of eccentric
loads had improved the performance from a corroded level. In all the cases, a significant improvement from the
corroded specimens was observed for both the specimens repaired in air and underwater. The slope in the load
displacement curve in the elastic region did not show much variation from the corroded specimens to the repaired
specimens (refer Table 3.2 for the elastic slope values). This confirms the fact that the stiffness of the repaired
39
The University of Western Australia Chapter 3
40
The University of Western Australia Chapter 3
In case of the repaired specimens, the axial load versus axial deflection behaviour showed certain jerks in
the unloading part of the curve, which were indicative of the sudden failure of the repair material. During testing
of the repaired specimens, cracking sounds were heard for all of them once the specimen has already past the
ultimate load. These cracking sounds, associated with minute irregularities (jerks) in the unloading part of the load
displacement curve, were indications of different failures in the repair itself. The failures in the repair could be
debonding of GFRP layer from the steel surface, rupture of GFRP, crushing of CFRP on the compression face or
41
The University of Western Australia Chapter 3
The comparison plots also enlightened how the repaired specimens were structurally able to recover to the
intact condition. The recovery was full in all cases except for the pure axial compression specimens where the
recovery was around 90% of the intact specimen. The load displacement behaviour of all the repaired specimens
showed that they were able to bring the corroded specimens into the intact levels for the most part. However, local
buckling in the repaired specimens triggering some portions in the structural collapse (post peak) curve to drop
below the intact levels was observed as well. Local buckling in a specimen could be identified by a sudden or
sharp change in the slope of the structural collapse curve. Even though the local buckling in repaired specimens
was shifted to occur at a larger displacement than the corresponding corroded specimens did, the repair was not
able to keep up with the behaviour of intact specimens once local buckling was initiated.
In addition, the difference in performance between a conventional repair (repair in air) and a subsea repair
(repair underwater) was also examined using the comparison plots shown in Figure 3.11 to 3.15. They behaved
almost identical in the elastic portion of the load displacement curve, whereas the underwater repairs had shown
slightly inferior ultimate strengths. The post peak load displacement behaviour of the underwater repairs was
marginally lower than the in air counterparts too. These small drawbacks of underwater repair compared to the in
42
The University of Western Australia Chapter 3
air repair could be attributed to lack of a proper rebondable surface in the underwater specimens, difficulties in
executing the repair process underwater and the difference in the resin system. Overall, as far as the elastic stiffness
and ultimate strength of specimens are concerned, the performance of underwater repairs was remarkably close to
their in air counterparts which is a promising sign for future application of FRP repair in subsea and marine
environments.
The slope of the post peak curve for the repaired specimens was less steep compared to the corroded
specimens indicating better energy absorption in the repaired specimens. The Energy Absorption (EA) of a
specimen was calculated as the area under the load displacement curve until it reaches the failure point with 5%
The calculated EA values are tabulated in Table 3.2 and confirm better energy absorption among the
repaired specimens. The energy absorption decreased from intact to corroded specimens as expected. The
maximum reduction of 48.3% was in case of the pure axial specimen, CAE0F0. The conventional in air (onshore)
repair improved the energy absorption for all specimens in comparison to the corroded specimens. The maximum
improvement of 95.4% was recorded for the specimen loaded at an eccentricity of 20 mm (CAE20F2A). Except
for the pure axial specimen (CAE0F2A), all the conventional repairs were able to regain their EA values to their
The underwater repair produced improvement in the EA values from the corroded specimens, but the
values were lower than that of the conventional repairs. The maximum EA for subsea repaired specimens, EA =
12 kJ, was observed for the pure axial specimen (CAE0F2W). In comparison to the intact specimens, the subsea
repairs were not able to reach the EA levels except for the specimen loaded with an eccentricity of 20 mm
(CAE20F2W). On average, the energy absorption value of a subsea repair was found to be about 79% of that of
the repair in air. The major point of difference between the in air and underwater repairs was in terms of their EA
values and this information would be very helpful in deciding how the repair design should vary for a subsea or
3.4.4 Ductility
The composites are brittle on their own and hence it can potentially be an issue of concern when such
repairs were suggested for onshore and offshore steel structures in which ductile behaviour is of quite important.
43
The University of Western Australia Chapter 3
Comparing load displacement behaviour of the repaired specimens with those for the intact or corroded specimens
show a reasonable ductile behaviour. However, ductile performance of the repaired specimens is qualitatively
studied by calculating Ductility Index (DI) for the specimens. Two different methods of calculating DI were
Method I for calculating DI is an energy based method which is a modified version of the method by Foster
and Attard (1997) [21]. In method I, the DI is calculated as given in equation (3.2) below.
DI (3.2)
where, ABC is the area under the load displacement curve up to Δ75, the deflection point obtained by
extending the linear elastic part of the curve (0 to 0.75Pu) to peak load (Pu), which is the elastic energy absorbed
by the specimen and ADE is the area under the load displacement curve up to an axial displacement equal to 15Δ75,
which is the total energy absorbed by the specimen. The identification of the five points, A to E is further illustrated
in Figure 3.17.
Method II used in the study for DI calculation is a displacement based method and DI is calculated as per
DI (3.3)
where, d1 is the displacement corresponding to 75% of the ultimate load (0.75Pu) and d2 is the
displacement corresponding to the local buckling point in the unloading part of the load displacement curve as
44
The University of Western Australia Chapter 3
The DI values calculated using both methods are tabulated in Table 3.3 with the ratios of them with the
corresponding corroded DI value. The ratio gave an insight into how the ductility has varied from the intact level
due to the corrosion and subsequently how the repairs were able to improve the ductility. In case of the intact
specimens, method II was not applicable as the local buckling points were not identifiable from the load
displacement graphs. Hence, logical assumptions were made for the local buckling points of intact specimens to
The DI calculated using method I showed the corroded specimens were less ductile than the intact
specimens by about 15% on average. The specimens repaired in air showed an improvement in DI from corroded
level and they all had better DI than their corresponding intact specimens except for CAE80F2A which was only
less than 3% lower than the intact DI. The mean improvement in DI for the repairs in air from the corroded level
was 26%. The specimens with underwater repairs showed DI values lesser than repairs in air, highlighting the fact
that the underwater repairs have lesser ductility than conventional repairs. On average, the underwater repaired
specimens possessed only 98% of the DI for corroded specimens. Hence, method I did not show ductility
DI values calculated from method II also gave some useful inferences. The corroded specimens showed
much lower DI than their corresponding intact specimens and this could be attributed to the selection of local
buckling points of intact specimens. The comparison of DI values of intact ones using method II to the rest of the
specimens is not appropriate for the same reason as above. All repaired specimens showed better ductility indexes
than corroded specimens using this method except for CAE60F2W which was only 3% lower than the corroded
DI. Moreover, the in air repaired specimens showed better DI than the specimens repaired underwater. The DIs of
in air repairs were 149% higher than corroded values on average whereas the average comparison of DIs of
underwater repairs was 43% higher than corroded values. Methdod II was able to identify the improved ductility
of underwater repairs from corroded ones, which was not possible with method I.
The comparison of how the DI from both methods varied is presented in 3.18 as a bar graph. The intact DI
calculations were adjusted to have a similar pattern from both methods. However, the DI of corroded and repaired
specimens showed an average jump down of around 8 from method I to II. The energy method (method I) had
shown better capability of picking up the ductility for the entire set of specimens which is evident through the
jump, although the displacement method (method II) was also able to portray some useful conclusions. Irrespective
of the DI calculation method, it can be concluded that the underwater repairs possess slightly less ductility than in
45
The University of Western Australia Chapter 3
air repairs. Future utilisation of this repair technology should be after careful consideration the lower ductility of
The load versus axial strain plots are presented in 3.19 (a), (b), (c) and (d) for intact, corroded, repaired in
air and repaired underwater respectively. The load versus hoop strain plots are presented in 20 (a), (b), (c) and (d)
for intact, corroded, repaired in air and repaired underwater respectively. The strain values observed from the
experimental specimens followed a unique pattern. The specimens, which were under pure axial load (i.e., no
eccentricity), showed compressive axial and tensile hoop strains on both faces of buckling until reaching the
ultimate load as shown in Figure 3.19 and Figure 3.20 respectively. Post peak load, the strains on the tension face
of the specimens with e = 0 mm reversed their nature. On the other hand, the specimens under eccentric loading
showed tensile and compressive axial strains on the tension and compression faces of buckling respectively. The
hoop strains showed the reverse nature of the axial strains for these specimens. As the eccentricity of the specimens
increased, the deviation between tensile and compressive strains in a specimen widened both in case of axial and
hoop strains. The general pattern of the strain values was consistent with the bare steel specimens as well as the
46
The University of Western Australia Chapter 3
47
The University of Western Australia Chapter 3
The strain values were measured at the ultimate load of each specimen and they are listed in Table 3.4. The
axial strains at the ultimate load among the bare steel (i.e. intact or corroded) specimens had a maximum of 1.13%
(compressive) and 0.33% (tensile) whereas the corresponding ones for the repaired specimens were 1.32%
(compressive) and 0.36% (tensile). Generally, the repaired specimens showed strain values slightly higher than
that of their bare steel counterparts. Similarly, the hoop strain values for the bare steel (i.e. intact or corroded)
specimens had a maximum of 0.12% (compressive) and 0.63% (tensile) and the repaired specimens had a
maximum of 0.21% (compressive) and 0.42% (tensile). The hoop strains were considerably lower due to Poisson
effect than the axial strains and this was obvious as the loading was predominantly in the axial direction.
The strain values measured at the local buckling point of each specimen were also investigated. However,
this data is very limited as most of the strain gauges have reached their limits before reaching the local buckling
point or they have been dislodged from their position before reaching this point. The available strain values at the
The mechanical properties of laminates with different types of fibres reported by the manufacturer are
provided in Table 3.5 [22]. These properties were based on tensile coupon tests performed on the fibre laminates.
It can be found out from this data that the strain at failure for a GFRP laminate at 42% fibre volume fraction (=
Fibre volume/Total lamina volume) was 3% whereas for a CFRP laminate at 40% fibre volume fraction was 1.4%.
The maximum strain from each specimen was also investigated and the values are listed in Table 3.4. The
extreme strains at the last stages of testing for some specimens were not captured due to damage of the strain
gauges which could lead to potentially some erroneous inferences from this data. The maximum compressive strain
of 1.89% was recorded as an axial strain on the compression face of CAE60F2A among all repaired specimens.
This strain value was larger than the recommended strain at failure, but this could be since the recommendation
was based on tensile strain at failure. Likewise, the maximum tensile strain of 1.39% was observed as the axial
strain on the tension face of CAE80F2W which was very close to the limits recommended by the manufacturer.
The average maximum axial strain values for the repaired specimens were found to be around 1% in both tension
and compression which was lower than the recommended value of 1.4%. These observations are mostly in
agreement with the design recommendations by the manufacturer, however, certain gaps in knowledge are also
identified. Updating the design recommendations with the help of more accurate test data and specifying separate
design values for tensile and compressive loading cases are necessary for further propagation of this technology.
48
The University of Western Australia Chapter 3
A Finite Element (FE) model was developed using commercially available software Abaqus version 6.14-
1 [23]. The model tried to predict the experiments presented in this paper and were successful to a great extent.
The FE model tried to model the composite repair on steel structures in a simplified and practical manner so that
the computational expense is minimised at the same time able to predict the load displacement behaviour
effectively.
The experimental specimens had pinned end connections facilitating rotation about the axis of the pipe.
The FE model in Abaqus was given boundary conditions in such a way to replicate this. The bottom support for
the pipe had all its rotations free and translations restricted. The top support had the rotations free and the
translations restricted except the translation along the axis of the pipe. The loading was provided by means of a
fixed displacement of 200 mm along this free translation. The axial load can be found out as the reaction force
created due to the introduced displacement. The load displacement plots were extracted accordingly and compared
The steel pipe was modelled a 4-node doubly curved shell element with reduced integration, hourglass
control and finite membrane strains (S4R). The shell was modelled with a thickness of 6 mm. The pipe was
portioned into the corrosion region for 300 mm in the mid span and thickness for this region was reduced by 20%.
Meshing with a global size of 5 mm was utilised for obtaining the best results in terms of accuracy and
computational time.
The repair laminas were modelled using the composite layup, which is an inbuilt tool in Abaqus for layering
up materials with different properties. Composite layup technique models the repair in such a way the lamina is
being tied to the steel surface. However, this tool would be able to avoid all the numerical instabilities with a
conventional surface to surface tie model. This tool has a limitation of not simulating the debonding and
delamination failures between the composite layers which was demonstrated by George et al. [24]. Modelling of
an adhesive layer as the interface between the composite layers or substrate and the first composite layer is another
alternative [25]. Due to the complex behaviour of resin with the steel surface and the fibre layers, accurate
modelling of the interface layer is a very difficult task and the introduction of a numerical model for such an
49
The University of Western Australia Chapter 3
interface needs complex type of elements with excessive computational efforts. Moreover, it was reported earlier
that no debonding or delamination was observed in the experiments until the ultimate load was reached. Composite
layup technique thus became the most appropriate and practical solution for modelling the composite repair in
these specimens where estimation of the ultimate strength is the main objective of the FE model.
The composite layup utilised to model the composite repair on the steel pipe had five different plies. The
first ply was the steel pipe followed by the ply for the corroded region which was an elastic filler layer to develop
a constant profile for the following repair laminas. The first repair ply was GFRP layer which was oriented at 90o.
First and second plies of CFRP laminas were added on top of the GFRP lamina ply. The orientation of these plies
was not significant as the CFRP lamina had the same properties in both directions. Since the composite layup tool
modelled the repair laminas as a single part with the base steel pipe, the element type and the mesh density were
the same as that of the steel pipe. More details about the geometry of the repair modelled in Abaqus can be found
Figure 3.21 Geometry for the FE model of the tubular member with composite repair
The steel pipe was defined as an elastic-plastic material with strain hardening. Steel properties were defined
based on the material testing done on the pipe material. The yield strength was used as 345 MPa with the stress at
a plastic strain of 0.5% as 360 MPa. The corroded region was filled with an elastic material with negligible
strength.
50
The University of Western Australia Chapter 3
The elastic properties of the repair laminas were defined based on the recommendations given by the
manufacturer of the composite materials used for the experimental specimens [22]. The elastic properties used for
the definition of GFRP and CFRP in the FE model are tabulated in Table 3.6. The GFRP lamina was defined with
elastic moduli, E1 and E2 as 19 GPa and 8 GPa respectively with ‘1’ as the stronger orientation. The GFRP lamina
was wrapped around the pipe with the stronger direction, ‘1’ around the pipe and ‘2’ along the length of the pipe.
This was done to simulate the actual repair where the weaker GFRP direction was used along the length of the
pipe as it was used as a galvanic corrosion insulation rather than a structural component. Shear moduli of 5.9 GPa,
taken from the manufacturer recommendation and a Poisson’s ratio of 0.32 were used for defining the GFRP
lamina [22]. Similarly, the CFRP lamina was defined with elastic moduli of 36 GPa in both direction as the carbon
fibres were quad axial woven cloth making the lamina equally strong in both directions. Based on a volume fraction
of 40%, the shear modulus was calculated as 13.5 GPa by the manufacturer. However, they recommended using a
conservative value of 2 GPa for shear moduli for design purposes [22]. Hence, shear moduli of 2 GPa was chosen
and a Poisson’s ratio of 0.32 was used for the CFRP lamina.
Damage in composite materials were typically defined utilising Hashin damage model [24-26]. Usually,
the use of Hashin damage concept in FE models for structural analysis of composite sections lead to numerical
complexities and instabilities in the analysis as well as computationally expensive simulations. As a result, FE
models with Hashin damage approach using reasonable computational was not able to progress beyond the
ultimate load stage. In order to negate the complications associated with the Hashin damage model, a simplified
plastic damage modelling was adopted in this study for the model. The utilized plastic damage model is simplifying
the different damage strengths of a composite material into a single plastic yield strength. Fundamentals for this
modelling are based on the plastic hinge analysis for steel and FRP composite members presented by Elchalakani
et al. [6, 14-15]. This modelling approach limits the ability of the model to simulate multiple modes of damage in
the composites simultaneously. This means that in the presented FE model the different modes of failure in the
composite repair like debonding, fibre rupture and fibre crushing will not be simulated and instead a plastic
yielding failure will be simulated which is only a representation of the actual behaviour. The plastic yield strength
was to be defined in such a way that it can capture the behaviour of the composite repair in the most efficient
manner. For the FRP laminas used in this study, no specific yield or ultimate strength is given by the manufacturer
and therefore a trial and error study was conducted to find the best suited plastic yield strength for the CFRP
laminas. A plastic yield strength of 135 MPa was found to be the most suitable for simulating the set of specimens
presented in this paper. However, due to the simplified damage modelling approach the ability of the developed
51
The University of Western Australia Chapter 3
FE model in predicting the post peak behaviour of the repaired specimens is considerably less than that one for
3.5.4 Analysis
The experiments carried out here were simulated in Abaqus utilising a two-step buckling analysis as
demonstrated by Hassanein et al. [27]. The first step was to conduct a linear elastic buckling analysis on the pipe
FE model to find the first Eigen mode of buckling. The first Eigen mode of buckling was used to simulate both
the local and global imperfections in the steel pipe [27]. The second step was to conduct the non-linear buckling
analysis in which the imperfection in the pipe was introduced by scaling the first Eigen mode shape with a factor
of L/1000, where L is the length of the pipe. The results from this analysis step were simulating the actual
Validation of the FE model for bare steel control specimens was done firstly. Model for IAE0F0 and
CAE0F0 (i.e., intact and corroded specimens under pure axial loads) were developed and the load displacement
behaviour was compared with the experimental results as shown in Figure 3.22. The FE models produced a good
match with the experimental results with the ratios of the ultimate load from the FE to that of the experiment
(PFE/PExp) being 0.99 and 1.00 for IAE0F0 and CAE0F0 respectively. The visual check of the global buckling and
the local behaviour from these models revealed close resemblance as shown in Figure 3.23 (a) and (b) respectively.
The negligible difference, mainly observed in the post buckling phase, is mostly due to the differences between
52
The University of Western Australia Chapter 3
Figure 3.23 Visual comparison of FE models with experimental bare steel specimens
As the behaviour of the in air and underwater repaired specimens showed very similar behaviour, only one
FE model was used to simulate both. A total of five different FE models were created for the repaired specimens
and their load displacement behaviour are presented in comparison with the corresponding experimental data in
Figure 3.24. A very close match can be observed for all the FE models with the experimental data except for the
experimental results of CAE80F2W, which is suspected to have some pre-existing damages in the GFRP layer.
Table 3.7 shows that the PFE/PExp ratios vary from 0.93 to 1.19 with an average ratio of 1.00 and a standard deviation
of 0.07. Furthermore, the visual comparison of the simulation and the experiment can be found in Figure 3.25.
These results proved the ability of the FE model to predict the behaviour of composite repair on steel pipes under
combined axial compression and bending loads. Such finite element models would be very helpful in
understanding and designing such repairs in the future. It is evident that the developed FE model is not able to
accurately present post buckling behaviour of the specimens, however, the FE model results are in very good
agreement with the test results up to the ultimate strength of the samples.
53
The University of Western Australia Chapter 3
54
The University of Western Australia Chapter 3
Figure 3.25 Visual comparison of the FE model (CAE0F2_FE) with the experimental repaired specimen
(CAE0F2A)
3.6 Conclusions
Series of tubular sections in intact, corroded, repaired with CFRP wraps in air and repaired with CFRP
wraps underwater under the combined action of axial load and bending moment were tested experimentally in this
study. The experimental study was a first of its kind in many aspects. The use of FRP wraps to repair heavily
loaded steel structural members in subsea conditions had not been investigated by previous researchers. The
comparison of how an underwater repair fares with a conventional in air repair is a significant finding of this study.
However, this study was addressing only a limited application of this technology. Future studies should be aimed
at expanding the understanding of long term performance of such repairs to develop a proper design methodology
for structural repair of offshore steel structures using FRPs. A numerical FE model using Abaqus software was
also presented in this paper aiming to initialise such developments. Study on long term performance and cyclic
behaviour of the underwater composite repairs needs to be carried out too. For the tubular sections, imposed
corroded wall thickness and the composite materials used in this research, the following conclusions can be drawn
The most important observation from the tests is that no premature debonding occur before the
attainment of the ultimate load in the repaired specimens. This reflects good surface preparation and
CFRP wraps (in air or underwater) can be used to structurally retrofit corroded tubular members under
axial compressive and bending loads to restore their capacity to non-corroded or intact levels. Use of
55
The University of Western Australia Chapter 3
a repair with 1 layer of GFRP and 2 layers of CFRP was enough to restore the ultimate capacity for
corroded sections with 20% corrosion in the wall thickness at mid span to the intact capacity.
Subsea or underwater composite structural repairs can be successfully executed with almost equivalent
Both underwater and in air CFRP repaired sections have better energy absorption capacity than
corroded members. Energy absorption of the in air and underwater repaired sections are higher and
CFRP retrofits have provided improved ductility for the repaired specimens from the corroded
condition. The DI values found for the underwater repairs were lower than the in air repairs
highlighting the difference in the performance of these two repairs. The lower ductility of the
The average maximum axial strain values for the repaired specimens were found to be around 1% in
both tension and compression which was lower than the recommended strain at failure of 1.4%. The
importance of updating the design recommendations and establishing separate guidelines for composite
application in tensile and compressive loads can be understood from the observed strain data from the
experiments.
The results from the simplified FE model demonstrated in the paper are in very good agreement with
the experimental test results in the estimation of ultimate strength for the repaired specimens.
3.7 Acknowledgements
The authors would like to thank Jason LeCoultre from ICR.IAS Joint Venture for providing composite
materials and expert labour for the preparation of the repaired specimens presented in the paper. The authors would
also like to acknowledge the efforts of Master of Professional Engineering (MPE) students, Liam McGurdy,
Michael Mason, Anjali Gayathri Konara and Kit Strickland in the conduct of the experiments through RiverLab
56
The University of Western Australia Chapter 3
3.8 References
[1] Saeed N. Composite Overwrap Repair System for Pipelines - Onshore and Offshore Application Queensland:
[2] ACI. Guide for the Design and Construction of Externally Bonded FRP Systems for Strengthening Concrete
[3] Bambach MR, Elchalakani M. Plastic mechanism analysis of steel SHS strengthened with CFRP under large
[4] Bambach MR, Elchalakani M, Zhao XL. Composite steel–CFRP SHS tubes under axial impact. Composite
Structures. 2009;87:282-92.
[5] Bambach MR, Jama HH, Elchalakani M. Axial capacity and design of thin-walled steel SHS strengthened with
[6] Elchalakani M, Fernando D. Plastic mechanism analysis of unstiffened steel I-section beams strengthened with
[7] Elchalakani M, Karrech A, Basarir H, Zhao X-L, Fawzia S, Hassanein MF. Strengthening of mild steel struts
using CFRP sheets subjected to uniform axial compression. Thin-Walled Structures. 2017;116:96-112.
[8] Alexander C, Cercone L, Lockwood J. Development of a Carbon-Fiber Composite Repair System for Offshore
Risers. ASME 2008 27th International Conference on Offshore Mechanics and Arctic Engineering 2008. p. 389-
405.
[9] Shouman A, Taheri F. An Investigation into the Behaviour of Composite Repaired Pipelines Under Combined
Internal Pressure and Bending. ASME 2009 28th International Conference on Ocean, Offshore and Arctic
[10] Shouman A, Taheri F. Compressive strain limits of composite repaired pipelines under combined loading
[11] Gao XY, Balendra T, Koh CG. Buckling strength of slender circular tubular steel braces strengthened by
57
The University of Western Australia Chapter 3
[12] Haedir J, Zhao X-L. Design of short CFRP-reinforced steel tubular columns. Journal of Constructional Steel
Research. 2011;67:497-509.
[13] Kabir MH, Fawzia S, Chan THT. Effects of layer orientation of CFRP strengthened hollow steel members.
Građevinar. 2015;67:441-51.
[14] Elchalakani M. Rehabilitation of corroded steel CHS under combined bending and bearing using CFRP.
[15] Elchalakani M, Karrech A, Basarir H, Hassanein MF, Fawzia S. CFRP strengthening and rehabilitation of
[16] Seica MV, Packer JA. FRP materials for the rehabilitation of tubular steel structures, for underwater
[18] Shamsuddoha M, Islam MM, Aravinthan T, Manalo A, Lau K-t. Effectiveness of using fibre-reinforced
polymer composites for underwater steel pipeline repairs. Composite Structures. 2013;100:40-54.
[19] Razavi Setvati M, Mustaffa Z, Shafiq N, Syed ZI. A Review on Composite Materials for Offshore Structures.
ASME 2014 33rd International Conference on Ocean, Offshore and Arctic Engineering 2014.
[20] ISO. Petroleum, petrochemical and natural gas industries - Composite repairs for pipework - Qualification
[21] Stephen JF, Mario MA. Experimental Tests on Eccentrically Loaded High Strength Concrete Columns. ACI
[22] Kotsikos G. Mechanical property characterisation of WTR laminate. Report submitted to Walker Technical
Resources: NewRail.Centre for Railway Research & School of Mechanical & Systems Engineering, Newcastle
University; 2010.
[24] George JM, Kimiaei M, Elchalakani M. Simplified FE model predicting the bending behaviour of corroded
tubular steel members rehabilitated using CFRP. 5th International Conference on Smart Monitoring, Assessment
and Rehabilitation of Civil Structures (SMAR 2019). Potsdam, Germany: NDT.net; 2020.
58
The University of Western Australia Chapter 3
[25] Kabir MH, Fawzia S, Chan THT, Gamage JCPH, Bai JB. Experimental and numerical investigation of the
behaviour of CFRP strengthened CHS beams subjected to bending. Engineering Structures. 2016;113:160-73.
[26] Hashin Z. Failure Criteria for Unidirectional Fiber Composites. Journal of Applied Mechanics. 1980;47:329-
34.
[27] Hassanein MF, Elchalakani M, Elkawas AA. Design of cold-formed CHS braces for steel roof structures.
59
The University of Western Australia Chapter 3
3.9 Tables
Pipe dimensions Mechanical properties of the steel cross section at mid span Repair details
Eccentricity, No. of No. of
Serial No. Specimen ID Specimen Group L Do t Lc tc Pyield Mys Mps LR
e (mm) I (mm4) Zs (mm3) Zp (mm3) GFRP CFRP
(mm) (mm) (mm) (mm) (mm) (kN) (kNm) (kNm) (mm)
laminates laminates
1 IAE0F0 0
2 IAE20F0 20
3 IAE40F0 Intact 40 1800 114 6.0 0 0.0 3.00 x106 5.25 x104 7.04 x104 704.3 18.1 24.3 - - -
4 IAE60F0 60
5 IAE80F0 80
6 CAE0F0 0
7 CAE20F0 20
8 CAE40F0 Corroded 40 1800 114 6.0 300 1.2 2.32 x106 4.15 x104 5.51 x104 557.2 14.3 19.0 - - -
9 CAE60F0 60
10 CAE80F0 80
11 CAE0F2A 0
12 CAE20F2A 20
13 CAE40F2A Repaired in air 40 1800 114 6.0 300 1.2 2.32 x106 4.15 x104 5.51 x104 557.2 14.3 19.0 1 2 600
14 CAE60F2A 60
15 CAE80F2A 80
16 CAE0F2W 0
17 CAE20F2W 20
18 CAE40F2W Repaired underwater 40 1800 114 6.0 300 1.2 2.32 x106 4.15 x104 5.51 x104 557.2 14.3 19.0 1 2 600
19 CAE60F2W 60
20 CAE80F2W 80
60
The University of Western Australia Chapter 3
61
The University of Western Australia Chapter 3
Ductility
Serial Specimen Method I Method II
Specimen Group
No. ID
Ratio to
ABC (J) ADE (J) DI Ratio to Corroded d1 (mm) d2 (mm) DI
Corroded
1 IAE0F0 1360 13285 9.8 1.05 3.39 44.00 12.0 7.18
2 IAE20F0 576 9430 16.4 1.07 2.63 48.00 17.2 3.27
3 IAE40F0 Intact 434 8612 19.8 1.28 2.42 51.47 20.3 3.76
4 IAE60F0 495 10301 20.8 1.29 3.47 70.59 19.4 2.93
5 IAE80F0 612 11818 19.3 1.08 4.83 86.40 16.9 2.26
6 CAE0F0 645 6020 9.3 1.00 2.35 6.27 1.7 1.00
7 CAE20F0 360 5487 15.2 1.00 2.12 13.27 5.3 1.00
8 CAE40F0 Corroded 430 6656 15.5 1.00 2.90 18.51 5.4 1.00
9 CAE60F0 386 6228 16.1 1.00 2.95 22.50 6.6 1.00
10 CAE80F0 403 7232 17.9 1.00 3.98 33.65 7.5 1.00
11 CAE0F2A 1047 11730 11.2 1.20 2.91 19.92 5.8 3.50
12 CAE20F2A 488 10369 21.3 1.39 2.06 31.53 14.3 2.71
13 CAE40F2A Repaired in air 532 10558 19.8 1.28 2.71 39.62 13.6 2.52
14 CAE60F2A 579 12904 22.3 1.38 3.64 52.41 13.4 2.03
15 CAE80F2A 780 14647 18.8 1.05 5.58 75.05 12.5 1.67
16 CAE0F2W 1262 11161 8.8 0.95 3.50 15.52 3.4 2.06
17 CAE20F2W 667 9755 14.6 0.96 2.72 20.43 6.5 1.24
18 CAE40F2W Repaired underwater 608 10004 16.5 1.06 3.23 24.72 6.7 1.23
19 CAE60F2W 678 10552 15.6 0.96 4.09 30.24 6.4 0.97
20 CAE80F2W 449 7977 17.8 0.99 3.98 52.94 12.3 1.65
62
The University of Western Australia Chapter 3
Strain Values at ultimate load (%) Strain Values at local buckling (%) Maximum Strain Values (%)
Serial Specimen
Specimen Group Face 1 Face 2 Face 1 Face 2 Face 1 Face 2
No. ID
Axial Hoop Axial Hoop Axial Hoop Axial Hoop Axial Hoop Axial Hoop
1 IAE0F0 -0.15 0.04 -0.06 0.04 NA NA NA NA 0.83 -0.76 -0.16 0.88
2 IAE20F0 0.04 -0.01 -0.18 0.06 NA NA NA NA 0.90 -0.68 -1.14 0.91
3 IAE40F0 Intact 0.13 -0.07 -1.00 0.13 NA NA NA NA 1.00 -1.00 -1.00 0.88
4 IAE60F0 0.17 -0.05 -0.14 0.04 NA NA NA NA 0.93 -1.20 -0.33 0.84
5 IAE80F0 0.33 -0.11 -0.15 0.05 NA NA NA NA 0.93 -1.26 -1.19 0.95
6 CAE0F0 -0.13 0.04 -0.17 0.05 0.36 -0.31 NA 0.72 0.94 -1.21 -1.09 0.72
7 CAE20F0 0.09 -0.08 -1.04 0.40 NA -1.00 NA NA 0.90 -1.23 -1.04 0.69
8 CAE40F0 Corroded 0.18 -0.12 -1.13 0.63 NA -0.89 NA NA 0.92 -1.26 -1.13 0.79
9 CAE60F0 0.21 -0.10 -0.18 0.02 NA NA NA NA 0.83 -1.26 -1.18 0.81
10 CAE80F0 0.22 -0.06 -0.18 0.05 NA NA NA NA 0.86 -0.83 -1.18 0.78
11 CAE0F2A -0.11 0.05 -0.49 0.07 NA -0.42 NA NA 1.15 -1.37 -0.65 0.40
12 CAE20F2A 0.15 -0.07 -0.33 0.24 NA -0.69 -0.57 NA 0.78 -0.72 -0.72 0.51
13 CAE40F2A Repaired in air 0.17 -0.06 -1.32 0.35 NA -0.43 NA NA 0.97 -0.62 -1.32 0.35
14 CAE60F2A 0.26 -0.09 -1.00 0.29 NA -0.72 NA NA 1.15 -0.94 -1.89 0.56
15 CAE80F2A 0.23 -0.15 -0.99 0.42 NA -0.61 NA NA 0.87 -1.35 -1.31 0.86
16 CAE0F2W -0.13 0.04 -0.58 0.07 0.70 -0.23 NA NA 0.99 -0.78 -1.06 0.79
17 CAE20F2W 0.11 -0.05 NA 0.21 NA NA NA NA 0.99 -0.33 NA 0.24
18 CAE40F2W Repaired underwater 0.23 -0.10 -0.63 0.20 1.05 -0.49 NA NA 1.20 -0.64 -0.63 0.42
19 CAE60F2W 0.27 -0.14 -0.98 0.21 NA -0.71 NA NA 0.93 -1.26 -0.98 0.41
20 CAE80F2W 0.36 -0.21 -0.73 0.21 NA -1.04 NA NA 1.39 -1.12 -0.83 0.64
Note: -ve sign shows compressive strain
63
The University of Western Australia Chapter 3
Table 3.6 Elastic properties used for the composite repair laminas in the FE model
Layer Type E1 (GPa) E2 (GPa) G12 (GPa) G13 (GPa) G23 (GPa) Nu12
Average 1.00
64
The University of Western Australia Chapter 4
consolidation methods
4.1 Abstract
Offshore structural steel members will need strengthening or repairing due to various reasons.
Strengthening is often required due to work over demands or increased environmental loads. Repairing becomes
essential as the steel members are usually severely corroded due to the marine conditions. Conventional hot work
strengthening and repairing solutions become very expensive and retrofitting using Fibre Reinforced Polymers
(FRPs) is an economical alternative. This paper investigates strengthening and repairing of offshore steel tubular
members using Carbon Fibre Reinforced Polymers (CFRPs) with the help of experimental testing and an analytical
study. Two different specimen sizes which were representative of members in offshore structures namely leg and
brace were examined in the study. Partial length strengthening of steel tubular members subject to combined
loading of axial compression and bending was carried out using different curing environments (in air and
underwater curing). Strengthening and repairing of leg and brace specimens with multiple layers of CFRP
retrofitted in underwater conditions were also investigated. Two different consolidation methods (underwater
vacuum consolidation and underwater stricture consolidation) were compared with the help of full length
strengthening of leg specimens. Although partial length strengthening proved to be less effective, partial length
repairing of corroded tubular members were successful in restoring their intact capacity. Full length strengthening
for underwater leg specimens were experimentally tested and it was found to improve the structural behaviour in
many aspects like ultimate strength, load displacement behaviour, elastic slope, energy absorption, and ductility.
A novel technique of underwater vacuum consolidation of retrofit laminates was demonstrated through a set of
full length strengthened leg specimens. Underwater vacuum consolidation proved to be advantageous compared
to the normal underwater stricture consolidation. An analytical check to identify whether a retrofitted tubular
member would reach its target section properties was also introduced. Overall, the paper demonstrates multiple
successful applications of FRP composites for underwater strengthening and repairing of offshore steel tubular
members. Studies like this expanding the current knowledge on how the FRP retrofitted steel tubular behave within
a real offshore structure are essential to increase the uptake of this technology in the oil and gas industry.
65
The University of Western Australia Chapter 4
4.2 Introduction
4.2.1 General
Offshore structures are subjected to extreme environmental conditions and the need for strengthening or
repairing arise very often during their lifetime. Strengthening becomes necessary when additional load-carrying
ability is required due to work over demands or an increase in environmental loads. Work over demands can be
due to additional equipment loads or change in usage of the structure which was not considered while designing
the structure. Offshore conditions are always changing and the environmental loads like wind, wave and current
loads can possibly become larger than what was anticipated in the original structural design. Repairing becomes
necessary in an offshore structure because of the corrosive marine environment eating up the steel cross-section
Offshore structural members when in need of strengthening or repairing are either replaced with new
members or steel sleeves will be welded or bolted to withstand the loading requirements. These conventional
solutions will create additional dead load to the structure which is disadvantageous. Another drawback of these
solutions is that in case required welding operation on the offshore platform, its normal production due to safety
concerns need to be interrupted creating monetary loss. Direct cost from these traditional repairs or maintenance
works is also quite high. Structural rehabilitation of steel structures using Fibre Reinforced Polymers (FRPs) is a
cost effective and structurally sound alternative for these repairs [1].
Generally, structural steel retrofitting using FRPs are carried out for onshore or in air applications. The
effectiveness of underwater steel structural retrofitting is a big gap in the knowledge. George et al. [1] investigated
underwater repair of steel tubular members under axial compression and bending loads and compared their
behaviour with in air repairs. This study is a continuation of the works presented in that paper. The paper focuses
more on the strengthening application using FRPs for offshore tubular members rather than repairing. On top of
that, two different consolidation techniques for FRP lamination on steel structures have been compared to
Structural retrofitting of steel structures using FRP composites is not a very novel idea. Researchers have
studied different steel structures which had been retrofitted for multiple reasons in the past. The common
parameters investigated were number of retrofit layers, length of FRP wrap, orientation of composite fibres, type
66
The University of Western Australia Chapter 4
As already mentioned, this paper is a continuation of the experimental studies from George et al. [1]. In
that paper, corroded tubular specimens were studied for their structural load bearing behaviour under combined
loading after being repaired in air as well as underwater. Similarly, in the first portion of this paper, the specimens
under combined load which are strengthened in air and underwater were investigated and compared. Strengthening
(i.e., wrapping intact sections with FRP) is studied in this paper as opposed to repairing (i.e., wrapping corroded
sections with FRP). Gao et al. [2] studied the buckling behaviour of steel tubular members strengthened with CFRP
sheets by employing axial compression tests on them. The strength improvement was found to be increasing from
26% to 84% for the strengthened members and a linear relationship from the results was developed between axial
load and number of CFRP layers [2]. The full length strengthening of specimens as in Gao et al. is done for a
number of specimens in the present study too, however the key point of difference is the underwater application
of FRPs and the consolidation techniques used for these specimens. The corrosion simulation (by reducing wall
thickness using machining) in this paper is adapted from the corrosion model in the experimental study on CFRP
The current use of FRPs to strengthen or repair offshore structural elements is very limited. Most of the
FRP wrapping application in offshore engineering industry is for corrosion protection of non-structural elements
in the topside, leakage protection of piping systems, pipelines, risers or conductors. Such research studies have
shown promising results in the past [5-9]. However, the dominant loading on the pipes was mostly internal pressure
which is different to the axial loads and bending moments in typical offshore structural members. Hence a study
with axial loads and bending moments on FRP retrofitted steel tubular members is required. Seica and Parker were
able to conduct bending tests on CHS (circular hollow section) retrofitted specimens with CFRP cured in air and
underwater, where underwater curing produced slightly inferior performance [10]. The combined loading tests of
axial compression and bending on CHS repaired underwater also revealed a small disadvantage for underwater
marine repair using FRP [1]. Overall, underwater retrofitting is showing very promising performance even though
its structural performance is slightly lower than retrofitting done in air. Hence, this technology should be studied
and investigated in detail for more offshore and marine structural applications.
Typical onshore application of FRPs for steel structural retrofitting have been studied by multiple
researchers and the results are very encouraging to adapt these results for offshore applications. Different types of
structural members retrofitted with FRPs undergoing axial loading tests have been investigated and their results
reveal significant improvement in strength and stiffness of the structure due to the FRP retrofitting [11-17].
Similarly, bending tests of CHS retrofitted with FRPs have also revealed good improvement in performance by
67
The University of Western Australia Chapter 4
multiple reserchers [3, 4, 10, 18]. The experimental study in this paper is trying to adopt similar structural
The main aim of the paper was to investigate the feasibility of using FRPs for structural strengthening and
repairing of offshore steel tubular members. The first objective was to experimentally compare in air and
underwater strengthening for tubular members under combined loading. The second objective was to investigate
the effect of number of FRP layers for underwater strengthened and repaired specimens under axial compression
load. The third objective was to compare underwater stricture and vacuum consolidation for full length
strengthened specimens. The final objective was to develop an analytical methodology to check whether retrofitted
tubular members under axial compressive loads would reach their target section properties.
4.3 Methodology
4.3.1 Specimens
The specimens were pipes made of steel of grade API 5L Grade X52. The length of the specimens, L, was
equal to 1800 mm. The Young’s modulus of the pipe material was 200 GPa. Two sets of pipe sections with different
Do/t ratio (outside diameter to wall thickness) representing typical leg and brace members in jacket platforms were
used. The outside diameter, Do, of leg pipes were 114.3 mm with a wall thickness, t, of 6 mm whereas brace pipes
had Do of 60.3 mm and t of 3.9 mm. The length of corroded area, Lc, was fixed as the 3D where D is the nominal
diameter of the pipe. The corresponding nominal diameter, D, of leg and brace pipes were 100 mm 50 mm
respectively. Hence Lc of 300 mm and 150 mm were machined in the mid-span of leg and brace pipes respectively
to get the corroded specimens. The thickness of corrosion, tc, to be machined out was fixed as 20% of the pipe
wall thickness (t) and tc of 1.2 mm and 0.8 mm were used for leg and brace pipes respectively. Strength loss due
to 20% wall loss in steel pipes under bending and bearing was reported to be fully recoverable by CFRP retrofitting
by Elchalakani et al. and hence the same level of corrosion was selected for this study [3, 4]. Table 4.1 summarises
these geometrical properties as well as the mechanical properties of the steel cross-section at the mid-span such as
moment of inertia (I), elastic section modulus (Zs), plastic section modulus (Zp), compressive yield strength (Pyield),
The entire experimental program consisted of 8 different sets of strengthening and repairing offshore
tubular structural members as depicted in Figure 4.1. In this paper, 28 different experimental specimens were
investigated as outlined in Table 4.1. Intact and strengthened specimens were named with ‘I’ in the beginning
68
The University of Western Australia Chapter 4
whereas a corroded and repaired specimens’ name started with ‘C’. Specimens that were repeated with a different
dimension of the pipe started their name with ‘R’ in front of ‘I’ or ‘C’. The axial loading scenario was described
in the second part of the specimen name. A pure axial specimen was named with ‘AE0’ to describe the loading
scenario as axially loaded with 0 mm eccentricity. Similarly, the specimen under axial load with ‘xx’ mm
eccentricity was named ‘AExx’. A bare steel specimen was named with ‘F0’ describing the usage of zero layers
of fibre. Similarly, repaired or strengthened specimens with 2, 4 and 6 layers of CFRP were named with ‘F2’, ‘F4’
or ‘F6’ respectively. The rehabilitation and curing process carried out in air or underwater were distinguished by
adding ‘A’ or ‘W’ at the end of the specimen names respectively. Full length strengthening specimens had two
different consolidation techniques namely stricture and vacuum consolidation which are identified at the very end
of their names as ‘S’ and ‘V’ respectively. Stricture consolidation was utilised for all specimens other than full
length strengthened specimens and therefore it was not mentioned in the specimen names.
4.3.2 Materials
Materials used for strengthening and repair of the specimens were TechnowrapTM products [19].
TechnowrapTM 2K, GFRP was used as the first layer for both strengthening and repairing to avoid direct contact
of CFRP with steel surface to prevent galvanic corrosion [8]. The TechnowrapTM 2K is a triaxial cloth with fibres
running in 0o and ±45o with 800 gsm. This cloth would give an average thickness of 0.8 mm as a lamina.
69
The University of Western Australia Chapter 4
TechnowrapTM SRS (Structural Rehabilitation System) was the CFRP used for preparing the specimens. These are
quad axial cloth with carbon fibres running in 0o, 90o and ±45o with 800 gsm. This CFRP cloth would give an
average thickness of 1 mm per layer as a lamina. The normal resin system, TechnowrapTM HA (High Ambient)
was used for the in air specimens and the subsea resin, TechnowrapTM Splashzone was utilised for the underwater
specimens. The underwater resin contained hydrophobic agents enabling underwater application and curing
without the need for additional curing aids like UV rays or heating mechanisms.
The laboratory retrofitting process for both strengthening and repairing were identical. The steel surface
was prepared with cleanliness of Sa2.5 and a roughness profile of 60 µm (minimum) in accordance with the ISO
guidelines [20]. Surface preparation of both in air and underwater specimen was done using air blasting which is
Once surface preparation was completed, the specimens were applied with multiple layers of CFRP
following the first GFRP layer as required to complete their specific retrofit scheme. All specimens had GFRP as
their first layer to prevent galvanic corrosion. A wet layup technique was utilised for applying the FRP layers on
the steel surface which was done in different stages. In case of the in air specimens, a tether cloth was placed on
top of each stage of FRP layers to develop a rebondable surface for the subsequent layers. Stricture bands were
employed to squeeze out the extra resin and trapped air from the FRP layers and this process is called stricture
consolidation. The strictures were removed after completing soft curing and the next stage of FRP layers were
applied using the same technique as mentioned above. Once all FRP layers in the retrofit scheme is completed,
they underwent a full curing for a day. Underwater specimens were also retrofitted with the same technique except
for the use of tether cloth but the entire process had taken place in a tank filled with saline water from Swan River
in Western Australia.
The length of repair, LR, was calculated as the sum of LC (= 3D), extended 1D after the corroded region at
both ends and appropriate tapering. LR of 600 mm was used for both leg and brace pipes which means 6D and 12D
for them respectively. Figure 4.2 shows how a corroded specimen would transition into a repaired specimen with
the corroded region within the repair. In order to have a basis for comparing the results from the partial length
strengthened specimens to the repaired ones, the length of strengthening, LS, was taken as 6D or 12D for leg or
brace pipes, respectively. Full length strengthened specimens had LS equal to L = 1800 mm.
70
The University of Western Australia Chapter 4
Full length strengthening specimens were prepared using an exactly similar underwater retrofitting process
as mentioned above and the only difference being the FRP layers laid over the full length of the specimens. Half
of the full length strengthening specimens used stricture consolidation like other underwater specimens. The other
Vacuum consolidation uses a vacuum pump to apply pressure on the laminates within the vacuum bag by
sucking the air out and forcing the extra resin used for wetting the laminates out into the flow medium which is
removed after the resin is set. This technique is normally used for in air repairs especially where there is a structure
with a shape or surface (like a flat surface or I or channel section) that will not allow stricture consolidation as
outlined in Section 4.3.3 [21]. Vacuum consolidation is more efficient than stricture consolidation as it reduces the
resin volume fraction in the laminate by eliminating extra resin thereby improving the properties of the laminate
as such. However, this method cannot be directly adopted for underwater specimens.
The main challenge for vacuum bagging underwater specimens is that when the specimens are bagged underwater,
the bag will contain water inside them which cannot be directly connected to a vacuum pump. Instead of the
vacuum pump is directly connected to the vacuum bag, a bypass tank was added to the normal set up this helped
to drain out the water which was trapped in the vacuum bag.
71
The University of Western Australia Chapter 4
Underwater vacuum consolidation was achieved with the following steps as described in Figure 4.3. After
application of the underwater retrofit laminate, a flow medium was placed on top of it to ensure the forcing out of
the extra resin from the laminate into it uniformly. The specimen was bagged using a polythene sheet and this bag
trapped air and water within it. The vacuum pump was connected to this bag and the trapped air and water were
vacuumed out with a pressure between 40 kPa to 60 kPa. The water was collected and drained out using the bypass
tank. The extra resin forced out into the flow medium was removed once the vacuum bag was opened after the
vacuum consolidation was completed. Underwater vacuum bagging was a novel idea which was tried out for the
Figure 4.3 Steps in underwater vacuum consolidation: (a) Application of flow medium, (b) Bagging with
polyethene sheets, (c) Bagged specimen containing air and water inside, (d) Bag connected with a hose, (e) Hose
connected to the vacuum pump via a bypass tank and
Specimen preparation was identical to the specimens presented by George et al. [1]. The exterior CFRP
surface of the specimens was marked out with white markers to easily visualise the local deformation and damage
location in the FRP retrofits. All specimens were gauged with four foil strain gauges on the exterior surface of the
final CFRP layer of the retrofitted specimens and on the steel surface if it is a bare steel specimen. On both faces
of buckling, two strain gauges each were placed at the mid-span, one in the axial direction (along the length of the
pipe) and one in the hoop direction (around the circumference of the pipe).
The testing equipment used for the experiments was an Amsler loading machine in the structural laboratory
at the University of Western Australia. The specimens were fabricated with 10 mm thick end plated welded to
72
The University of Western Australia Chapter 4
them. The concentrically loaded specimens were welded at the centre of the end plate whereas the eccentrically
loaded specimens were welded at an offset equal to the eccentricity from the centre of the endplate. The specimens
were connected to roller plates at the top and bottom to create hinge support conditions for the specimen. The
specimen was later lifted and fitted into the testing equipment and the compressive load was applied and then
gradually increased until the overall failure of the specimen. Figure 4.4 shows the testing set up with a specimen
at different stages of the experiment. Once the ultimate load was reached, the loading continued until the axial
displacement has reached at least 5% of the length of the specimen (L = 1800 mm).
Figure 4.4 Testing set up: (a) Specimen fitted in the loading equipment before testing and (b) Specimen buckling
during the test
The experimental results were analysed to find certain key points as described here. The failure mode of
the specimen was studied in detail with the different failures in the FRP retrofit. The ultimate load was identified
and was used to make a comparison with the control cases. The load displacement behaviour was investigated and
comparative studies were carried out to understand certain patterns. The slope of the linear elastic portion of the
load displacement curve was found as Elastic Slope (ES). Energy Absorption (EA) of a specimen was calculated
73
The University of Western Australia Chapter 4
as the area under the load displacement curve until it reaches 5% strain (which is equal to 90 mm displacement for
specimens with L = 1800 mm length). Ductility Index (DI) was calculated based on an energy based method which
is a modified version of the method by Foster and Attard (1997) [22]. This is exactly same as the “method I”
presented by George et al. [1] and was calculated as given in equation (4.1) below.
DI (4.1)
where ABC is the area under the load displacement curve up to Δ75, the deflection point obtained by
extending the linear elastic part of the curve (0 to 0.75Pu) to peak load (Pu), which is the elastic energy absorbed
by the specimen and ADE is the area under the load displacement curve up to an axial displacement equal to 15Δ75,
which is the total energy absorbed by the specimen. The identification of the five points, A to E is further illustrated
in Figure 4.5. The strain values were plotted against load for comparison and the strain values at ultimate load and
4.4.1 Strengthened leg specimens in air vs. underwater under combined loading
All the leg specimens presented in this section were partial length strengthened with only 2 layers of CFRP
as their retrofit scheme. Combined loading applied on the specimens was a combination of axial compression and
bending moment which was generated through eccentricities in the loading point and the centre of the specimen
namely 0 mm, 20 mm, 40 mm, 60 mm and 80 mm. Figure 4.6 shows the strengthened specimens of both groups,
74
The University of Western Australia Chapter 4
Figure 4.6 Strengthened leg specimens after testing: (a) in air & (b) underwater
The failure location for the partial length strengthened leg specimens was different to that of bare steel and repaired
specimens. Bare steel and repaired specimens showed a symmetrical buckling (i.e., with the point of failure in the
mid span of the specimen), whereas the strengthened specimens showed unsymmetrical buckling as depicted in
Figure 4.7. The point of failure was shifted just outside the retrofitted zone with most of the FRP retrofit remaining
undamaged throughout the testing. The partial length strengthened specimen was looking for the weakest point to
yield and since the mid span region was strengthened it moved to just outside the retrofit zone to facilitate the
yielding.
Figure 4.7 Unsymmetrical buckling in partial length strengthened leg specimens (IAE20F2W)
The ultimate load of the partial length strengthened specimens showed that the in air and underwater
specimens do not have a significant difference between them and with their corresponding intact capacity extracted
from George et al. [1]. Table 4.2 summarises the results of the strengthened specimens in air versus underwater
under combined loading. IAE0F2A is the only specimen that had an ultimate strength smaller than their
75
The University of Western Australia Chapter 4
corresponding intact specimens. The variation in ultimate strength for IAE0F2A was probably due to negligible
variation of the steel cross section between these two specimens. On average, strengthened in air specimens had
5.8% improvement and strengthened underwater specimens had an improvement of 6.1% than intact specimens.
Improvement in the ultimate strength of the partial length strengthened leg specimens compared to the intact ones
was very negligible. For the intact specimens the failure location was in the mid span while for the partial length
strengthened ones it was always on the edge of the retrofit zone, far from the mid span, there was less secondary
moment due to the smaller lateral displacement in unsymmetrical buckling and hence these strengthened
specimens showed slightly higher strength than the corresponding intact ones.
The load displacement behaviour of strengthened specimens under combined loading is presented in Figure 4.8
and it was very similar to that one for the intact specimens governed by yielding at mid-span as presented by
George et al. [1]. The strengthened specimens showed an unsymmetrical buckling governed by yielding but the
point of failure shifting to just outside the retrofit zone. The comparison between in air and underwater specimens
was depicted through these load displacement graphs. There was no distinguishable difference between the
behaviour of in air and underwater specimens. This could be attributed to the fact that failure of the specimens
happened due to unsymmetrical buckling outside the retrofit zone and the behaviour was more dependent on the
Figure 4.8 Load displacement graphs of strengthened leg specimens under combined loading
The elastic slope of the strengthened leg specimens showed mixed results. The elastic slope was expected to
increase from the intact levels due to the added stiffness of the FRP wraps in the strengthened area. However,
some specimens showed a decrease in their elastic slope as shown in Table 4.2. This variation should be a result
76
The University of Western Australia Chapter 4
of differences in the steel material properties and FRP retrofitting differences from specimen to specimen. The
differences in elastic slopes are not significant enough to make a distinction between in air and underwater
specimens as well.
The energy absorbed (EA) by all the strengthened leg specimens had improved from their corresponding intact
levels as presented in Table 4.2. In the case of strengthened in air specimens, the EA improved the maximum for
IAE80F2A by 23.2% and on average 17.6% improvement in EA was observed. In the case of strengthened
underwater specimens, the EA improved the maximum for IAE20F2W by 20.5% and on average 12.5%
improvement in EA was observed. In a similar way to the repaired specimens as presented by George et al. [1],
the strengthened specimens also showed better performance for in air specimens than underwater specimens in
terms of energy absorbed by the specimens. The specimens strengthened underwater showed about 96% on
The Ductility Index (DI) of all strengthened leg specimens increased with respect to their intact DI (DIintact) values.
The ratio of DI to DIintact was also calculated and all these results are tabulated in Table 4.2. The DI values increased
as the eccentricity of loading increased for the strengthened in air specimens. The pattern was the same for
specimens strengthened underwater as well except for IAE80F2W. The average ratio to intact was 1.15 and 1.09
for specimens strengthened in air and underwater respectively. Strengthening in air was producing a better ductility
for specimens under combined loads of axial compression and bending moment than strengthening underwater.
The strain values obtained from the partial length strengthened leg specimens were not very relevant as the strain
gauges were attached at the mid-span further away from the point of failure. However, they gave an insight into
how the specimens behaved and whether there was some participation of the retrofit layers in the behaviour of
these specimens. Strain values at the ultimate load observed by each specimen and the maximum strain values for
the specimens are tabulated in Table 4.3. The maximum ultimate axial strain of 0.54% (compressive) was observed
in IAE60F2W and the maximum ultimate hoop strain of 0.15% (tensile) was observed in IAE80F2A. Considering
the axial strains observed for all of these specimens, the maximum axial strain was observed in IAE20F2W as a
compressive strain of 1.06%. The maximum hoop strain among all the hoop strains observed in these specimens
was a tensile strain of 0.59% in IAE0F2A. Even though these observations were not taken at the point of failure,
they prove that the retrofit layers have participated in the load sharing for these specimens.
77
The University of Western Australia Chapter 4
Specimens with different retrofit schemes were parametrically investigated in this study. A bare steel
specimen with 0 layers of CFRP was presented as the control case in each group. Three different retrofit schemes
were studied and they contained 2, 4 and 6 layers of CFRP in them. The results of specimens IAE0F0, CAE0F0
and CAE0F2W were also added from George et al [1] to have a complete set of data for the parametric study. The
retrofitting process of all the specimens mentioned in this section was carried out underwater. Table 4.4
summarises the key results and Table 4.5 presents the strain values from these experimental specimens.
The partial length strengthened underwater leg specimens had undergone unsymmetrical buckling governed
by yielding of the cross section just outside the retrofit zone as shown in Figure 4.9. The strengthened underwater
specimens showed an ultimate load value very close to the control bare steel specimen, IAE0F0 as tabulated in
Table 4.4. No improvement in the ultimate strength was observed for these specimens and hence it could be
concluded that this partial length strengthening configuration was not useful for structural members where the
ultimate strength was eventually dependent on the yielding capacity of the steel cross section. The number of
retrofit (CFRP) layers also did not have any impact on the ultimate load.
78
The University of Western Australia Chapter 4
The load displacement behaviour of the specimens strengthened underwater as illustrated in Figure 4.10
reveals that all the strengthened specimens showed consistent behaviour which was slightly different to that of
IAE0F0. This difference was due to the symmetric buckling in IAE0F0 and unsymmetric buckling in the
strengthened specimens. The axial load capacity of the strengthened underwater leg specimens was better until a
79
The University of Western Australia Chapter 4
The results of the partial length strengthened underwater leg specimens for their ultimate strength, elastic
slope, energy absorption and ductility index are tabulated in Table 4.4. Elastic slopes of the strengthened
underwater specimens are all higher than that of the bare steel specimen. The elastic slope increased from
IAE0F2W to IAE0F4W, but IAE0F6W showed the least among the three strengthened specimens. This
inconsistency probably could be due to an error in the experimental set up generating a slight eccentricity in the
loading for this specimen. The energy absorption of these specimens remained almost consistent among them at
an average of 5.9% improvement from intact level. The ductility of the strengthened underwater specimens
increased from the intact level and the DI increased as the number of retrofit layers increased. Even though this
strengthening did not serve the original purpose of improving the specimen strength, parameters like load
displacement behaviour, elastic slope, EA and DI had improved as a result of strengthening. The strain at ultimate
load and maximum strains for these specimens are tabulated in Table 4.5. Since the strains were recorded away
from the point of failure for the strengthened specimens, they were much lower than what is recorded for the intact
specimen.
The idea of brace specimens, with a different dimension compared to leg specimens, was to study the
effectiveness of the strengthening or repairing with a different mode of buckling. The failure of leg specimens was
governed by yielding of steel cross section but the brace specimens were designed to have Euler buckling. The
critical load for Euler buckling, Pcr calculated for a brace specimen was 168.5 kN and its yield load, Pyield was 239
kN. As Pcr is notably lower than Pyield for the brace specimens, their failure was dominated by the Euler buckling.
80
The University of Western Australia Chapter 4
The intact brace specimen, RIAE0F0 had an ultimate load of 137.74 kN which was close enough to Pcr confirming
The partial length strengthened underwater brace specimens all showed unsymmetrical buckling with the
point of failure right outside the retrofit zone as shown in Figure 4.11. The ultimate load of the strengthened
underwater specimens improved except for RIAE0F2W. The ultimate load of RIAE0F4W and RIAE0F6W was
almost identical with an average improvement of 19.1% in ultimate load from intact level. There was no variation
in ultimate load as the retrofit layers were increased. The strengthening of brace specimens seemed to be slightly
The load displacement behaviour of the strengthened underwater brace specimens is presented in Figure
4.12. RIAE0F2W showed a remarkable difference in the load displacement behaviour especially in the elastic
slope indicating that potentially this specimen was unexpectedly loaded with a large eccentricity. Due to this
potential experimental error, RIAE0F2W was not utilised for drawing any conclusions in this section. In
comparison with RIAE0F0, both RIAE0F4W and RIAE0F6W showed significantly better structural behaviour as
81
The University of Western Australia Chapter 4
The test results for the strengthened underwater brace specimens are tabulated in Table 4.4. Elastic slopes
of the strengthened underwater brace specimens were better than that of the bare steel specimen for RIAE0F4W
and RIAE0F6W. The energy absorption of all these specimens was better than intact level even RIAE0F2W. On
average 52.5% improvement in energy absorption was observed for the strengthened brace specimens. The
ductility of the strengthened underwater brace specimens also increased from the intact level. Overall,
strengthening of these brace specimens showed a better result than strengthening the leg specimens. The strain at
ultimate load and maximum strains for these brace specimens are tabulated in Table 4.5. Even though strain values
in the strengthened specimens were recorded away from the point of failure, they were in the similar range of
strains observed in the intact repeat specimen proving that the retrofit layers had participated in the load sharing.
Specimens repaired underwater showed failure modes all over the spectrum. The results of CAE0F2W were
already discussed in George et al. [1]. This specimen showed symmetrical buckling in the middle of the retrofit
zone within the corrosion region. The ultimate load was 29.4% better than CAE0F0 but less than intact strength.
When the number of retrofit layers was increased to 4 as for CAE0F4W, it showed an unsymmetrical buckling
with the point of failure outside the corrosion region but within the retrofit zone. This one showed the maximum
ultimate strength as it utilised both the full steel thickness as well as all retrofit layers at its point of failure.
CAE0F4W showed 40.2% improvement in ultimate load than CAE0F0 and slightly better than intact strength.
82
The University of Western Australia Chapter 4
CAE0F6W showed a typical unsymmetrical buckling with the point of failure outside the retrofit zone like the
partial length strengthened specimens. CAE0F6W showed the least ultimate strength among these three specimens
but still was 24.4% better than CAE0F0. One major inference from these specimens was that more retrofit layers
will not always translate into an improved ultimate strength of a tubular member. The point of failure of the
retrofitted specimen also had a significant influence on their behaviour. The failure points observed after testing
and load displacement behaviour for these specimens repaired underwater are shown in Figure 4.13 and Figure
4.14 respectively. All of them show better behaviour than CAE0F0. CAE0F4W showed the best behaviour and
CAE0F6W the worst among the repaired underwater specimens as expected based on the different modes of failure
involved.
83
The University of Western Australia Chapter 4
All key observations from the repaired underwater leg specimens are tabulated in Table 4.4. Elastic slope
of CAE0F2W was lower than that of CAE0F0 indicating slight eccentricity in the test set up for this specimen.
The elastic slope for CAE0F4W and CAE0F6W were better than the control case and the slope increased as the
retrofit layers increased showing additional retrofit layers increase the stiffness of the member. The energy
absorption of all these specimens were better than the intact level with the best EA observed for CAE0F4W at
124.8% of CAE0F0. DI of the repaired underwater specimens also increased from the intact level except for
CAE0F2W which was only 0.95 times the DI of CAE0F0. The ratio of DI for CAE0F4W and CAE0F6W to that
of CAE0F0 were 2.75 and 1.28 respectively. CAE0F4W showed the best results for all these parameters except
The strain values of CAE0F0 and CAE0F2W were presented and discussed by George et al. [1]. The strain values
observed for CAE0F4W and CAE0F6W were at a slight disadvantage as they are measured away from their points
of failure. Hence a direct comparison of the strain values was not feasible. Strain values at the ultimate load and
maximum strain values for the underwater repaired leg specimens are also tabulated in Table 4.5.
Brace specimens repaired underwater had unsymmetrical buckling with their point of failure right outside
the retrofit zone as shown in Figure 4.15. The ultimate of these specimens improved on average 20.6% from the
control case RCAE0F0 and all of them had better strength than the intact specimen too. Even though the point of
84
The University of Western Australia Chapter 4
failure was outside for all 3 specimens, they showed a slight improvement in strength as the number of layers
increased. The improvement was not a direct influence of the improving strength of the retrofit layers rather it is
an outcome of increasing stiffness of retrofit layers as Euler buckling was observed in them.
The load displacement behaviour of the repaired underwater brace specimens as shown in Figure 4.16
revealed something different. The behaviour declined in the post ultimate load portion as the number of retrofit
layers increased. The behaviour of RCAE0F6W dropped even below the bare steel control level. The exact reason
cannot be pinpointed but lesser involvement of the retrofit layers in the post ultimate load behaviour as the retrofit
zone became stiffer was evident. More detailed study focusing on retrofitting of Euler buckling specimens are
85
The University of Western Australia Chapter 4
The elastic slopes increased as the number of retrofit layers increased in these specimens showing better
stiffness with more retrofit layers. The energy absorption declined as the retrofit layers increased with the least
energy absorbed by RCAE0F6W even lower than RCAE0F0. In the case of Ductility Index, RCAE0F4W showed
the best ratio to the corroded case at 1.87. All these results are tabulated with more details in Table 4.4. Strain
values measured for these specimens were taken at a location away from the point of failure and hence irrelevant.
Strain values at ultimate load and maximum strain values observed in specimens repaired underwater (repeat) are
The leg specimens strengthened for their full length were also experimentally tested in this study. All the
specimens discussed in this section had their retrofit layers consolidated using stricture banding like all specimens
discussed earlier. The full potential of strengthening was demonstrated through these specimens. The specimens
buckled with the point of failure at mid-span or very close to it as shown in Figure 4.17 and hence it could be
concluded that symmetrical buckling was observed in these specimens just like in the bare steel control specimens.
CFRP layers were crushed on the compression face and CFRP layers were ruptured on the tension face of the
specimens as presented in Figure 4.17. A combined load sharing between the steel pipe and retrofit layers was
demonstrated through these specimens. The ultimate load of all leg specimens strengthened full length underwater
(stricture) were better than the control case IAE0F0 and they progressively increased as the retrofit layers
86
The University of Western Australia Chapter 4
increased. The maximum ultimate load of 902.13 kN was observed in IAE0F6WS which is 53.8% better than the
intact specimen. On average 35.2% improvement in ultimate strength was observed for these specimens
Figure 4.17 Full length strengthened underwater (stricture) leg specimens after testing
The load displacement behaviour shown in Figure 4.18 demonstrated better performance for all the
strengthened leg specimens from the intact level. The post ultimate load portion of the curves have minute jerks
and steps indicating the retrofit layers failing in different modes. The smooth behaviour of IAE0F0 showed how
differently yielding of pure steel member occurs as opposed to a retrofitted member. During load testing
experiments, large cracking sounds were heard corresponding to these jerks observed in the load displacement
behaviour indicating matrix cracking and/or fibre rupture in the retrofit laminates. It was also observed the jerks
in the unloading part of the load displacement curve became more pronounced with increasing number of retrofit
layers. As the unloading progressed, all the full length strengthened leg specimens gradually dropped down to
reach the intact level i.e., after axial displacement of 80 mm all the retrofitted specimens showed almost the same
behaviour as the intact one which indicated full utilisation of retrofit layers by this point.
87
The University of Western Australia Chapter 4
Figure 4.18 Load displacement behaviour of full length strengthened underwater (stricture) leg specimens
As presented in Table 4.4, elastic slopes of the full length strengthened (stricture) leg specimens were 30.1%
better than intact on average. However, they showed a decreasing trend as the number of retrofit layers increased.
Energy Absorption of these specimens increased by 62.3% on average from the intact level with EA increasing as
the retrofit layers going up. The mean DI ratio for these strengthened specimens to intact was 1.27 and the best DI
of 13.3 was observed for IAE0F2WS. All these parameters for the strengthened full length underwater (stricture)
Strain values of leg specimens strengthened full length underwater (stricture) were exhibiting the same
trend as the intact specimen as shown in Figure 4.19. The axial strain values were slightly different than a typical
steel specimen, but the nature of the strains remained the same as intact. Strains at ultimate load and maximum
strains for these specimens are tabulated in Table 4.5. Axial strain and hoop strain at ultimate load were observed
the largest in IAE0F6WS on its compression face at 0.28% (compressive) and 0.12% (tensile) respectively. The
maximum axial strain observed among these specimens was 1.66% (tensile) in IAE0F4WS. The maximum hoop
strain for these specimens was recorded for IAE0F6WS as 0.39% (compressive). The minute jerks observed in the
load displacement curves are also reflected in the strain value plots confirming the presence of microscopic failures
with the retrofit laminates. Overall, the strain values confirmed the fact that each retrofitted specimen would have
a slightly different set of microscopic failures involved than other retrofitted specimens. Even though there was a
88
The University of Western Australia Chapter 4
general trend in the behaviour of these retrofitted specimens, the prediction of the exact location and magnitude of
Figure 4.19 Strain values of full length strengthened underwater (stricture) leg specimens
A different technique for the application of FRP layers on a steel tubular member using vacuum
consolidation was studied here. Full length strengthening of leg specimens was carried out with vacuum
consolidation and the results were compared with normal stricture consolidation. This process provided retrofit
laminates with better surface finish and smaller thickness per FRP layer as the extra resin was eliminated more
efficiently.
89
The University of Western Australia Chapter 4
The full length strengthened underwater (vacuum) leg specimens all had symmetrical buckling with the
point of failure at mid-span or adjacent to it as shown in Figure 4.20. As observed in the stricture consolidated
specimens, they had FRP crushing on their compression face and FRP rupture on their tension face. Due to an
error during the preparation of IAE0F6WV, a small amount of water unexpectedly got trapped between laminates
and it was expelled out of the retrofit once the compression testing had commenced. The structural performance
Figure 4.20 Full length strengthened underwater (vacuum) leg specimens after testing
The ultimate load of all full length strengthened underwater (vacuum) leg specimens was higher than the
intact capacity and on average 41.3% better than intact specimen which was 6.1% better than stricture consolidated
specimens. The ultimate load carried by these specimens also improved as the retrofit layers increased. All vacuum
consolidated specimens had considerably better ultimate load than their stricture consolidated counterparts except
for IAE0F6WV. Even though IAE0F6W had an error during preparation, with obvious adverse effect on overall
load bearing capacity of the specimen, it recorded the largest ultimate load of 879.19 kN among the vacuum
consolidated specimens.
The load displacement behaviour as shown in Figure 4.21 demonstrated better performance for all the
strengthened specimens from the intact level and progressive improvement as the retrofit layers increased. The
comparison of the load displacement behaviour between stricture consolidated and vacuum consolidated
specimens is presented in Figure 4.22. The behaviour was better for IAE0F2WV and mostly better for IAE0F4WV
as opposed to their stricture consolidated counterparts. In the unloading part of IAE0F4W, the stricture
90
The University of Western Australia Chapter 4
consolidated specimen performed slightly better than the vacuum bagging. The behaviour of IAE0F6WV, up to
the ultimate load and then beginning of the unloading part, was obviously inferior to IAE0F6WS, but due to the
observed error in preparation of IAE0F6WV no inferences should be drawn. The minute jerks in the load
displacement behaviour corresponding to matrix failures are slightly smoother for vacuum consolidated
specimens. Hence, vacuum consolidation reduced the intensity of microscopic failures in the retrofit laminates
which is an advantageous feature. In general, the load displacement behaviour was better for vacuum consolidated
specimens.
Figure 4.21 Load displacement behaviour of full length strengthened underwater (vacuum) leg specimens
91
The University of Western Australia Chapter 4
Figure 4.22 Comparison of load displacement behaviour of leg specimens strengthened full length using stricture
Refer to Table 4.4, elastic slope of the leg specimens had improved from the intact level but there is no
clear pattern of change for the elastic slope with increasing retrofit layers. In the case of vacuum consolidation,
the thickness of the CFRP laminates decreased but stiffness and strength of the retrofit were supposed to improve.
92
The University of Western Australia Chapter 4
The better ultimate loads of vacuum specimens from stricture consolidated indicate that the loss in thickness was
more than compensated by the improvement in strength. However, in the case of elastic slope this was not accurate
as confirmed by lower elastic slopes of vacuum consolidated specimens than stricture consolidated specimens.
Elastic slope of specimens is a function of their axial stiffness (EA) and bending stiffness (EI). Hence it could be
inferred that vacuum consolidated specimens had lower elastic slope than stricture consolidated specimens due to
The energy absorption of the specimens also improved from intact level and had an ascending pattern as
the retrofit layers increased. IAE0F4WV had slightly lower energy absorption than the general trend due to the
sudden drop in the unloading portion of the load displacement behaviour. Mean improvement of 64.3% in EA
was observed in the vacuum consolidated specimens with 2% better results than stricture consolidated specimens.
DI values of these specimens were greater than intact DI. The best DI ratio of 1.30 was recorded for IAE0F6WV
Strain values of leg specimens strengthened full length underwater (vacuum) were showing identical pattern
as the stricture consolidated specimens as shown in Figure 4.23. The axial strain values on the compression face
of IAE0F6WV was slightly different from the rest of the axial strains. Strains at ultimate load and maximum strains
for these specimens are tabulated in Table 4.5. Axial strain and hoop strain at ultimate load were largest on the
compression face of IAE0F4WV at 0.25% (compressive) and on the compression face of IAE0F6WV at 0.09%
(tensile) respectively. The maximum axial strain observed among these specimens was 1.74% (tensile) in
IAE0F4WV. The maximum hoop strain for these specimens was recorded for IAE0F6WS as 1.47% (tensile) on
93
The University of Western Australia Chapter 4
Figure 4.23 Strain values of full length strengthened underwater (vacuum) leg specimens
Overall, vacuum consolidation proved to be more efficient in terms of ultimate load, load displacement
behaviour and energy absorption than stricture consolidated specimens. Ductility index and elastic slope were
slightly better for stricture consolidated specimens. Underwater vacuum consolidation being a more complex
procedure than stricture consolidation, the potential for fabrication errors as in IAE0F6WV is high. However, the
benefits outweigh the drawbacks for vacuum consolidation technique. Perfecting this novel technique will be
highly beneficial for underwater repairing and strengthening of steel tubular members in the future.
The main purpose of this analytical check was to determine whether the repaired or strengthened section
capacity could reach the target section properties. The check examined approximately how many layers of retrofit
94
The University of Western Australia Chapter 4
was required to repair a corroded member back to its original performance or strengthen a member to a certain
target level. Another important aim of this analytical check was to predict whether the point of failure will occur
within the retrofit zone or not. Different failure locations had been observed in the experimental specimens and it
would be very intuitive if an analytical check could predict it, even approximately, before occurring especially for
A typical cross section of a retrofitted tubular member could be schematically represented as in Figure 4.24.
The thickness of steel is based on the type of specimen (leg or brace) and corrosion level. Based on the data
provided by the manufacturers, the thickness of every layer of GFRP and CFRP, namely tg and tc, were taken as
0.8 mm 1 mm respectively. They represent stricture consolidation thickness and hence vacuum consolidated
specimens would not be represented by these analytical calculations. The area of the cross section and moment of
inertia are calculated based on the geometry of the cross section. A, Ag and Ac are the area of cross section of steel
pipe, GFRP layer and CFRP layers respectively. Similarly, I, Ig and Ic are the moment of inertia of steel pipe,
The properties of the retrofit layers were derived based on the mechanical properties of laminates provided
by TechnowrapTM as in Table 4.6 [23]. The modulus of elasticity of CFRP layers, Ec was considered as 36 GPa.
The tensile strength of CFRP laminate was calculated as 511 MPa in the fibre direction [23]. However, the
experiments were looking into axial compression of the specimens and the Finite Element Model for the repaired
specimens by George et al. [1] revealed that 135 MPa was more suitable for modelling the strength for compression
loading. Hence the strength of CFRP layers, σc was taken as 135 MPa which was approximately a quarter of the
tensile strength reported in the manufacturer’s data. Since the GFRP layers were applied with their transverse
direction along the length of the pipe, elastic modulus of the GFRP layer was taken as 8 GPa. Like in the CFRP
layers, the compressive strength was taken as a quarter of their tensile strength (240MPa) for GFRP layer too. So,
95
The University of Western Australia Chapter 4
the strength of GFRP layer, σg was taken as 60 MPa. The steel properties were taken as elastic modulus, E of 200
The basic idea utilised for this analytical check was that strength and stiffness of the retrofitted cross section
should be at least equal to the non-corroded steel cross section for repairing cases or to the equivalent target steel
cross section for strengthening cases. Hence, strength gain, α and stiffness gain, β of a retrofitted tubular member
were simply defined using Equations (2) and (3) respectively as given below.
α (4.2)
( )
The ratio of the yield strength of retrofitted tubular cross section to that of the target tubular steel cross
section, (σA)target was taken to calculate strength gain. The yield strength calculated for the retrofitted cross section
will be higher than the ultimate load of the retrofitted member due to member slenderness reduction effects [24].As
the target member and the retrofitted member would have similar member slenderness reduction factors, the
strength gain should provide a reasonable estimate of the ultimate strength of a retrofitted member. If an
unsymmetrical buckling with the point of failure outside the retrofit zone occurs, the strength gain would not be
able to predict the ultimate strength of the retrofitted tubular member. The ratio of the bending stiffness of
retrofitted tubular cross section to that of the target tubular steel cross section, (EI)target, was taken to calculate
stiffness gain as defined in Equation (3). If the stiffness gain is found to be greater than 1 it means that the retrofitted
member has higher bending stiffness than the target level and more than likely the point of failure would be
β (4.3)
( )
Strength and stiffness gains calculated based on the above discussion for the different strengthening and
repair cases presented in the paper are tabulated with the details of their calculation in Table 4.7. The target level
for strengthening cases was taken as intact steel cross section which meant all gains were greater than 1 for these
cases, but still useful inferences were drawn from them. The target for repaired corroded members was also taken
In the case of strengthened leg pipes, the retrofit layers of 2, 4 and 6 layers produced strength gains of 1.17,
1.31 and 1.47 respectively which were very similar to the 19.9, 31.8% and 53.8% improvement from the intact
capacity for the full length strengthened underwater (stricture) leg specimens as outlined in Table 4.4. Hence these
strength gains were able to provide a good analytical estimate of the corresponding ultimate member strength.
96
The University of Western Australia Chapter 4
The partial length strengthening of leg and brace pipes presented in the paper would not become applicable
under this analytical check as they had unsymmetrical buckling with point of buckling outside the retrofit zone.
The strength gain defined here would not be applicable to strengthened brace specimens governed by Euler
buckling too. More extensive studies are required to analytically describe effects of retrofitting on such specimens.
The best set of inference from this analytical study was obtained for the repaired underwater leg specimens.
CAE0F2W had symmetrical buckling with the buckling within the retrofit zone which agreed with fact that the
strength and stiffness gains were less than one. The strength gain of 0.95 for this case reasonably predicted with
the ultimate load of 542.88 kN for CAE0F2W which was 92.5% of the ultimate strength of the intact control
specimen (IAE0F0), see Table 4.4. The stiffness gain of 0.85, lower than target level might have influenced the
ultimate load to be slightly inferior for both these specimens. CAE0F4W had a strength gain of 1.10 but its stiffness
gain was only 0.93. This specimen showed a failure mode which was unique with unsymmetrical buckling with
the point of failure just within the retrofit zone. The strength gain calculated would not be able to precisely predict
the ultimate member strength as unsymmetrical buckling was observed. The ultimate load for CAE0F4W was only
0.25% greater than the intact capacity. As the stiffness gain of 0.93 for CAE0F4W was about to reach the target
level, the point of failure had started to shift outside the retrofit zone. CAE0F6W had both strength and stiffness
gains greater than one and this specimen showed a clear unsymmetrical buckling with the point of failure outside
the retrofit zone. This was an ideal scenario for a repaired specimen where the analytical check was completely
satisfied and the failure point moved outside the retrofit zone.
Similarly, the repaired underwater brace specimens were investigated based on their strength and stiffness
gains. RCAE0F2W had a strength gain of 1.04 and a stiffness gain of 0.89. Even though the stiffness gain was
lower than 1 for RCAE0F2W, an unsymmetrical buckling with the failure point outside the retrofit zone was
observed. A possible explanation for this out of trend behaviour was that stiffness gain being close enough to the
target level for RCAE0F2W and behaviour like CAE0F4W occurred for this specimen too but was not quite
distinguishable due to the small dimension of brace pipes. RCAE0F4W and RCAE0F6W had their strength and
stiffness gains greater than one and they both showed unsymmetrical buckling during the experiment with the
point of failure outside the retrofit zone as expected in the analytical check.
4.6 Conclusion
A total number of 28 specimens, comprising 8 different sets of strengthening and repairing offshore tubular
structural members, were investigated in this paper. Strengthening in air and underwater were compared using
97
The University of Western Australia Chapter 4
experimental specimens under axial compression loads and bending moments to understand the adaptability of
FRP retrofitting in offshore or marine conditions. Effect of the number of retrofit layers was investigated for partial
length underwater strengthening and underwater repairing for both leg and brace size specimens under pure axial
compression loads. Full length underwater strengthened leg size specimens with two different consolidation
techniques namely stricture and vacuum were also tested and examined under axial compression loads. Finally, an
analytical check for retrofitted tubular members under axial compressive loads was developed. Inferences drawn
in the paper are dependent on specimen sizes and composite material properties used in this study and careful
consideration should be taken in generalising them. The key conclusions drawn based on the experimental and
Partial length strengthening of leg specimens did not produce significant improvement in strength and
they all had unsymmetrical buckling with their point of failure just outside the retrofit zone.
The performance of in air and underwater strengthened specimens were non distinguishable and it
happened because the point of failure for all these specimens was outside the retrofit zone.
Partial length strengthening of brace specimens showed a mean strengthening improvement of 19.1%
compared to intact specimens. Brace specimens which were governed by Euler buckling responded
Underwater repairing produced the best ultimate strength for the leg specimen with 4 layers of CFRP
at 40.2% improvement which was more than the intact capacity. When the repair was able to match
both the strength and stiffness of lost steel due to corrosion (in the case of CAE0F6W), unsymmetrical
buckling occurred taking the point of failure just outside the retrofit zone. The performance of the
Underwater repairing of brace specimens showed significant improvement in performance where all
the specimens recorded an unsymmetrical buckling with point of failure outside the retrofit zone
indicating the repair was successful in mitigating the corrosion. The maximum ultimate load was
recorded in RCAE0F6W which represented 54.9% improvement. The performance improved as the
Full length strengthened underwater (stricture) leg specimens had shown significant improvement in
their strength and other parameters from the intact levels. The improvement in ultimate load increased
98
The University of Western Australia Chapter 4
by 19.1%, 31.8% and 53.8% for 2, 4 and 6 layers of CFRP retrofitting respectively. They all exhibited
specimens. For the tested leg specimens, the full potential of the strengthening was only achievable
through full length strengthening although it might become an expensive solution than partial length
strengthening.
Full length strengthened underwater (vacuum) leg specimens also showed improved performance from
the intact level and mostly had better results than stricture consolidated specimens. The ultimate load
improvements for 2, 4 and 6 layers of CFRP in these strengthened specimens were 34.1%, 40% and
49.9%, respectively. The novel underwater vacuum consolidation technique was demonstrated
An analytical check for retrofitted tubular members undergoing axial compressive loads was presented.
This method was able to successfully predict the behaviour of the experimental specimens.
Overall, the paper demonstrated that strengthening and repairing of offshore tubular structural members in
marine environmental conditions, in air or underwater, could be achieved with the help of FRP composites. Future
studies should try to expand the potential of this technology for different loading scenarios and wider range of
section shapes and sizes, investigate reliability of the achieved results and evaluate the long term performance of
4.7 Acknowledgements
The authors would like to thank ICR.IAS Joint Venture for providing composite materials, manufacturer
data and expert labour for the preparation of the retrofitted specimens presented in the paper. The authors would
also like to acknowledge the efforts of Master of Professional Engineering (MPE) students, Bowei Sun, Dong
Hyun Kim, Vignesh Sairam Giridhar, Liam McGurdy, Michael Mason, Anjali Gayathri Konara and Kit Strickland
in the conduct of the experiments through RiverLab research fund at the University of Western Australia
(www.uwa.edu.au/projects/oceanworks-project-pages/riverlab).
4.8 References
[1] George JM, Kimiaei M, Elchalakani M, Fawzia S. Experimental and numerical investigation of underwater
composite repair with fibre reinforced polymers in corroded tubular offshore structural members under concentric
99
The University of Western Australia Chapter 4
[2] Gao XY, Balendra T, Koh CG. Buckling strength of slender circular tubular steel braces strengthened by CFRP.
[3] Elchalakani M. Rehabilitation of corroded steel CHS under combined bending and bearing using CFRP.
[4] Elchalakani M, Karrech A, Basarir H, Hassanein MF, Fawzia S. CFRP strengthening and rehabilitation of
[5] Alexander C, Cercone L, Lockwood J. Development of a Carbon-Fiber Composite Repair System for Offshore
Risers. ASME 2008 27th International Conference on Offshore Mechanics and Arctic Engineering 2008. p. 389-
405.
[6] Shouman A, Taheri F. An Investigation Into the Behaviour of Composite Repaired Pipelines Under Combined
Internal Pressure and Bending. ASME 2009 28th International Conference on Ocean, Offshore and Arctic
[7] Shouman A, Taheri F. Compressive strain limits of composite repaired pipelines under combined loading
[8] Shamsuddoha M, Islam MM, Aravinthan T, Manalo A, Lau K-t. Effectiveness of using fibre-reinforced
polymer composites for underwater steel pipeline repairs. Composite Structures. 2013;100:40-54.
[9] Saeed N. Composite Overwrap Repair System for Pipelines - Onshore and Offshore Application Queensland:
[10] Seica MV, Packer JA. FRP materials for the rehabilitation of tubular steel structures, for underwater
[11] Bambach MR, Elchalakani M. Plastic mechanism analysis of steel SHS strengthened with CFRP under large
[12] Fawzia S, Al-Mahaidi R, Zhao XL, Rizkalla S. Strengthening of circular hollow steel tubular sections using
[13] Bambach MR, Elchalakani M, Zhao XL. Composite steel–CFRP SHS tubes under axial impact. Composite
Structures. 2009;87:282-92.
100
The University of Western Australia Chapter 4
[14] Bambach MR, Jama HH, Elchalakani M. Axial capacity and design of thin-walled steel SHS strengthened
[15] Elchalakani M, Fernando D. Plastic mechanism analysis of unstiffened steel I-section beams strengthened
[16] Haedir J, Zhao X-L. Design of short CFRP-reinforced steel tubular columns. Journal of Constructional Steel
Research. 2011;67:497-509.
[17] Elchalakani M, Karrech A, Basarir H, Zhao X-L, Fawzia S, Hassanein MF. Strengthening of mild steel struts
using CFRP sheets subjected to uniform axial compression. Thin-Walled Structures. 2017;116:96-112.
[18] Kabir MH, Fawzia S, Chan THT. Effects of layer orientation of CFRP strengthened hollow steel members.
Građevinar. 2015;67:441-51.
[20] ISO. Petroleum, petrochemical and natural gas industries - Composite repairs for pipework - Qualification
[21] George JM, Kimiaei M, Elchalakani M. Experimental Study on Structural Rehabilitation of Severely
Damaged I-Beams Using Fibre Reinforced Polymers. Proceedings of the ASME 2021 40th International
[22] Stephen JF, Mario MA. Experimental Tests on Eccentrically Loaded High Strength Concrete Columns. ACI
[23] Kotsikos G. Mechanical property characterisation of WTR laminate. Report submitted to Walker Technical
Resources: NewRail.Centre for Railway Research & School of Mechanical & Systems Engineering, Newcastle
University; 2010.
[24] Zhao X-L, Wilkinson T, Hancock G. Chapter 4 - Members Subjected to Compression. In: Zhao X-L,
Wilkinson T, Hancock G, editors. Cold-Formed Tubular Members and Connections. Oxford: Elsevier; 2005. p.
67-89.
101
The University of Western Australia Chapter 4
4.9 Tables
Pipe dimensions Mechanical properties of the steel cross section at mid span Retrofit details
1 IAE0F2A 0
2 IAE20F2A 20
Partial length
3 IAE40F2A Leg In air Stricture 40 1800 114.3 6.0 0 0.0 3.00E+06 5.25E+04 7.04E+04 704.3 18.1 24.3 1 2 600
Strengthening
4 IAE60F2A 60
5 IAE80F2A 80
6 IAE0F2W 0
7 IAE20F2W 20
8 IAE40F2W 40 1 2
Partial length
9 IAE60F2W Leg Underwater Stricture 60 1800 114.3 6.0 0 0.0 3.00E+06 5.25E+04 7.04E+04 704.3 18.1 24.3 600
Strengthening
10 IAE80F2W 80
11 IAE0F4W 0 1 4
12 IAE0F6W 0 1 6
13 RIAE0F0 - - -
16 RIAE0F6W 1 6
17 CAE0F4W 4
Leg Repairing Underwater Stricture 0 1800 114.3 6.0 300 1.2 2.32E+06 4.15E+04 5.51E+04 557.2 14.3 19.0 1 600
18 CAE0F6W 6
19 RCAE0F0 - - -
20 RCAE0F2W 1 2
Brace Repairing Underwater Stricture 0 1800 60.3 3.9 150 0.8 2.10E+05 7.17E+03 9.62E+03 187.4 2.5 3.3
21 RCAE0F4W 1 4 600
22 RCAE0F6W 1 6
23 IAE0F2WS 1 2
Full length
24 IAE0F4WS Leg Underwater Stricture 0 1800 114.3 6.0 0 0.0 3.00E+06 5.25E+04 7.04E+04 704.3 18.1 24.3 1 4 1800
Strengthening
25 IAE0F6WS 1 6
26 IAE0F2WV 1 2
Full length
27 IAE0F4WV Leg Underwater Vacuum 0 1800 114.3 6.0 0 0.0 3.00E+06 5.25E+04 7.04E+04 704.3 18.1 24.3 1 4 1800
Strengthening
28 IAE0F6WV 1 6
102
The University of Western Australia Chapter 4
Table 4.2 Results of Strengthened leg specimens in air vs. underwater under combined loading
103
The University of Western Australia Chapter 4
Table 4.3 Strain values of Strengthened leg specimens in air vs. underwater under combined loading
104
The University of Western Australia Chapter 4
105
The University of Western Australia Chapter 4
Table 4.5 Strain values of specimens with different number of retrofit layers
106
The University of Western Australia Chapter 4
107
The University of Western Australia Chapter 4
Ac, Area of CFRP layers (mm ) 2 741 1507 2298 726 1477 2253 401 828 1280 391 808 1250
Ic, Moment of Inertia of CFRP layers 128752 271056 427831 121050 255115 403095 20512 45120 74335 19010 41920 69230
(mm4) 7 9 9 3 3 8 5 4 9 8 9 0
Pc, Yield strength of GFRP layer (kN) 100 203 310 98 199 304 54 112 173 53 109 169
EcIc, EI of CFRP layers (kNm2) 0.0464 0.0976 0.1540 0.0436 0.0918 0.1451 0.0074 0.0162 0.0268 0.0068 0.0151 0.0249
α, Strength Gain 1.17 1.31 1.47 0.95 1.10 1.25 1.27 1.51 1.76 1.04 1.28 1.53
β, Stiffness Gain 1.08 1.17 1.26 0.85 0.93 1.02 1.14 1.30 1.49 0.89 1.04 1.22
108
The University of Western Australia Chapter 5
with CFRP wraps and their performance after exposure to real marine
conditions
5.1 Abstract
Offshore structural members may require repairing or strengthening due to various reasons as they age.
Repairing is often required due to wall thinning due to the corrosive marine environment. Strengthening may
become essential due to work over demands or additional strength requirements due to change in usage. This paper
investigated strengthening and repairing of offshore steel tubular members using Carbon Fibre Reinforced
Polymers (CFRPs) subjected to four-point bending in the first part. Underwater retrofitting was also compared
with conventional in air retrofitting process within this part. The results indicated underwater retrofitting works
equally good as in air retrofitting. The effect of the number of layers of retrofit was investigated parametrically for
repairing and strengthening. Repairing corroded steel tubular members with CFRP was found to be very effective
under bending. Strengthening of the steel tubular members also showed significant improvement in performance,
but the improvement remained at the same level even when the number of CFRP layers increased. The results
were presented including ultimate strength, load displacement behaviour, energy absorption, ductility and strains.
In the second part of the study, the durability of CFRP retrofitted tubular members was examined using ‘Real
Corrosion’ test where the retrofitted members were kept in a river (saline water) for a period of one year. While
corrosion was observed in the non-retrofitted members, the CFRP layers prevented corrosion of retrofitted
specimens. The bending test of the Real Corrosion specimens indicated slight improvement in their performance
as well. It was observed that the retrofit did not lose any structural integrity due to exposure to marine or underwater
conditions for a prolonged period of one year. Lastly, a lower-bound interaction equation was developed for
repaired steel tubular members and was found satisfactory. Future studies are essential to understand longer term
109
The University of Western Australia Chapter 5
5.2 Introduction
5.2.1 General
Generally, offshore jacket structural elements are tubular as they can be equally effective against wave,
wind and current loads acting upon them in any direction. These structural steel members may need repairing or
strengthening during their design life. Repairing may be required due to corrosion of steel due to the extreme
marine conditions. Repairing could also be essential due to damage of these members from vessel impact or
dropped objects as well. Strengthening of offshore structural members could become necessary due to
unanticipated load carrying requirements arising due to additional equipment, change in usage of the structure or
In traditional practice, offshore industry would weld or bolt steel sleeves or patches to members for the
purpose of repairing or strengthening or ultimately would replace them with new steel sections. The main
disadvantage for all these solutions is that they need complex fabrication and installation procedure and would
typically add dead weight to the structure. The new steel structural members or attachments would be also subject
to corrosion in the same way as the old member got corroded. The performance of bolted sleeves would also
degrade over time due to loss of friction between the structural member and the sleeve. Moreover, to achieve most
of these retrofitting solutions the operations on the offshore facility needs to be shut down as they are hot works.
This will incur an indirect economic loss on top of the actual expenditure for retrofitting. A cost-effective and
structurally sound alternative for these traditional retrofitting measures is the use of Fibre Reinforced Polymers
(FRPs) to retrofit steel structures. However, this technology is not very popular in the offshore industry yet due to
the lack of knowledge about how this can be adopted in underwater or marine conditions.
Retrofitting of steel structural members using FRPs has been successfully utilised for onshore conditions.
The biggest knowledge gap in adopting this technology was whether this could be adapted for an underwater or
marine scenario. George et al. demonstrated that underwater repairing and underwater strengthening using CFRP
(Carbon Fibre Reinforced Polymers) of structural members under concentric and eccentric axial loading would be
beneficial [1, 2]. This paper examines the same retrofitting technique utilised for structural members under bending
loads and hence becomes a continuation of the earlier works. The effect of actual marine exposure on the durability
and performance of CFRP retrofitting is investigated for the first time by any researcher in this paper. A collation
of all experimental works from previous papers and this study is achieved through the interaction behaviour for
110
The University of Western Australia Chapter 5
Structural retrofitting of steel structures using FRP composites is an established technology with most of
the research focusing on onshore applications. Earlier researchers have addressed some of the key aspects of this
technology. This paper is trying to bridge some of the existing main gaps in the research on structural retrofitting
As already mentioned above, this paper is a continuation of the experimental studies presented by George
et al in the previous papers [1, 2]. In those papers, tubular members repaired and strengthened underwater subjected
to concentric and eccentric axial loads were investigated in detail. Whereas this paper addresses underwater
retrofitting of structural members subjected to pure bending loads. The bending behaviour of retrofitted steel
tubular members were successfully demonstrated by other researchers. Retrofitted steel tubular members under
bending and bearing loads were studied by Elchalakani et al. [3, 4]. Effect of the number of layers and the
orientation of FRP retrofit on steel tubular members under bending was studied by Kabir et al. [5, 6]. Behaviours
of other typical steel cross-sections retrofitted with FRPs under a variety of loading scenarios were investigated
by other researchers [7-11]. However, none of these works presented a study looking into underwater or marine
retrofitting of steel structural members. Seica and Packer demonstrated the feasibility of rehabilitation of steel
tubular members taking bending loads using FRPs in underwater conditions and compared the results with typical
in air rehabilitation [11]. They tried to use products which were meant to be used for in air application and curing
in underwater conditions. Even though, they were able to show improvement in flexural capacity for specimens
wrapped and cured underwater than the bare steel control specimen, the underwater strengthened specimens
showed considerably lower flexural capacity than specimens wrapped and cured in air. This was understandable
as the products were not designed to be used for underwater application and curing. This study is trying to expand
this knowledge by experimenting with retrofit materials that are specifically designed to perform well in
underwater conditions.
Some other studies have identified successful use of FRP retrofitting technology for specific offshore
applications like pipelines and risers [12-15]. These applications had their load primarily accounted from the
internal pressure of a pipe which is fundamentally different to the structural applications addressed in this paper.
However, successful demonstration of repairing or strengthening for underwater steel tubular structures ensures
111
The University of Western Australia Chapter 5
One of the most critical reasons why FRP retrofitting technology is not widely adopted yet in the offshore
industry is due to the concerns about the long term performance and durability of FRP retrofits in harsh marine
environments. Durability of FRP retrofitted steel plates were studied by accelerated corrosion testing by
Batuwitage et al. [16]. This study revealed degradation of CFRP material properties reducing the effectiveness of
such retrofits in marine conditions. However, accelerated corrosion may not realistically replicate the actual
scenario occurring in a retrofitted member in marine conditions. The material loss or corrosion of a metal substrate
can be simulated at accelerated pace using an electrochemical reaction, but the same reaction will not be useful in
mimicking the change in properties of a FRP layer over time. Leong et al. who monitored an actual FRP retrofit
on an offshore riser in the splash zone reported no visual degradation or damage of the FRP [17]. These
contradicting results called for a more detailed study into this phenomenon. Hence, the durability and performance
of CFRP retrofitted steel structural members exposed to marine conditions is studied through ‘Real Corrosion’
The main aim of the paper was to investigate the feasibility of using Fibre Reinforced Polymers for
structural repairing and strengthening of offshore steel tubular members under bending loads. The first objective
was to experimentally investigate the structural performance of in air and underwater repaired steel tubular
members. Similarly, the second objective was to experimentally investigate the structural performance of in air
and underwater strengthened steel tubular members. The third objective was to qualitatively understand the effect
of CFRP retrofitted specimens when they are exposed to underwater or marine conditions through ‘Real Corrosion’
tests. The last objective was to examine the interaction behaviour of tubular members repaired with CFRP under
5.3 Methodology
5.3.1 Specimens
The specimens were pipes made of steel of grade API 5L Grade X52. The length of the specimens, L was
equal to 1800 mm with overhangs of 100 mm on both ends. The nominal Young’s modulus and yield strength of
the pipe material was 200 GPa and 345MPa respectively. The outside diameter, Do of the pipes were 114.3 mm
with a thickness, t of 6 mm and the length of corrosion, Lc was fixed as 3D where D is the nominal diameter of the
pipe (100 mm). The thickness of corrosion, tc = 1.2 mm (20% of pipe thickness) was machined out for all corroded
specimens. Table 5.1 summarises these geometrical properties as well as the mechanical properties of the steel
112
The University of Western Australia Chapter 5
cross-section at the mid-span such as moment of inertia (I), elastic section modulus (Zs), plastic section modulus
(Zp), compressive yield strength (Pyield), elastic moment capacity (Mys) and plastic moment capacity (Mps).
The convention utilised to name the specimens is similar to what is done in George et al. [1]. Intact and
strengthened specimens were named with ‘I’ in the beginning whereas corroded and repaired specimens’ names
started with ‘C’. The bending load condition was represented using ‘B’ which distinguished these specimens from
other axially loaded specimens. A bare steel specimen was named with ‘F0’ describing the usage of zero layers of
fibre. Similarly, repaired and strengthened specimen were named with ‘F2’, ‘F4’ or ‘F6’ to indicate the number of
layers of CFRP used. The rehabilitation process done in air and underwater were distinguished by adding ‘A’ and
‘W’ at the end of their names respectively. The Real Corrosion specimens were named with ‘RC’ at the end of
their names and since there were repeat tests in Real Corrosion samples, they were named with ‘RC1’ or ‘RC2’ to
5.3.2 Materials
Materials used for strengthening and repair of the specimens were TechnowrapTM products [18].
TechnowrapTM 2K, GFRP was used as the first layer for both strengthening and repairing to avoid direct contact
of CFRP with steel surface to prevent galvanic corrosion (electrochemical reaction leading to loss of steel when
coming in direct contact with carbon fibres). TechnowrapTM SRS (Structural Rehabilitation System) was the CFRP
used for preparing the specimens. The normal resin system, TechnowrapTM HA (High Ambient) was used for the
in air specimens and the subsea resin, TechnowrapTM Splashzone was utilised for the underwater specimens.
The length of repair, LR was calculated as the sum of LC (= 3D), extended 1D after the corroded region at
both ends and appropriate tapering. LR can be approximated as 6D or in numerical terms, 600 mm. The length of
strengthening, LS was also taken as 6D for the specimens. The laboratory retrofitting process for both strengthening
and repairing were identical. The steel surface was prepared with a cleanliness of Sa2.5 and a roughness profile of
60 µm (minimum) following the ISO guidelines [19]. Surface preparation of both in air and underwater specimens
was done using air blasting which is a limitation of the underwater specimens in the paper. The rest of the process
was exactly like the retrofitting process using stricture consolidation mentioned by George et al. [1, 2]. Retrofitting
of underwater specimens was carried out in a tank filled with saline water taken from the Swan River in Western
113
The University of Western Australia Chapter 5
The durability of a retrofitted pipe, when exposed to an underwater or marine condition for a long period,
is one of the main objectives of this study. For the sake of brevity, the exposure of specimens to underwater or
marine conditions was termed as ‘Real Corrosion’ in this paper. Real Corrosion was achieved by hanging
specimens in an approximate 2m water depth of the Swan River in Western Australia for a period of 1 year. As
per ISO , the minimum design life of a composite repair is 2 years and a design life up to 20 years can be provided
for a composite repair [19]. However, due to practical restrictions, the Real Corrosion experiment in this study
was limited to a period of 1 year only, which would still be useful in providing some useful indications on how
durable the composite retrofitting is in underwater or marine conditions. An alternative methodology to check the
durability of metallic substrates retrofitted with FRP composites is through accelerated corrosion [16]. In the
accelerated corrosion method, an electrochemical reaction would be introduced to speed up the natural corrosion
process in metals thus able to achieve degradation equivalent to long periods of real corrosion within a short period.
Batuwitage et al. demonstrated that CFRP retrofit would undergo severe degradation due to exposure through
accelerated corrosion [16]. Nonetheless, a CFRP retrofit might behave differently in a real marine condition which
A total of 6 experimental specimens were utilised for Real Corrosion testing as described in Table 5.2. The
specimens were installed with endcaps to prevent water ingress causing internal corrosion and hung into the river
from a floating deck as shown in Figure 5.1. The specimens were placed in the river in a horizontal position hung
with 2 ropes in a similar pattern in which they would be tested later. Once retrieved from the river, the weight and
thickness of marine growth attached to specimens measured. Later, the marine growth was removed from the
specimens manually at first and cleaned completely using water jet later. It was comparatively easier to remove
the marine growth on the retrofitted zone than on the steel surface, which could be a virtue for these underwater
FRP retrofits. The weight loss due to Real Corrosion was recorded once marine growth was completely removed.
The different stages involved in the retrieval of Real Corrosion specimens to make them ready for structural testing
are presented in Figure 5.2. The retrieved specimens were then structurally tested under four-point bending. The
results of these structural tests were compared with that of the experimental specimens which did not undergo Real
Corrosion exposure to understand the effect of Real Corrosion on the structural integrity of a retrofitted member.
114
The University of Western Australia Chapter 5
Figure 5.1 Real Corrosion specimens: (a) prepared with endcaps to be placed in the river and (b) lowered into
river hung with two ropes
Figure 5.2 Retrieved Real Corrosion specimens: (a) covered with full marine growth, (b) manual removal of
marine growth, (c) water jet washing and (d) fully cleaned for structural testing
Specimens were gridded using white markers on the exterior CFRP to capture and observe the local
deformation and damage location in the FRP retrofits easily. All bending specimens were gauged with 2 foil strain
gauges one in the axial direction and another in the hoop direction at their midspan in the tension face. The four-
point bending test set up used for these specimens is as shown in Figure 5.3. The specimens were placed on the
support rollers set at a span of 1800 mm and the loading rollers were placed on top of the specimen at a span of
600 mm in the middle. 100 mm wide saddle shape brackets were used at the support and loading points to ensure
bearing of the pipe did not occur during the structural testing. The load was increased gradually and the beam
response (i.e., load-deflection) and strain gauges readings were monitored continuously. The testing was continued
115
The University of Western Australia Chapter 5
even past the ultimate load until a midspan displacement reached a value of at least 180 mm, 10% of the free span
of the specimen. Certain specimens did not reach ultimate load even after reaching this displacement limit and the
testing had to be stopped when they started to slip off the support rollers.
The point of failure was investigated with different failure modes observed in the retrofit layers. The
ultimate load was identified as the maximum load observed during the testing. The load displacement curves were
also recorded for all specimens and compared with each other to reach useful inferences. Energy Absorption (EA)
of a specimen was estimated from the area under the load displacement curve until midspan displacement reaches
10% of the free span of the specimen (i.e., 180 mm). The strain values recorded were plotted against their
corresponding loads for easy analysis and comparison. Strain values at ultimate load and maximum strain values
were recorded for each specimen. Ductility Index (DI) of the specimens was calculated with a modified version of
the method introduced by Foster and Attard [20]. This energy-based method calculates DI as given below.
DI (5.1)
where ABC or elastic energy absorbed by the specimen which is calculated as the area under the load
displacement curve up to Δ75, the displacement point obtained by extending the linear elastic part of the curve (0
to 0.75Pu) to peak load (Pu) and ADE or total energy absorbed by the specimen is calculated as the area under the
load displacement curve up to displacement equal to 9Δ75. The identification of the five points, A to E is further
116
The University of Western Australia Chapter 5
Repairing was found to be successful in regaining the structural properties of the pipe due to wall thinning.
The structural performance improved for both in air and underwater repaired specimens with 2 layers of CFRP.
The structural response also improved as the number of repair layers increased. The repaired specimens under
bending loads behaved in different ways as shown in Figure 5.5. The point of failure was within the corrosion zone
for repairs with 2 layers of CFRP, but it shifted outside the corrosion zone but well within the retrofit zone for
repair with 4 CFRP layers. Points of failure shifted outside the retrofit zone as the repair layers were increased to
6 layers. The retrofit zone of CBF6W did not undergo any significant deformation and remained straight. Figure
5.5 depicts the point of failure in all the repaired specimens after bending tests. This behavioural trend was very
similar to the repaired specimens under axial compressive loads presented by George et al. [2]. Crushing of the
CFRP laminates on compression face and rupturing of CFRP on the tension face was observed at the point of
failure for the repaired specimens except for CBF6W where the points of failure just outside the retrofit zone were
showing yielding of the steel. Debonding of the repair layers from the steel pipe along the tension face of the
117
The University of Western Australia Chapter 5
The ultimate loads from all repaired specimens showed higher values than the bare steel control specimen.
The in air repair specimen, CBF2A and underwater repair specimen, CBF2W portrayed very comparable ultimate
loads of 91.32 kN and 91.52 kN respectively. There was nothing much to distinguish between an underwater repair
with an in-air repair in terms of ultimate strength. Both these ultimate strengths were greater than the corroded
specimen (CBF0) as well as the intact specimen (IBF0). Hence, with 2 layers of CFRP, the corroded member (i.e.,
1.2mm corrosion out of 6 mm wall thickness) was able to reach an ultimate strength greater than that of an intact
member. The ultimate load also progressively increased as the number of retrofit layers increased from 2 layers to
4 and 6. CBF4W and CBF6W had ultimate loads of 93.42 kN and 105.00 kN respectively. The improvement
percentages of the ultimate load from the control case are tabulated in Table 5.3.
The load displacement curves of the repaired specimens, as shown in Figure 5.6 and Figure 5.7, were
investigated and it was observed that the load increased with small displacement in the elastic portion of the curve.
118
The University of Western Australia Chapter 5
When the specimen started to yield the displacement started to pick up without taking up as much load in the
elastic portion. This continued or even plateaued until the specimen was fully plasticised where the ultimate load
was recorded. Post ultimate load, an unloading portion in the curve was observed where the load was dropping but
the displacement was still increasing. The load displacement curve of CBF2A and CBF2W as shown in Figure 5.6
were almost identical showing underwater repair can work equally good as in air repair. They were also
significantly better than the behaviour of CBF0 indicating the effectiveness of FRP repair under bending loads.
Similarly, load displacement behaviour of all underwater repaired specimens is compared in Figure 5.7. All of
them show better performance than CBF0 and the performance improved as the number of repair layers increased.
CBF4W showed significant improvement from CBF2W especially in the unloading portion of the curve. CBF6W
showed further improvement in the load displacement behaviour. The unloading portion of the CBF6W was not
Figure 5.6 Load displacement behaviour of repaired in air vs. underwater specimens
119
The University of Western Australia Chapter 5
Energy Absorbed (EA) by the repaired specimens were found out and their improvement from CBF0 is
tabulated in Table 5.3. CBF2A absorbed 11.7% more energy than CBF0 whereas CBF2W had 12.9% improvement
showing a slight advantage for the underwater repair. CBF4W and CBF6W had an improvement of 19.0% and
36.3% respectively showing the energy absorbed by a specimen increased with the number of repair layers.
As FRP composites are brittle on their own, the ductility of the repaired specimens should be critically
investigated. Ductility Index (DI) of the repaired specimens were calculated and their ratio to the control case,
CBF0 was found out and tabulated in Table 5.3. DI ratios of CBF2A and CBF2W were 1.06 and 1.05 respectively
showing better ductility than the corroded specimen. However, as the repair layers further increased the DI values
dropped below the corroded specimen. CBF4W and CBF6W had DI ratios of 0.90 and 0.94 respectively. Even
though the total energy absorbed (ADE in Figure 5.4) by these 2 specimens was much higher than the control case,
the elastic energy absorbed (ABC in Figure 5.4) was also significantly larger due to the higher stiffness of these
Strain values of the repaired specimens are shown in Figure 5.8. Axial strains are tensile and hoop strains
are compressive in nature for all specimens. They tend to increase steadily in the elastic portion and suddenly
increases as the specimen starts to yield. CBF0 had a peculiar behaviour in the strain values around the yielding
point, the strain values decreased slightly before increasing significantly. This behaviour was consistent with all
other bare steel specimens too. The hoop strain values of CBF2A and CBF2W were very similar, but the axial
120
The University of Western Australia Chapter 5
strain in CBF2A is much lower and inconsistent with CBF2W. Considering the observed similarities between
CBF2A and CBF2W in other readings (ultimate load or hoop strain), the recorded dissimilarity between them for
the axial strain could possibly be due to some errors with these readings. In the case of the underwater repaired
specimens, we can see the strain values were lower for both axial and hoop strains as the repair layers increased.
The strain values at ultimate load and the maximum strain values recorded during the experiment are tabulated in
Table 5.3, however it should be noted that most of the strain gauges were damaged due to excessive deformations
during the tests before the specimens reached the ultimate load.
Strengthening was found to be very successful in improving the bending performance of the specimens.
For most of the strengthened specimens, points of failure were observed with the steel pipe yielding just outside
121
The University of Western Australia Chapter 5
the retrofit zone as shown in Figure 5.9. The retrofit zone did not bend and remained almost straight for these
specimens. IBF2A had shown an out of pattern behaviour from the rest of the strengthened specimens. IBF2A had
a point of failure within the retrofitted zone with the CFRP crushing on the compression face. This could have
occurred due to an initial imperfection at this location in the steel pipe used for preparing IBF2A. Hence, the results
of IBF2A were not used to draw inferences. IBF2W also had the point of failure within the retrofit zone with
rupture of the CFRP laminate on the tension face and had shown slight bending within the retrofit zone. IBF4W
and IBF6W had points of failure outside the retrofit zone with no significant bending in the middle. All
strengthened specimens also had observed debonding from the steel surface on the tension face similarly to the
repaired specimens.
The ultimate loads from all strengthened specimens showed higher values than the bare steel control
specimen. The in air and underwater strengthened specimens, IBF2A and IBF2W, showed different ultimate loads
122
The University of Western Australia Chapter 5
of 106.13 kN and 103.04 kN respectively. The difference in the ultimate loads was purely due to the different
failure modes in these two specimens as described earlier. IBF2A was the one that showed better ultimate load as
the retrofit layers participated in the yielding at the point of failure. However, the retrofit zone participated
negligibly in the points of failure for the underwater strengthened specimens as the failure points were pushed
outside due to the higher strength and stiffness in the retrofit zone. The increase in the number of strengthening
layers did not have a significant impact on the ultimate strength for the same reason. All 3 underwater strengthened
specimens had comparable ultimate strengths. More details about the ultimate strength and improvement from
The load displacement curves of the strengthened specimens when investigated also inferred something
similar. They all had significant improvement from the control case, IBF0. The testing of IBF0 was not taken to
larger displacement due to a limitation in the experimental set up and hence the unloading portion of the curve was
not captured completely. The load displacement curve of IBF2A and IBF2W as shown in Figure 5.10 portrayed
the different failure modes in these two specimens. IBF2A had showcased better performance than IBF2W
especially after yielding where it is indicating larger involvement of retrofit layers in yielding. IBF2W had a load
displacement curve with a sharp jerk towards the end of the plateau region corresponding to the cracking of the
CFRP in this specimen. The comparison of the load displacement curves of all underwater strengthened specimens
as shown in Figure 5.11 proved that increasing the number of retrofit layers did not improve the performance of
these specimens. A more significant benefit of strengthening of a bending member might be possible when the
entire length of the structural member is retrofitted as found out by George et al. for axially loaded specimens [2].
123
The University of Western Australia Chapter 5
Figure 5.10 Load displacement behaviour of in air and underwater strengthened specimens
Energy Absorbed (EA) by the strengthened specimens were also calculated and their improvement from
IBF0 is tabulated in Table 5.3. The improvement calculated was higher due to the incomplete data for IBF0. The
EA for all three strengthened underwater specimens remained almost the same for all. Ductility Index (DI) of the
strengthened specimens were calculated and their ratio to the control case, IBF0 was found out and tabulated in
Table 5.3. DI ratios of IBF2A and IBF2W were 1.05 and 1.01 respectively showing better ductility than IBF0.
124
The University of Western Australia Chapter 5
However, DI of IBF4W and IBF6W were less than that of IBF0 indicating the lower ductility of the strengthened
specimens.
The strain values recorded from the strengthened specimens as shown in Figure 5.12 followed the general
trend of the axial strain being tensile and hoop strains being compressive in nature. Strain values from IBF0
resembled the strains from CBF0 and confirmed that the bare steel specimens (intact or corroded ones) behave in
a similar pattern. Axial strains of IBF2A and IBF2W were very different as they had different modes of failure.
IBF2W had the lowest axial strain among the three underwater strengthened specimens, possibly due to the slightly
different mode of failure in this specimen. Ignoring IBF2W results, it was observed that the axial strain values
decreased as the retrofit layers increased. Similarly, the hoop strain values also decreased as the retrofit layer
increased. This behaviour was consistent with that of underwater repaired specimens as shown in Figure 5.8. Axial
and hoop strain values at ultimate load and maximum recorded during the tests are tabulated in Table 5.3.
125
The University of Western Australia Chapter 5
The physical observation when the Real Corrosion specimens were retrieved from the river after 1 year are
tabulated in Table 5.4. The weight of each specimen before deploying them into the river was taken. The specimens
when retrieved from the river had marine growth on them with the average diameter ranging from 159.3 mm to
183.5 mm. This marine growth had substantial weight and weights of the specimens with wet and drained marine
growth was recorded. The recorded weight will indicate that the specimens were carrying on average around 45%
of their self-weight.
The weight of the specimens after completely removing the marine growth with pressure cleaning was also
found out and compared with the weight before Real Corrosion exposure. This comparison gave information about
the specimens losing weight due to corrosion. CBF0RC, the bare steel corroded specimen lost 1.39% of its weight.
However, CBF2RC1 and CBF2RC2, the repaired specimens lost only 0.66% of their weight on average which is
about 50% of what is lost from the bare steel specimen. This is a clear indication that the repaired specimens
reduced the corrosion of the steel tubular section underneath the repair layers. IBF0RC (i.e., the bare steel intact
specimen) lost 1.94% of its weight and it is significantly higher than CBF0RC (i.e., the bare steel corroded
specimen) as one of the end caps of this specimen was completely corroded away and letting water inside the
tubular member causing a small degree of internal corrosion too. Nonetheless, IBF2RC1 and IBF2RC2, the
strengthened specimens only lost 0.65% of their weight proving that the retrofit layers have prevented the steel
pipe underneath from corroding away. This property of CFRP retrofit layers would be highly beneficial for
repairing and strengthening offshore or marine structural members as they would insulate the substrate from any
Corroded and repaired specimens after structural testing looked as presented in Figure 5.13. The point of
failure was in the middle of the corrosion zone for CBF0RC as expected. The point of failure for the repaired
specimens (CBF2RC1 and CBF2RC2) seemed to be just outside the corrosion zone but still within the retrofit
zone. Even though these specimens should have behaved like CBF2W due to the same repair layer count, the
failure mode was more resembling the failure mode in CBF4W. This could be attributed to a slight improvement
in the retrofit properties as the CFRP layers were exposed to continuously applied circumferential confinement
due to external hydrostatic pressure from the river water for one year. Overall, this means that the CFRP
retrofitting will not lose structural properties after Real Corrosion exposure as suspected.
126
The University of Western Australia Chapter 5
Figure 5.13 Real Corrosion - Corroded & Repaired specimens after testing
The ultimate load values found from these Real Corrosion specimens and their change from corresponding
non exposed cases are tabulated in Table 5.5. The ultimate load of CBF0RC had dropped by 12% from CBF0
indicating wall thinning due to Real Corrosion. The ultimate load of CBF2RC1 and CBF2RC2 were greater than
CBF2W by 8.0% and 3.9% respectively and they were even higher than that of CBF4W. This clearly means
The load displacement behaviour of Real Corrosion - corroded and repaired specimens are presented in
Figure 5.14. CBF0RC showed much lower performance than CBF0 owing to the additional corrosion during one
year of exposure. The load displacement behaviour of the repaired specimens after Real Corrosion was like that
of CBF4W until they started to yield and post ultimate load, they have behaved similarly to CBF2W. The load
displacement from yielding to ultimate load was more pronounced for these specimens than CBF2W and CBF4W
proving slight improvement in the retrofit properties. Overall, the load displacement behaviour for the Real
Corrosion specimens repaired with 2 layers of CFRP was in between the behaviour of CBF2W and CBF4W.
127
The University of Western Australia Chapter 5
Figure 5.14 Load displacement behaviour of Real Corrosion - Corroded & Repaired specimens
Energy absorbed (EA) by the Real Corrosion specimens as shown in Table 5.5 revealed a similar trend.
CBF0RC absorbed much lower energy than CBF0 as expected. CBF2RC1 and CBF2RC2 on average had 15.5%
more energy than CBF0 which was higher than EA by CBF2W and lower than CBF4W. DI calculated for the Real
Corrosion specimens were tabulated with their ratios with the control case CBF0 in Table 5.5. CBF0RC had DI
almost like CBF0 with a ratio of 1.01 as they were both bare steel specimens. The DI values for the repaired
specimens which underwent Real Corrosion were much lower than that of CBF0 and indicating lower ductility of
retrofitted specimens. The DI values of both CBF2RC1 and CBF2RC2 were also lower than that of CBF2W.
Strain values of the Real Corrosion - corroded and repaired specimens are presented in Figure 5.15. The
strain values showed a similar pattern with CBF0 and CBF2W. The repaired specimens after Real Corrosion
showed slightly larger axial and hoop strains than CBF2W. The strain values at the ultimate load and the maximum
recorded during the test were presented for these specimens in Table 5.5.
128
The University of Western Australia Chapter 5
Figure 5.15 Strain values of Real Corrosion - corroded & repaired specimens
IBF0RC, the intact Real Corrosion specimen was discarded from structural testing as it had internal
corrosion due to water ingress through the corroded endcap. The strengthened Real Corrosion specimens had
points of failure outside the retrofit zone with the steel yielding as shown in Figure 5.16. The retrofitted zone in
both IBF2RC1 and IBF2RC2 did not bend significantly but had developed small wrinkles in them towards the end
of structural testing. Their failure mode was very similar to the non-exposed underwater strengthened specimens.
129
The University of Western Australia Chapter 5
The ultimate load values found from the Real Corrosion strengthened specimens and their change from
IBF2W is tabulated in Table 5.5. The ultimate load of IBF2RC1 and IBF2RC2 were slightly smaller than IBF2W
by 2.2% and 1.0% respectively. The slight depreciation in ultimate load could be attributed to the corrosion of the
The load displacement behaviour of Real Corrosion strengthened specimens is presented in Figure 5.17.
The load displacement curves of IBF2RC1 and IBF2RC2 followed the same pattern as each other. Their curve
showed higher performance than that of IBF2W after yielding indicating higher involvement of CFRP retrofitting.
The Real Corrosion strengthened specimens reach the ultimate load at a lower displacement value and the
unloading portion of the load displacement curve showed slightly lower performance than IBF2W. This again was
due to the corrosion of steel outside the retrofit zone due to exposure in the river.
130
The University of Western Australia Chapter 5
Energy absorbed (EA) by the Real Corrosion strengthened specimens are shown in Table 5.5. IBF2RC1
and IBF2RC2 on average had 39% more energy than IBF0 which was higher than EA by all 3 underwater
strengthened specimens. This again pointed to slightly better performance by Real Corrosion specimens due to
improved structural properties of the retrofit layers after 1 year of Real Corrosion. DI calculated for the Real
Corrosion strengthened specimens were tabulated with their ratios with the control case IBF0 in Table 5.5.
IBF2RC1 and IBF2RC2 had DI values greater than DI of IBF0 but lower than IBF2W. This result cannot be
inferred as strengthened specimens had better ductility than steel as the failure mode in these specimens were
Strain values of the Real Corrosion strengthened specimens are presented in Figure 5.18. The strain values
showed a similar pattern with IBF0 and IBF2W. The Real Corrosion strengthened specimens showed the slightly
larger axial strain and smaller hoop strains than IBF2W. The difference could be attributed to the slightly different
failure mode in IBF2W with CFRP rupture in the middle. The strain values at the ultimate load and the maximum
recorded during the test of Real Corrosion strengthened specimens are presented in Table 5.5.
131
The University of Western Australia Chapter 5
The interaction behaviour of retrofitted steel tubular members between axial compressive load, N and
bending moment, M was investigated. Experimental results of specimens under concentric and eccentric
compressive loads from George et al. were utilised in this part of the study [1]. The results of intact and corroded
members under axial loads were taken along with the bending test results presented earlier in this paper to form
the control interaction behaviours. Specimens repaired in air and repaired underwater with 2 layers of CFRP were
studied to understand how the interaction behaviour was different to bare steel cases. The strengthened specimens
under axial loads as presented in George et al. had their point of failure outside the retrofit zone which made them
not suitable to be included in this part of the study [2]. Similar was the case for the strengthened specimens under
bending where the points of failure were outside the retrofit zone.
132
The University of Western Australia Chapter 5
The data used for the N-M interaction behaviour is tabulated in Table 5.6. The axial load, N for each case
was taken as the ultimate load recorded for a particular specimen. In the case of a concentrically loaded specimen,
the bending moment, M is zero as there would be no bending moment. In the case of a specimen loaded
eccentrically, N is the ultimate load and M is calculated as the product of its ultimate load and eccentricity in
loading (i.e., 20 mm, 40 mm, 60 mm or 80 mm as noted in George et al. [1]). In the case of a bending specimen,
M is calculated as the bending moment corresponding to the ultimate load recorded for the specimen and N is
The N-M interaction behaviour for all 4 specimen groups is plotted in Figure 5.19. A general trendline for
each specimen group is plotted. All the interaction trendlines showed a similar pattern with a slight dip in their
behaviour due to combined loading based on the experimental results. The difference between intact and corroded
specimens is more significant as the interaction diagram is closer to pure axial load, P0 (i.e., N when M = 0 kNm).
The difference between intact and corroded interaction curves reduced and became the minimum at pure bending
load, M0 (i.e., M when N = 0 kN). The interaction behaviour of repaired in air and repaired underwater specimens
remained almost identical to each other showcasing both behave the same. The interaction behaviour of the
repaired specimens showed that they had improved the performance from the corroded level significantly.
However, the repair with 2 layers of CFRP was not completely successful in bringing back the structural
performance to the intact level. The performance of repaired specimens was slightly inferior to the intact specimens
under pure axial load, but repaired specimens had better performance than those intact ones as more bending load
was taken by them. The repaired specimens showed better behaviour than the intact ones after M/N > 20 mm.
133
The University of Western Australia Chapter 5
Working with dimensional N-M interaction curves as typically shown in Figure 5.19 is not a practical
solution, and hence to generalise the N-M interaction behaviour, a normalized N-M interaction behaviour was
generated based on these experimental results. This normalized N-M interaction behaviour would have certain
limitations. This would only be useful if the steel tubular members are failing under axial loads due to the yielding
of the cross section as for these experimental specimens. This interaction behaviour would become inapplicable
once the point of failure shifts outside the retrofit zone as in the case of the strengthened specimens.
N and M values were normalised by dividing them by their corresponding pure axial load (N0) and pure
bending load (M0) respectively. N/N0 and M/M0 are tabulated in Table 5.6. Once plotted the normalized N-M
interaction behaviour was plotted, all experimental results were following a second order polynomial trendline as
presented in Figure 5.20. Practically, using this second order equation to find whether a retrofitted tubular member
would withstand a combination of N and M seemed difficult. Hence, a simplified lower-bound control line was
established as shown in Figure 5.20. Based on this a retrofitted tubular member could withstand a combined
0.80 (5.2)
134
The University of Western Australia Chapter 5
The only drawback for this is that we need a good estimate of the pure axial load, N0 and the pure bending
moment, M0 of a retrofitted tubular member through experimental or analytical methods. The N-M interaction
behaviour presented here would help for structural design of retrofitted tubular members under combined loading.
The interaction lower-bound equation presented here is very relatable to the interaction equations of steel
structural members under axial compressive and bending loads presented in AS 4100 [21] where the lower-bound
equation for biaxial bending in axially loaded members is as given in Equation (5.3).
∗ ∗ ∗
1 (5.3)
∅ ∅ ∅
N*, Mx* and My* are the design axial load and design moments in x and y direction respectively. Similarly,
Ns, Msx and Msy are the nominal section axial load capacity and nominal section moment capacities in x and y
directions respectively. Ø is the capacity factor defined within the code for different scenarios.
Simplifying Equation (3) to make it applicable for the experiments presented in the paper, Equation (5.4)
can be generated. Since there are no moments in the y direction the last term can be removed. The capacity factor,
Ø for members subjected to combined action is given as 0.90 in AS 4100 [21] which when taken to the right hand
∗ ∗
0.90 (5.4)
Equation (5.4) now takes a similar form as the lower-bound equation (i.e., Equation (5.2)) generated for
repaired steel tubular members under combined loads. The main difference between them is that the denominators
are member capacities in Equation (5.2) and section capacities in Equation (5.4). Currently, the gap in knowledge
is lack of understanding of section capacities of a steel tubular cross section repaired with FRP composites. More
studies should be conducted to address this in the future to make more effective design tools for repairing steel
5.6 Conclusions
The paper focussed on the bending behaviour of retrofitted steel tubular members. Repairing of artificially
corroded structural members with CFRP wraps and strengthening intact structural members with CFRP wraps to
improve their performance were investigated. Comparison on how repairing and strengthening work in air and
underwater was carried out with the help of bending tests. Effect of the number of retrofit layers on underwater
135
The University of Western Australia Chapter 5
repaired and underwater strengthened specimens under bending loads was also studied. The performance of
underwater retrofitting after exposure to an actual marine environment for a period of one year was also
investigated through Real Corrosion testing. An interaction behaviour for repaired steel tubular members for a
combination of axial compressive load and bending moment was developed in this paper. The key conclusions
drawn from all these experimental studies on specimens with their given geometrical and mechanical
Repairing of corroded tubular members under bending loads had improved their structural performance
Underwater repairing and in air repairing had produced almost identical results showcasing underwater
repairing could work equally good as conventional in air repairing for structural members under
bending.
The structural performance of the underwater repaired specimens improved as the number of retrofitted
layers increased and as the stiffness of the repair increased the point of failure was shifted outside the
retrofit zone.
Strengthening of intact structural members under bending was found to be very successful with 2 layers
of CFRP, but the points of failure had shifted outside the retrofit zone. Hence, a further increase in
Real Corrosion results proved that retrofitting layers will prevent steel underneath from coming in
Repaired specimens after Real Corrosion exposure showed slightly better structural performance than
non-exposed repaired specimens. This could possibly be inferred due to the improvement of structural
properties of the retrofit laminates after constant exposure to external hydrostatic pressure in
underwater condition.
Strengthened specimens after Real Corrosion exposure performed almost as good as the non-exposed
strengthened specimens and the variation in performance was due to corrosion of the steel at the points
136
The University of Western Australia Chapter 5
of axial compressive load (N) and bending moment (M) could be withstood by a repaired steel tubular
member was developed. This would be applicable for compact tubes that would fail predominantly due
In conclusion, the application of underwater repairing and strengthening of offshore steel tubular
members using CFRP was successfully demonstrated through experiments. Exposure of these CFRP retrofits to
marine environments would not necessarily decrease their structural performance. More similar studies should be
carried out in this research area to develop confidence in this underwater retrofit technology to be widely adopted
5.7 Acknowledgements
The authors would like to thank ICR.IAS Joint Venture for providing composite materials and expert labour
for the preparation of the retrofitted specimens presented in the paper. The authors would like to extend their
gratitude to Royal Perth Yacht Club (www.rpyc.com.au) for using their marina for the conduct of Real Corrosion
tests. The authors would also like to acknowledge the efforts of Master of Professional Engineering (MPE) students
in the conduct of the experiments through RiverLab research funds at the University of Western Australia
(www.uwa.edu.au/projects/oceanworks-project-pages/riverlab).
5.8 References
[1] George JM, Kimiaei M, Elchalakani M, Fawzia S. Experimental and numerical investigation of underwater
composite repair with fibre reinforced polymers in corroded tubular offshore structural members under concentric
[2] George JM, Kimiaei M, Elchalakani M, Fawzia S. Underwater strengthening and repairing of tubular offshore
structural members using Carbon Fibre Reinforced Polymers with different consolidation methods. Thin-Walled
[3] Elchalakani M. Rehabilitation of corroded steel CHS under combined bending and bearing using CFRP.
[4] Elchalakani M, Karrech A, Basarir H, Hassanein MF, Fawzia S. CFRP strengthening and rehabilitation of
137
The University of Western Australia Chapter 5
[5] Kabir MH, Fawzia S, Chan THT. Effects of layer orientation of CFRP strengthened hollow steel members.
Građevinar. 2015;67:441-51.
[6] Kabir MH, Fawzia S, Chan THT, Gamage JCPH, Bai JB. Experimental and numerical investigation of the
behaviour of CFRP strengthened CHS beams subjected to bending. Engineering Structures. 2016;113:160-73.
[7] Bambach MR, Elchalakani M. Plastic mechanism analysis of steel SHS strengthened with CFRP under large
[8] Bambach MR, Elchalakani M, Zhao XL. Composite steel–CFRP SHS tubes under axial impact. Composite
Structures. 2009;87:282-92.
[9] Elchalakani M, Fernando D. Plastic mechanism analysis of unstiffened steel I-section beams strengthened with
[10] George JM, Kimiaei M, Elchalakani M. Experimental Study on Structural Rehabilitation of Severely
Damaged I-Beams Using Fibre Reinforced Polymers. Proceedings of the ASME 2021 40th International
[11] Seica MV, Packer JA. FRP materials for the rehabilitation of tubular steel structures, for underwater
[12] Alexander C, Cercone L, Lockwood J. Development of a Carbon-Fiber Composite Repair System for
Offshore Risers. ASME 2008 27th International Conference on Offshore Mechanics and Arctic Engineering 2008.
p. 389-405.
[13] Shouman A, Taheri F. An Investigation Into the Behaviour of Composite Repaired Pipelines Under Combined
Internal Pressure and Bending. ASME 2009 28th International Conference on Ocean, Offshore and Arctic
[14] Shouman A, Taheri F. Compressive strain limits of composite repaired pipelines under combined loading
[15] Shamsuddoha M, Islam MM, Aravinthan T, Manalo A, Lau K-t. Effectiveness of using fibre-reinforced
polymer composites for underwater steel pipeline repairs. Composite Structures. 2013;100:40-54.
[16] Batuwitage C, Fawzia S, Thambiratnam D, Al-Mahaidi R. Durability of CFRP strengthened steel plate
138
The University of Western Australia Chapter 5
[17] Leong A, Leong KH, Tan YC, Liew PF, Wood C, Tian W et al. OVERWRAP COMPOSITE REPAIRS OF
[19] ISO. Petroleum, petrochemical and natural gas industries - Composite repairs for pipework - Qualification
[20] Stephen JF, Mario MA. Experimental Tests on Eccentrically Loaded High Strength Concrete Columns. ACI
139
The University of Western Australia Chapter 5
5.9 Tables
Pipe dimensions Mechanical properties of the steel cross section at mid span Retrofit details
Serial Specimen Wrapping No. of No. of
Retrofit type L Do t Lc tc Pyield Mys Mps LR
no. ID and curing I (mm4) Zs (mm3) Zp (mm3) GFRP CFRP
(mm) (mm) (mm) (mm) (mm) (kN) (kNm) (kNm) (mm)
laminates laminates
1 CBF0 - Bare steel - - -
2 CBF2A In air Repairing 1 2 600
3 CBF2W 2000 114.3 6.0 300 1.2 2.32E+06 4.15E+04 5.51E+04 557.2 14.3 19.0 1 2 600
4 CBF4W Underwater Repairing 1 4 600
5 CBF6W 1 6 600
6 IBF0 - Bare steel - - -
7 IBF2A In air Strengthening 1 2 600
8 IBF2W 2000 114.3 6.0 0 0.0 3.00E+06 5.25E+04 7.04E+04 704.3 18.1 24.3 1 2 600
9 IBF4W Underwater Strengthening 1 4 600
10 IBF6W 1 6 600
Pipe dimensions Mechanical properties of the steel cross section at mid span Retrofit details
Serial Specimen Wrapping No. of No. of
Retrofit type L Do t Lc tc Pyield Mys Mps LR
no. ID and curing I (mm4) Zs (mm3) Zp (mm3) GFRP CFRP
(mm) (mm) (mm) (mm) (mm) (kN) (kNm) (kNm) (mm)
laminates laminates
1 CBF0RC - Bare steel - - -
2 CBF2RC1 2000 114.3 6.0 300 1.2 2.32E+06 4.15E+04 5.51E+04 557.2 14.3 19.0 1 2 600
Underwater Repairing
3 CBF2RC2 1 2 600
4 IBF0RC - Bare steel - - -
5 IBF2RC1 2000 114.3 6.0 0 0.0 3.00E+06 5.25E+04 7.04E+04 704.3 18.1 24.3 1 2 600
Underwater Strengthening
6 IBF2RC2 1 2 600
140
The University of Western Australia Chapter 5
Pu Change EA Change
ABC (J) ADE (J) DI Ratio Axial Hoop Axial Hoop
(kN) (%) (J) (%)
CBF0 79.51 - 12168 - 644 6854 10.6 1.00 - - 0.89 -1.32
CBF2A 91.32 14.9 13588 11.7 841 9530 11.3 1.06 - - 0.17 -0.40
CBF2W Corroded & Repaired 91.52 15.1 13744 12.9 860 9601 11.2 1.05 - - 0.73 -1.01
CBF4W 93.42 17.5 14481 19.0 1366 13083 9.6 0.90 - - - -
CBF6W 105.00 32.1 16585 36.3 1100 11054 10.0 0.94 0.40 -0.18 0.40 -0.18
IBF0 86.77 - 11888 - 699 7439 10.6 1.00 - - 0.80 -1.29
IBF2A 106.13 22.3 17425 46.6 1012 11272 11.1 1.05 0.52 - 0.53 -
IBF2W Intact & Strengthened 103.04 18.8 16068 35.2 1086 11711 10.8 1.01 - -0.83 0.31 -0.83
IBF4W 102.91 18.6 16235 36.6 1333 13105 9.8 0.92 0.42 -0.31 0.42 -0.33
IBF6W 103.50 19.3 15940 34.1 1278 12652 9.9 0.93 - - 0.24 -0.14
Note: -ve strain values are compressive strains
141
The University of Western Australia Chapter 5
Pu Change EA Change
ABC (J) ADE (J) DI Ratio Axial Hoop Axial Hoop
(kN) (%) (J) (%)
CBF0RC 69.98 -12.0 9374 -23.0 455.78 4893.99 10.7 1.01 - - 1.68 -0.26
Real Corrosion -
CBF2RC1 Corroded & 98.87 8.0* 13958 14.7 1628.06 13854.34 8.5 0.80 1.38 - 1.38 -0.26
Repaired
CBF2RC2 95.05 3.9* 14161 16.4 1195.53 12101.40 10.1 0.95 - - 1.61 -0.25
IBF0RC - - - - - - - - - - - -
Real Corrosion -
IBF2RC1 Intact & 100.76 -2.2** 16596 39.6 939.43 10342.96 11.0 1.04 1.11 -0.25 1.13 -0.25
Strengthened
IBF2RC2 102.02 -1.0** 16445 38.3 947.13 10171.97 10.7 1.01 0.60 - 0.69 -0.21
*Compared to CBF2W in Table 5.3
**Compared to IBF2W in Table 5.3
Note: -ve strain values are compressive strains
142
The University of Western Australia Chapter 5
143
The University of Western Australia Chapter 5
144
The University of Western Australia Chapter 6
6.1 Abstract
Fixed and Floating Offshore structures commonly utilize I-beams as structural components withstanding
distributed loads on their decks or inside hulls. These structural members get damaged due to the corrosive marine
environment leading to a condition in which they need replacement or rehabilitation. Such situations are not
desirable as it will incur monetary losses directly with replacement or repair costs and indirectly through
operational losses due to shut down for hot repair works. A safe and economical alternative for structural
rehabilitation of damaged I-beams is using Fibre Reinforced Polymer (FRP) composites. An experimental
investigation on the feasibility of repair of a heavily damaged I-beam using two different types of FRPs is
presented. The severe damage in the I-beam was artificially introduced by removing both flanges and the web for
300 mm in the mid-span of 1800 mm long I-beam. Four-point bending tests under static loads were performed
until failure of the beam. The first repair was done using carbon fibre reinforced polymers (CFRPs) and the second
one utilized glass fibre reinforced polymers (GFRP). The CFRP repaired specimen showed 277% improvement
from the damaged state whereas the GFRP repair improved 248% in terms of the ultimate strength. A comparison
of the behaviour between CFRP and GFRP repair is also highlighted in the study. Various parameters like stiffness,
ductility, load-displacement behaviour and failure modes of these FRP repairs for damaged I-beams are discussed
in detail. Overall, the results from the study portray the adequacy of an FRP rehabilitation to reinstate the strength
6.2 Introduction
I-beams are commonly used in the offshore industry as structural components both in fixed and floating
platforms carrying distributed loads on their decks or hulls. Due to the extremely corrosive marine environment
these I-beams get damaged. Web and flange sections of an I-beam may get corroded away in such a way that their
structural load carrying ability is critically affected. Replacement or structural strengthening of these damaged I-
beams is subjected to hot works on the platform which are usually not desirable as it will create additional indirect
monetary loss due to operational shutdowns of the platforms. A cost-effective solution for structural rehabilitation
of steel in these situations is by using Fibre Reinforced Polymers (FRPs). This paper aims to experimentally
investigate the feasibility of structural rehabilitation of severely damaged I-beams using FRPs under static loads.
145
The University of Western Australia Chapter 6
FRP composites are fibrous reinforcement embedded in a polymeric matrix. Carbon and glass typical fibres
used in them in association with thermosetting resins such as epoxy or vinylestre [1]. Moreover, they are very
Previous researchers have demonstrated the use of FRP composites to structurally rehabilitate steel
members. Elchalakani and Fernando (2012) showed undamaged steel I-beams can be strengthened using CFRP
[2]. George et al. (2021) had successfully demonstrated the use of FRPs for the repair of corroded tubular offshore
structural members [3]. There is enough evidence from other researchers as well suggesting that the FRP
composites will be able to rehabilitate the corroded steel members [4-6]. However, none of the researchers has
investigated the possibility of rehabilitating severe damage covering both flanges and web in steel I-beams. In this
I-beams used in the experimental study were made with hot rolled steel of grade AS/NZS 1594 - HA250
which guarantees a minimum yield strength of 250 MPa. They were universal beams, 310UB40. Each beam was
cut with a length of 2000 mm so that 1800 mm free span can be achieved in the experimental set up with 100 mm
overhangs on either end. The damage in the beam was modelled as a cut section of 300 mm x 100 mm in the mid-
span as shown in Figure 6.1. The web was completely removed and part of the flanges removed as well. A total of
4 I-beams were used in the study, one damaged, one intact and two rehabilitated.
146
The University of Western Australia Chapter 6
6.3.2 Specimens
Four specimens were part of the study. In order to distinguish the specimens, a naming system was
employed as described below. Damaged I-beams started their name with 'DI' and 'II" was used to name Intact I-
beams. All specimens were subjected to bending which is indicated by the next letter, 'B' in their names. The last
part of the name is indicating the rehabilitation carried out on the specimen. The control specimens were named
with ‘C0’ indicating 0 layers of CFRP. Similarly, ‘C8’ and ‘G8’ for 8 layers of CFRP and GFRP respectively in
the rehabilitated specimens. The tested specimens with their corresponding names are presented in Figure 6.2.
6.3.3 Rehabilitation
Rehabilitation of the damaged I-beams was done using FRP composite products from TechnowrapTM [7].
An E-glass fibre (GFRP) triaxial woven cloth, TechnowrapTM 2K and a carbon fibre (CFRP) quad axial woven
cloth, TechnowrapTM SRS were used for the rehabilitation of the damaged I-beams. The mechanical properties of
the repair products provided by the manufacturer are added in Table 6.1 [8]. The repair length was 700 mm above
the damaged region of the I-beams as shown in Figure 6.3. Firstly, the damaged region was covered using a
composite patch to easily adhere the repair laminates on top. A typical cross-section of the rehabilitation scheme
147
The University of Western Australia Chapter 6
is also presented in Figure 6.3 for more clarity. The repair laminates were not extended over web side 2 and bottom
flange to simulate an actual scenario where some parts of the structural member become inaccessible for
rehabilitation.
Longitudinal Transverse
Product name FVF (%)
E (GPa) εf (%) E (GPa) εf (%)
TechnowrapTM 2K 42 19.78 3 8.26 3.01
TechnowrapTM SRS 40 36.5 1.4 36.01 1.35
The composite patch was fabricated adjacent to the damaged region to develop a similar shape of the I-
beam and contained 4 layers of GFRP. Surface preparation was carried out to achieve a cleanliness of Sa2.5 and a
roughness profile with a minimum of 60 µm. The fabricated composite patch was moved and adhered over the
damaged region to close it out. The repair laminates were applied on top of the patch using wet layup technique
and consolidated using vacuum bagging in multiple stages. Hardness of the repair is checked to reach a minimum
threshold value to ensure the curing of the repair laminates is completed. Different steps in the rehabilitation
process are shown in Figure 6.4. The composition of the repair laminates varied for rehabilitated specimens and is
148
The University of Western Australia Chapter 6
The repair laminates on the rehabilitated specimens were gridded in order to easily visualize large
deformations and localize any failure in the repair. The specimens were gauged with 4 strain gauges. All of them
were placed along the length of the beam at the mid-span or adjacent possible in case of the damaged control
specimen at the center of top flange, bottom flange, web side 1 and web side 2. Four-point bending under gradually
increasing static load was carried out on the test specimens. The specimens were set up in the equipment as shown
in Figure 6.5 with a free span of 1800 mm. The loading beam was placed on rollers separated by 600 mm in the
mid-span. The load applied on the specimen was read by the sensor within the loading equipment. The vertical
displacement of the mid-span of the specimen was measured using in-built displacement gauge of the equipment.
149
The University of Western Australia Chapter 6
The results from the experiments are primarily analyzed based on the load-displacement behaviour which
is presented in Figure 6.6. The load increased mostly linearly with respect to the displacement in the elastic portion
and increased further to reach the ultimate load. Once reaching the ultimate load, the unloading part of the graph
started and it plateaued after reaching a certain level. The general pattern of the load-displacement is consistent
150
The University of Western Australia Chapter 6
One of the control specimens, DIBC0 (damaged I-beam) showed a failure mode as local yielding of the top
flange which is presented in Figure 6.7. The other control specimens, IIBC0 (intact I-beam) showed typical failure
mode of lateral torsional buckling as depicted in Figure 6.8. These failure modes are reflected in the load-
displacement graphs too. The yielding of DIBC0 is observed as a smoothly curved region around the ultimate load
as opposed to the sharp peak for lateral torsional buckling of IIBC0. The elastic slope of DIBC0 is significantly
lower than that of IIBC0 indicting the lack of overall bending stiffness of the beam in the damaged specimen (see
Table 6.3).
151
The University of Western Australia Chapter 6
The overall failure mode in the rehabilitated specimens were like intact I-beam showing lateral torsional
buckling in these specimens. The repair laminates did not indicate any external signs of failure in themselves but
got debonded from the bottom flange as shown in Figure 6.9. The debonding happened around the peak of the
load-displacement behaviour.
The rehabilitated specimens showed better load-displacement behaviour in terms of stiffness and ultimate
strength than the damaged I-beam at the same time they did not reach the intact level either. Moreover, the load-
displacement curve of the rehabilitated specimens showed minute jerks in them especially for DIBC8 indicating
the participation of the FRP laminates in the load sharing and these jerks are corresponding to microscopic failures
within the laminates. Both rehabilitated specimens improved their stiffness from the damaged level as indicated
by the elastic slopes presented in Table 6.3. DIBG8 showed lower stiffness than DIBC8 this can be attributed to
the lower elastic modulus of GFRP in comparison to CFRP. DIBC8 showed stiffness greater than the intact I-beam
152
The University of Western Australia Chapter 6
indicating rehabilitation scheme for this specimen is more than enough for replicating stiffness of the steel lost due
to damage.
Ultimate strengths of the specimens give a clear indication of how the rehabilitation had performed. The
damaged specimen had the lowest ultimate strength due to the large section cut out from this specimen to simulate
a severe damage. The ultimate strength of the intact specimen was 337% of damaged strength. Specimens
rehabilitated with CFRP and GFRP improved by 277% and 248% from the damaged level respectively in terms
of the ultimate load indicating that the rehabilitation had enhanced the structural performance of damaged I-beams.
The Strength Regain Ratio (SRR) was also calculated as per Equation (6.1) which gave an insight into how
the rehabilitation had been able to bring the ultimate strength from the damaged to the intact condition.
_
= (6.1)
_
SRR presented in Table 6.4 indicate that the examined repair schemes were able to reach at least 0.8 times
the intact capacity. SRR for DIBC8 is 0.86 which is higher than the SRR for DIBG8 of 0.80 indicating the
superiority of CFRP repair. The improvement in the ultimate strength of these rehabilitations can be improved
further by increasing the number of layers in the repair laminates. However, such improvements may be very
limited as the ultimate strength was more dependent on the interface properties of the repair laminates with the
steel surface. This fact can be verified by the fact that even though DIBC8 had 40% greater stiffness than DIBG8
153
The University of Western Australia Chapter 6
FRPs by themselves show brittle behaviour and the repairs done using them for steel sections need to be
ductile enough to replicate the behaviour of steel members. The ductility of these specimens can be quantified as
Ductility Indexes (DI) using a modified version of Foster and Attard (1997) [9]. This is an energy-based method
DI = (6.2)
In Equation (2), ABC is the area under the load-displacement curve up to Δ75, the deflection point obtained
by extending the linear elastic part of the curve (0 to 0.75Pu) to peak load (Pu), which is the elastic energy absorbed
by the specimen and ADE is the area under the load-displacement curve up to a mid-span displacement equal to
2Δ75, which is considered the total energy absorbed by the specimen. The method is illustrated in detail in Figure
6.10.
DI calculated for the specimens are tabulated in Table 6.5 along with ABC and ADE values in Joules (J).
The control bare steel specimens show very similar Ductility Indexes (DI) but these are not comparable as both
these specimens have different failure modes. The DI of DIBC8 is higher than both the control cases indicating
the rehabilitation was able to achieve enough ductility. DIBG8 showed a slightly inferior DI than IIBC0. However,
this can be possibly improved with additional layers of GFRP to this specimen.
154
The University of Western Australia Chapter 6
The axial strain values observed on the top and bottom flanges of the specimens are presented in Figure
6.11. The strain values increased as the load increased and decreased as the unloading started. The top flange
strains are compressive (negative) and the bottom flange strains are tensile (positive) but show a similar trend as
the experiment progressed except for the DIBC0 which had the yielded top flange accounting for the out of pattern
strain values.
The axial strain values on the web (side 1 and 2) are presented in Figure 6.12. The control specimens show
increase in compressive strains as the specimen is loaded and decreases as the unloading starts. On the contrary,
the rehabilitated specimens show increase in tensile strains until the ultimate load is reached and as the unloading
progresses the strain decreases and becomes compressive eventually. This indicates that the repair laminates are
acting slightly different manner as opposed to steel to take part in the load sharing.
155
The University of Western Australia Chapter 6
The strain values observed corresponding to the ultimate load are tabulated in Table 6.6. The ultimate strain
on the flanges were the largest for DIBC0 due to the local yielding failure mode of the flange. In case of the rest
of the specimens, the intact I-beam, IIBC0, showed the least followed by DIBC8 and DIBG8 respectively as
expected based on their performance during the experiments. The ultimate strains on the web sides were
compressive for the control specimens and tensile for the rehabilitated specimens. In case of the control specimens,
IIBC0 had slightly higher ultimate web strains. DIBG8 had much larger web and flange strains than DIBC8
showcasing the lower stiffness of glass fibres compared to the carbon fibres.
The ultimate strain values in the rehabilitated specimens are much higher than the allowable strain at failure,
εf defined for the composites by the manufacturers (refer to Table 6.1) and at the same time the experiments did
not see much damage in the composite repair laminates except for the debonding from steel surface. This inevitable
conservatism in definition of the allowable strain for failure of composite sections can be utilized more effectively
156
The University of Western Australia Chapter 6
6.5 Conclusions
The paper demonstrated the successful use of Carbon and Glass Fibre Reinforced Polymers (CFRPs and
GFRPs) for the structural rehabilitation of heavily damaged I-beams. The key conclusions drawn from the study
are as follows:
FRP composites are successful in restoring the structural strength of severely damaged I-beams to at
CFRP repair showed a 277% improvement in ultimate strength as opposed to 248% increase for GFRP
repair.
Overall, CFRP repair showed preferable properties over GFRP repair for structural rehabilitation
6.6 Acknowledgements
The authors would like to acknowledge ICR.IAS Joint Venture for funding this research. The authors would
also like to thank Master of Professional Engineering (MPE) students, Vignesh Sairam Giridhar, Donghyun Kim
and Bowei Sun for assisting with the conduct of experiments at the University of Western Australia.
6.7 References
[1] Saeed, N., 2015, "Composite Overwrap Repair System for Pipelines - Onshore and Offshore Application "
[2] Elchalakani, M., and Fernando, D., 2012, "Plastic mechanism analysis of unstiffened steel I-section beams
strengthened with CFRP under 3-point bending," Thin-Walled Structures, 53, pp. 58-71.
[3] George, J. M., Kimiaei, M., Elchalakani, M., and Fawzia, S., 2021, "Experimental and numerical investigation
of underwater composite repair with fibre reinforced polymers in corroded tubular offshore structural members
under concentric and eccentric axial loads," Engineering Structures, 227, p. 111402.
[4] Bambach, M. R., and Elchalakani, M., 2007, "Plastic mechanism analysis of steel SHS strengthened with CFRP
[5] Elchalakani, M., 2016, "Rehabilitation of corroded steel CHS under combined bending and bearing using
157
The University of Western Australia Chapter 6
[6] Elchalakani, M., Karrech, A., Basarir, H., Hassanein, M. F., and Fawzia, S., 2017, "CFRP strengthening and
rehabilitation of corroded steel pipelines under direct indentation," Thin-Walled Structures, 119, pp. 510-521.
services/technowrap-engineered-composite-solutions.
[8] Kotsikos, G., 2010, "Mechanical property characterisation of WTR laminate," NewRail.Centre for Railway
[9] Stephen, J. F., and Mario, M. A., 1997, "Experimental Tests on Eccentrically Loaded High Strength Concrete
158
The University of Western Australia Chapter 7
7 Conclusion
7.1 Summary
The thesis has presented research on rehabilitation of aged offshore structures using composite materials.
The aim of the thesis was to study the feasibility of using FRP (Fibre Reinforced Polymer) composite materials
for structural rehabilitation of offshore steel structures. This thesis had attempted to lay the ground work for a
framework for structural analysis and design of FRP retrofitted offshore structures. This section summarises the
specific conclusions and outcomes based on each of the objectives of the thesis as defined in Chapter 1.
The 1st objective of the thesis was to compare the performance of an underwater or subsea retrofit to
conventional in air or onshore retrofit in different loading conditions. This was fulfilled with the help of repairing
and strengthening steel tubular members with 2 layers of CFRP utilising both underwater and in-air retrofitting
techniques and testing them under different loading conditions. The following conclusions were drawn based on
Subsea or underwater composite structural repairs can be successfully executed with almost
equivalent performance in the ultimate strength to that of in-air repairs when loaded in concentric
Energy absorption (EA) and Ductility Indices (DI) found for the underwater repairs were slightly
lower than the in-air repairs highlighting the difference in the performance of these two repairs.
The performance of in-air and underwater partial length strengthened specimens were non
distinguishable and it happened because the point of failure for all these specimens was outside
The 2nd objective of the thesis was to experimentally investigate the effectiveness of repairing and
strengthening an offshore structural member using FRP under different loading conditions. Two different
applications of FRP retrofitting were investigated as part of this objective: Repairing and Strengthening. Repairing
was done on specimens with corrosion artificially simulated with machining out of 20% wall thickness or
simulating damage by removing the cross section (in case of I-beams). In the case of the tubular specimens under
159
The University of Western Australia Chapter 7
concentric axial compressive loads, two different specimen sizes (representing leg and brace specimens) were
employed. The key conclusions drawn based on the results from the repaired specimens are as given below.
An observation from the structural tests of the underwater repaired tubular specimens is that no
premature debonding occurs before the attainment of the ultimate load. This reflects good surface
In the case of the leg specimens under concentric and eccentric axial compressive loads, repaired
with 2 layers of CFRP was enough to restore the ultimate capacity for corroded sections with 1.2
mm corrosion in the wall thickness (20% of wall thickness) at mid span to the intact capacity. On
average about 30% improvement in ultimate load was observed for these specimens from the
corroded condition. The energy absorption (EA) of these repaired specimens were better than the
In the case of the leg specimen under bending, repaired with 2 layers of CFRP (CBF2W) was
enough to improve the ultimate strength from 1.2 mm corroded level by 15.1% to restore the
capacity to a level greater than intact capacity. The energy absorption (EA) and Ductility Index
(DI) were better than both corroded and intact levels. Hence, the CFRP repair would behave better
in restoring the structural performance for a member under pure flexural loads than concentric or
eccentric axial compressive loads. This reflects the efficiency of CFRP in resisting tensile loads
Underwater repairing produced the best ultimate strength for the leg specimen under concentric
axial compressive load with 4 layers of CFRP at 40.2% improvement which was more than the
intact capacity. Underwater repairing produced the best ultimate strength for the leg specimen
under bending with 6 layers of CFRP at 32.1% improvement from the corroded capacity.
When the underwater repair was able to match both the compressive strength and bending stiffness
of lost steel due to corrosion with 6 layers of CFRP, unsymmetrical buckling occurred taking the
point of failure just outside the retrofit zone for the specimen under axial compressive load and
points of failure were shifted outside the retrofit zone for the specimen under bending .
The performance of the underwater repair of leg specimens under axial compressive load
improved as the number of retrofit layers increased. At 6 layers of CFRP, the repair had reached
160
The University of Western Australia Chapter 7
its maximum performance and a further increase in repair layers did not have an impact on
Underwater repairing of brace specimens under axial compressive load showed significant
improvement in performance where all the specimens recorded an unsymmetrical buckling with
point of failure outside the retrofit zone indicating the repair was successful in mitigating the
corrosion. All these specimens performed better than the intact brace specimen. The maximum
ultimate load was recorded for the repaired brace specimen with 6 CFRP layers under no eccentric
axial load which represented a 54.9% improvement. The performance of the brace member
improved as the retrofit layers increased and could further improve with more retrofit layers due
to the occurrence of Euler buckling in the brace specimens as opposed to cross sectional yielding
in leg specimens.
The experimental program revealed a successful use of Carbon and Glass Fibre Reinforced
Polymers (CFRPs and GFRPs) for the structural rehabilitation of heavily damaged I-beams. FRP
composites were successful in restoring the structural strength of severely damaged I-beams to at
least 80% of their intact capacity. CFRP repair showed 277% improvement in ultimate strength as
opposed to 248% increase for GFRP repair. Overall, CFRP repair showed preferable material
The second application of FRP retrofitting, i.e., strengthening was done on specimens that are in their intact
condition without any artificial corrosion or damage. Partial length strengthening and full length strengthening
were investigated. Like the repaired tubular specimens under axial compressive load, the partial length
strengthening was also carried out for two different specimen cross sections representing leg and brace members
in fixed offshore platforms. The difference between underwater stricture consolidation and underwater vacuum
consolidation was also studied as part of this objective. The key conclusions drawn based on the results from
Partial length strengthening of leg specimens under concentric and eccentric axial compressive
loads did not produce significant improvement in strength and they all had unsymmetrical
buckling with their point of failure just outside the retrofit zone.
161
The University of Western Australia Chapter 7
Partial length strengthening of leg specimens under bending was found to be very successful with
2 layers of CFRP increasing the ultimate load by 18.8%, but the points of failure had already
shifted outside the retrofit zone. Hence, a further increase in retrofit layers did not result in a
Partial length strengthening of brace specimens under concentric axial compressive load showed
a mean strengthening improvement of 19.1% compared to intact specimens. Brace specimens that
were governed by Euler buckling responded more favourably towards partial length strengthening.
Full length strengthened underwater (stricture) leg specimens under concentric axial compressive
load had shown significant improvement in their strength and other key performance parameters
from the intact levels. The improvement in ultimate load increased by 19.1%, 31.8% and 53.8%
for 2, 4 and 6 layers of CFRP retrofitting respectively. They all exhibited symmetrical buckling
(i.e., point of failure in the mid span of the specimen) as opposed to the unsymmetrical buckling
(i.e., point of failure out of the retrofit zone in the mid span of the specimen) in partial length
strengthened specimens. For the tested leg specimens, the full potential of the strengthening was
only achievable through full length strengthening although it might become an expensive solution
compared to partial length strengthening, mainly for long structural members. Hence the cost
effectiveness of a full length strengthening will have to be carried out on a case by case basis.
Full length strengthened underwater (vacuum) leg specimens under concentric axial compressive
load also showed improved performance from the intact level and mostly had better results than
stricture consolidated specimens. The ultimate load improvements for 2, 4 and 6 layers of CFRP
in these strengthened specimens were 34.1%, 40% and 49.9%, respectively. The novel underwater
The 3rd objective of the thesis was to understand the effect of exposure of FRP retrofitted steel structure to
actual marine conditions. Repaired and strengthened specimens were placed in the Swan River in Western
Australia for a period of one year to conduct the ‘Real Corrosion’ test. The specimens were retrieved from the
162
The University of Western Australia Chapter 7
river after this period and structurally tested under pure bending. Real Corrosion studies produced certain insights
about the exposure retrofitted steel structures in marine conditions as described below.
Real Corrosion results proved that retrofitting layers will prevent steel underneath from coming in
direct contact with water and marine growth thus preventing corrosion.
Repaired specimens after real corrosion exposure showed slightly better structural performance
than non-exposed repaired specimens. This was inferred to happen due to the improvement of
structural properties of the retrofit laminates after constant exposure to external hydrostatic
Strengthened specimens after real corrosion exposure performed almost as good as the non-
exposed strengthened specimens and the variation in performance was due to corrosion of the steel
7.1.4 Numerical and analytical models for retrofitted tubular steel members
The 4th and final objective of the thesis was to develop numerical models and analytical tools to help design
FRP retrofitting for offshore structural members. Simplified Finite Element (FE) models which would simulate
the experimental behaviour of FRP retrofitted steel members in different loading conditions had been presented.
Limited simplified analytical design tools for retrofitted steel tubular members were also developed based on the
experimental results. Key conclusions based on all the numerical and analytical works in this thesis are
A simplified FE model for steel tubular members retrofitted with CFRP under bending and bearing
was developed. The model was accurate enough to predict the load displacement behaviour of
three point bending experiments of CFRP retrofitted steel specimens. On average, there was only
a variation of 4% in the ultimate load predicted by the FE model to that of the experimental results.
A further modified FE model was developed to simulate the experiments performed on repaired
specimens with 2 layers of CFRP under concentric and eccentric axial compressive loads. The
model predicted the behaviour of the experimental specimens with a good accuracy until the
ultimate load but the post ultimate load behaviour was not fully captured due to the simplifications
made in the composite material definition in the model. The ratio between the ultimate load
predicted by the FE model and the experiments ranges from 0.93 to 1.19 with an average ratio of
163
The University of Western Australia Chapter 7
1.00 and a standard deviation of 0.07. Even with the limitations, this FE modelling technique
would help in understanding and designing FRP repairs for steel tubular members in the future.
An analytical check for retrofitted tubular members undergoing axial compressive loads was
presented. The check examined approximately how many layers of retrofit was required to repair
a corroded member back to its original performance or strengthen a member to a certain target
level. With the definition of strength gain and stiffness gain, the analytical check successfully
combination of axial compressive load (N) and bending moment (M) could be withstood by a
repaired steel tubular member was developed. This would be applicable for compact tubular
sections (like the leg specimens) that would fail predominantly due to the yielding of their cross
section.
The research presented in the thesis had advanced the knowledge in rehabilitating aged offshore structures
using FRP composites. There remains a large scope of works that could be done to advance this technology so that
it could be utilised in the offshore industry with more confidence. Main recommended areas for future research
The research presented in this thesis had mostly showed the behaviour of compact structural
members (leg specimens) rehabilitated with CFRP. More experimental studies should be
conducted with the retrofit on slender specimens (like the brace specimens) to understand further
The corroded specimens presented in the thesis are all having 20% wall thickness loss in them.
Larger and more critical corrosion or severe damages could occur in some structural members in
an offshore platform. The effect of larger levels of corrosion or damage in steel tubular members
and the effectiveness of FRP retrofitting to bring back their structural performance should be
The feasibility of using FRP retrofitting for rehabilitating other typical structural elements in an
offshore structure like non-tubular members, plate girders and plates could be studied. The
164
The University of Western Australia Chapter 7
Offshore structural members are under cyclic loads throughout their design life. They could also
be subject to impact loading from dropped objects or vessel impact. Further studies into the effect
of cyclic loading and impact resistance of FRP retrofitted structural members will be very
The real corrosion test presented in this thesis only studied the durability of FRP retrofits for a
period of one year only. Longer term performance of FRP retrofitted steel structures in marine
conditions, above or under water, should be carried out to gain more confidence in this technology.
The current FE models cannot capture debonding of FRP layers from the steel surface or
delamination in between the FRP layers. Finite Element (FE) modelling which would capture more
The current research output has given solid inputs for developing a design framework for
retrofitted offshore steel structural members under the combination of axial compressive loads and
bending moments. More detailed analytical and numerical FE studies based on the experimental
results in this thesis can be carried out to provide the required design guidelines for repair and
Further studies should be conducted to advance this knowledge to retrofitted members under other
types of loading (e.g., tension, shear) and then to study retrofitted steel frame offshore structures
as a whole.
Future research focus should be given into inspection techniques for the retrofit layers and the
steel substrate under the retrofit. This will be very critical in monitoring the performance of these
retrofits and managing risks associated with them effectively. Current available inspection
techniques like microwave scanning, radiography, infrared thermography, etc. are promising
avenues for future research adapting them to into inspection of FRP retrofitted steel structures.
The cost benefit analysis of FRP retrofitting for offshore structural elements should be carried out
to clearly quantify the economic advantage of FRP retrofitting over common repair and
165