You are on page 1of 38

Design of A Fibre-Reinforced Polymer (FRP) Bridge

B. Leo1

University of New South Wales at the Australian Defence Force Academy

Fibre-reinforced polymer (FRP) composite materials are commonly used in the


aerospace, marine and defence industries. Due to their inherent advantages, FRP
composite materials are increasingly making their way into civil engineering
applications. In the field of bridge engineering, FRP composite materials are used
mainly for strengthening and rehabilitation purposes but there are few bridge
superstructures which are designed entirely out of FRP composite materials. This is
mainly due to the high initial costs of FRP composite materials and the lack of design
guidelines and standards. This paper will first review deficiencies in existing bridges
constructed out of conventional materials such as timber, concrete, cast iron and steel.
FRP composite materials offer the potential of being a choice as a structural material
due to their inherent advantages over conventional materials. The current uses of FRP
composite materials in bridge engineering will also be reviewed. This paper will then
propose an innovative design of a FRP bridge, using Strongwell Systems™ pultruded
profiles as the bridge deck and composite FRP-concrete beams as girders. Finally, the
performance of this bridge system will be analysed using finite element modelling with
ANSYS.

Contents
Nomenclature 2
I. Introduction 3
A. Deficiencies in Current Bridges 3
B. Advantages of FRP Materials 4
1. High strength to weight ratio 4
2. Durability 4
3. Green technology 5
C. Current Applications of FRP Materials in Bridges 5
1. Rehabilitation 6
2. Replacement of bridge decks 6
3. Whole of bridge superstructure 8
II. Proposed Bridge Structure 8
A. Bridge Deck 9
B. Girder 10
III. Design Methodology 12
A. Methodology of Design – An Overview 12
B. Step 1 - Determine Design Loads 12
1. Type of loads 12
2. Evolution of traffic loads from an Australian perspective 13
3. Load factors 14
4. Load combinations for Ultimate Limit State (ULS) and Serviceability Limit State (SLS) 14
5. Allowable deflections 14
C. Step 2 - Determine Number of Girders Required 14
1. Input data parameters 15
2. Discussion of results 15
D. Step 3 – Design of Hybrid Beam 16
1. Initial dimensions 16
2. Deflections 17

1
Second Lieutenant (Singapore Armed Forces), School of Aerospace, Civil & Mechanical Engineering,
ZACM 4252/4251 Civil Engineering: Project, Thesis & Practical Work Experience A/B.

1
Final Thesis Report 2009, SEIT, UNSW@ADFA
3. Lateral stability 19
4. Web buckling load 19
5. Strength capacity 20
6. Discussion of results 21
7. Final design of hybrid beam 22
IV. Finite Element Modelling of Bridge Using ANSYS 23
A. Physical Modelling 23
B. Element Types and Real Constants 24
1. Bridge deck 24
2. Girder 25
C. Material Properties 27
1. Girder 27
2. Bridge deck 27
D. Meshing 28
E. Load 29
1. Displacements 29
2. Force and pressure 30
V. Discussion of Results 30
A. Computational Results for a Single Type B Beam 30
B. Mesh Size Control 32
C. Deflections 33
1. Three-layered approach 33
2. Equivalent deck approach 33
VI. Conclusion 34
VII. Future Work and Recommendations 34
Acknowledgements 35
References 36
Appendices 38

APPENDICES
Appendix A. Project Documentation
Appendix B. Matlab (Release 2008) Results
Appendix C. Maple (Version 10) Results

Nomenclature
αs = shear coefficient
δ bend = vertical deflection of girder in z-axis due to bending [mm]
δ shear = vertical deflection of deck in z-axis due to shear deformation [mm]
δ total = total vertical deflection of deck in x-axis due to bending and shear [mm]
∆T = vertical deflection of bridge deck in y-axis [mm]
τbw = web shear buckling stress [MPa]
w = uniformly distributed load (UDL) [kN/m]
A = area [mm2]
b = width of concrete layer and hybrid beam without ±45° GFRP laminate [mm]
c = depth of concrete layer [mm]
d = clear distance of web in pultruded profile [mm]
D = depth of hybrid beam [mm]
Ec = Young’s modulus of concrete [MPa]
Ep = longitudinal (x-direction) Young’s modulus of pultruded profile [MPa]
Eg = longitudinal (x-direction) Young’s modulus of ±45° GFRP laminate [MPa]
El = longitudinal (x-direction) Young’s modulus of CFRP laminate [MPa]
fs = form or shape factor
Gc = shear modulus of concrete [MPa]
Gp = shear modulus of pultruded profile [MPa]
Gg = shear modulus of ±45° GFRP laminate [MPa]
h = depth of pultruded profile [mm]

2
Final Thesis Report 2009, SEIT, UNSW@ADFA
H = height of hybrid beam [mm]
L = span of bridge [mm]
Mext = external moments generated due to loading of beam [Nmm]
Mint = internal resisting moments [Nmm]
P = concentrated/axle load [kN]
Ptriaxle = combination of tri-axle load [kN]
Pwb = web buckling load [kN]
Q = first moment of area [mm3]
t = thickness of CFRP laminate [mm]
tg = thickness of ±45° GFRP laminate [mm]
tw = thickness of web and flange of pultruded profile [mm]
Vu = average shear stress [MPa]

I. Introduction

A material in which more than one constituent is present can be described as a ‘composite’, e.g., reinforced
concrete (steel/concrete) and timber (cellulose fibre/lignin matrix)(Canning et. al, 2007). In the context of
this paper, composite materials are those composed of synthetic fibres and a polymer resin. These composite
materials are formed by embedding continuous fibres in a resin matrix which binds the fibres together. Common
fibres include carbon, glass and aramid fibres and common resin include epoxy, polyester and vinyl ester resins
(Teng et. al, 2003).

The earliest composite materials were glass fibres embedded in polymeric resins and they originated during
World War Two as a result of the burgeoning petrochemical industry (Bakis et. al, 2002). With newer
manufacturing processes and decrease in cost of production, the use of composite materials became more
extensive from 1970s onwards, especially in the defence and aerospace industries. From the late 1980s and
throughout the 1990s, the growth of composite materials in construction was aided by research and
demonstration projects funded from the construction industries and governments. As the need for aggressive
infrastructure renewal became more evident in the developed world, composite materials are becoming more
accepted as a construction material (Bakis et. al, 2002).

Bridges are vital networks of a transport system worldwide as they facilitate the transport of people, goods
and services. However, many of them are in a state of deterioration and in need of urgent repair or replacement.
The growing concern of maintaining these bridges has motivated many researchers to find innovative and cost
effective solutions for repairing or replacing these structurally deficient bridges. Fibre-reinforced polymers
(FRP) are increasingly used for this purpose as an alternative to conventional materials (Kumar et al, 2002). In
fact, the world’s first all FRP composite vehicle bridge can be dated back to 1982, when a single span two-lane
bridge was constructed in China. Glass-reinforced composite box girders of the Miyun Bridge were made from
the hand-lay process and this bridge was constructed by 20 workers within two weeks, assisted by simple
machinery only (McCormick, 1993).

This paper will highlight the deficiencies in existing bridges, and discuss why FRP composite materials are
more attractive than conventional materials in the rehabilitation of structurally deficient bridges and in the
replacement of deteriorating bridge decks. In Chapter 2, an innovative FRP bridge design based on a fictitious
scenario will be proposed. The FRP bridge will be using an ‘off-the shelf’ bridge deck and composite FRP-
concrete beams based on a hybrid concept, combining the high compressive performance of concrete with the
specific advantages of FRP composites developed at UNSW@ADFA (Khennane, 2009). The design
methodology adopted to design this FRP bridge will be presented in a stepped and logical process presented in
Chapter 3. The FRP bridge will be modelled using a finite element software, ANSYS 11.0© and the modelling
process will be outlined in Chapter 4. The computational results obtained with ANSYS will be discussed in
Chapter 5. Lastly, this paper will present the conclusions and recommendations for future work in Chapters 6
and 7 respectively.

A. Deficiencies in Current Bridges


As of 2003, the total number of bridges in the United States is 615 718 and nearly 26.3% of these bridges
are either structurally deficient or functionally obsolete (Alnahhal, Aref, 2008). The United States Federal
Highway Administration has estimated the cost of repairing these deteriorated bridges, and bringing them up to
an acceptable level to be US $110 billion (Aref & Parsons, 2000).

3
Final Thesis Report 2009, SEIT, UNSW@ADFA
In Australia, there are approximately 40 000 road bridges of 7 m span or longer, with a total asset value
estimated at around AS$10 billion. A quarter of these are of timber construction and 60% were built before the
1940s and have been designed for lower load standards than that would be required today (van Erp et. al, 2002).
The original design load has increased significantly from the T44 truck which is based on a 20 tonne vehicle to
the M1600 moving traffic load or the S1600 stationary traffic load which are based on a 72 tonne vehicle (Scott
& Wheeler, 2001). This means that the margins of safety of these timber bridges are unacceptable due to the
increased traffic loading even without considering the state of deterioration.

It is evident from the above statistics that many bridges around the world are experiencing structural
deficiencies and are in urgent need of repair or replacement. According to Karbhari & Zhao (2000), Karbhari &
Seible (2000), the common deficiencies within bridges occur as a result of various factors: (1) aging of
structural components, (2) weather deterioration (de-icing salts, temperature gradients and freeze-thaw actions),
(3) increased traffic ratings from changing traffic patterns, (4) insufficient detailing at time of original design
based on older standards, (5) problems associated with dynamic loads (wind and seismic activities), and (6)
inadequate maintenance and rehabilitation measures taken through the life of the bridge structure.

It is evident there is a growing need for the widening of bridges to accommodate greater flow of traffic or
heavier vehicles at higher speed since deteriorated bridges can have a great impact on society in terms of
economic losses (Karbhari & Zhao, 2000). Rehabilitation and replacement of bridges can be done using
conventional materials but FRP materials are becoming more attractive in bridge applications. This is due to
their superior material properties and it has been shown that designing FRP materials in conjunction with
conventional structural materials can achieve technical efficiency within competitive economical constraints
(van den Einde, et al, 2003).

B. Advantages of FRP Materials


The composites industry has grown approximately 460% over the past 30 years, from 360 000 tons in 1970
to 1.68 million tons in 2000 (Alnahhal, Aref, 2008). FRP materials are widely used in aerospace and maritime
industries as those industries have recognised that FRP materials have superior material properties over
conventional materials in their field of application. However, in a predominantly conservative construction
industry, particularly in the case of bridge engineering, the high strength to weight ratio and durability of FRP
materials are two important advantages which FRP materials have over conventional materials.

1. High strength to weight ratio


FRP materials display high strength-to-weight and
high stiffness-to-weight ratios in comparison with most
metals and alloys (Karbhari & Zhao, 2000). This means
that FRP composite structures offer the potential to be
significantly lighter than structures made from
conventional materials. Designers can now develop
designs at lower weights and these weight savings can
equate to other forms of cost savings in terms of
transportation and labour costs. In civil infrastructures,
weight savings could also result in various advantages
such as better seismic resistance, ease of application of
reinforcing plates during rehabilitation and a decrease in
need for large foundations (Karbhari & Zhao, 2000). In
addition, the drive to increase traffic ratings means that
Figure 1. Carrying a composite beam to
there is a huge potential to replace older and deteriorated
demonstrate its light weight (Garden, 2004)
bridge structures with FRP materials since weight savings
from FRP materials can improve the live load capacity
without the expense of new support structures and approach works (Heldt et al, 2004). Figure 1 shows the light
weight nature of a composite beam manufactured by researchers Taylor Woodrow and Dragados (Garden, 2004).

2. Durability
The long term durability of FRP materials is often stated as the main reason for the use of these materials.
As such, they have been successfully used in the aerospace, automotive and maritime industries. In general,
FRP materials do not corrode and have a better chemical resistance to the environment. However, it must be
noted that the durability of these materials depends intrinsically on the choice of constituent materials, methods
of manufacturing and fabrication, and the surrounding environmental conditions during their service lives
(Karbhari, 2003). Only with the development of appropriate factors of safety from validated data of FRP

4
Final Thesis Report 2009, SEIT, UNSW@ADFA
materials applied in a civil infrastructure environment, can we realise the potential of low maintenance and low
life cycle costs of FRP materials (Karbhari, 2003). This factor plays an important role in the decision making
process of choosing a suitable construction material. The whole life cost of a bridge should be examined instead
of just considering the initial cost of a construction material since FRP materials have higher initial costs but
might be a cheaper option in the long run due to lower maintenance costs. However, the long term durability
and the whole of life cost studies are not considered in this paper.

3. Green technology
As a result of global warming, the environmental impact of the built environment and the carbon content of
construction materials has become a very important issue. Concrete is one of the most widely used construction
materials but the manufacture of concrete contributes to the global output of greenhouse gases. It has been
found that the production of Portland cement, an important constituent of concrete produces a significant
amount of carbon dioxide, a greenhouse gas. One tonne of Portland cement produces about one tonne of carbon
dioxide and other greenhouse gases (Naik, 2008).

A study was done to compare


the energy consumption of
using various materials such as
steel, aluminium, composites
and reinforced concrete for
construction. Figure 2 show
that composites consume
approximately half the total
amount energy as any other
construction material
considered in the above study
(Daniel, 2003). The study of
energy consumed was
Figure 2. Energy Consumption of Construction Materials (Daniel, 2003)
conducted from the stages of
extraction of raw materials, production and fabrication of the material, delivery of the material to the
construction site and maintenance throughout the design life of the structure. In this manner, FRP material
promises to be a greener construction option over concrete and steel.

C. Current Applications of FRP Materials in Bridges


Conventional materials have a number of advantages, namely the relatively low cost of materials and a long
history of application through tried construction methods. However, in some cases, these materials lack
longevity and are susceptible to rapid deterioration. Some cases of design are constrained by limitations of the
material used, e.g., weight constraints in a clear span concrete bridge. In other cases, some conventional
materials are deemed as ineffective in terms of functionality. In all of such (and other) cases, there is a need for
emerging new materials and technologies that are functional and efficient (Karbhari & Zhao, 2000).

According to Aref & Parsons


(2000), Holloway (2003) and
Karbhari (2004), current
applications of FRP materials in
bridges can be broadly grouped
into three categories: (1)
rehabilitation schemes (retrofit,
repair and strengthen), (2)
replacement of deteriorated bridge
decks, (3) design and construction
of new composite bridges. Figure
3 shows the various applications
of FRP materials within the
Innovative Bridge Research and
Construction (IBRC) Program
conducted by the United States
Federal Highway Administration.
About one quarter of the bridges
Figure 3. FRP applications in IBRC program Bridges (Harries, 2008) involve the use of FRP as an

5
Final Thesis Report 2009, SEIT, UNSW@ADFA
externally bonded structural element and another quarter of the bridges use FRP decks as replacement for
existing bridge decks on steel girders. This shows that although FRP materials have already been used
extensively within the bridge industry, the majority of their use has been limited to rehabilitation and the
replacement of bridge decks.

1. Rehabilitation
The term ‘rehabilitation’ in this paper refers to the restoration or enhancement of functionality or safety of
existing concrete structures. Therefore, terms such as repair and strengthening, are covered within the scope of
rehabilitation (Karbhari & Zhao, 2000). In the European
Union, an annual budget of £215 million has been
allocated to rehabilitate 84 000 existing reinforced and
pre-stressed concrete bridges. In the UK, more money is
spent on maintenance and upgrading of existing
reinforced concrete, steel and cast iron structures as
compared to the construction of new bridges. In the US,
for a 7 year period between 1990 and 1996, 17% of the
Federal-aid highway funds were spent on rehabilitation,
whereas only 4.5% was spent for construction of new
bridges (Holloway, 2003). The above statistics show that
in most of the developed countries, there is an urgent
need to develop cost effective and durable rehabilitation Figure 4. Adhesive bonding of pultruded
methods since a large portion of bridge funds are CFRP strips to underside of slab (Karbhari
allocated to rehabilitation. and Seible, 2000)

One common method of strengthening of deteriorated bridge decks involves the addition of epoxy bonded steel
plates to the soffit of the bridge deck. Although this method is simple and has been extensively used in Europe,
it suffers from a number of disadvantages that include difficulty in handling and problems relating to durability
and corrosion (Karbhari & Seible, 2000). Generally, the weight of the steel plates makes transportation and their
placement difficult. Jacks, winches, cranes and extensive scaffolding are needed to install these steel plates
while elaborate and expensive falsework is required to maintain the steel plates in position during bonding. This
method is also often associated with the uncertainties of durability and corrosion effects (Hollaway, 2003).

On the other hand, FRP materials provide significant flexibility on site when they are applied to the decks
through methods such as adhesive bonding, wet lay-up and resin infusion. These methods of application are
usually less labour and machinery intensive, hence, providing an efficient and easy mean of strengthening the
bridge deck without interrupting traffic (Karbhari & Seible, 2000).

The external application of FRP materials to bridge decks can result in increased load carrying capacity, flexural
and shear strength of the original deck when conducted in an appropriate manner (Karbhari, Seible, 2000).
However, it must be noted that the efficiency of such external reinforcement depends on the appropriate
selection of FRP material based on stiffness, strength requirements and the integrity of bond between the
concrete and composite (Hollaway, 2003). Nevertheless, this application of FRP strips has already became
commonplace for the rehabilitation of bridge decks in several western European countries, and in certain parts
of North America (Meier, 2001). Figure 4 shows the ease of application of prefabricated CFRP strips to the
soffit of a slab. No form of jacks, winches or extensive scaffolding are required to install them.

2. Replacement of bridge decks


A bridge deck is defined as a structural element that transfers loads transversely to supports such as
longitudinal running girders. The bridge deck requires the most maintenance of all elements in a bridge
superstructure as it is affected by most of the bridge deterioration problems. One of the major problems is the
degradation of the wearing surface itself. In addition, there may a need to increase the load ratings and number
of lanes on the existing bridge to accommodate increasing traffic flow or conform to new bridge codes
(Hollaway, 2003). The loss of time and resources associated with rehabilitation of the bridge deck has provided
a significant impetus for the development of new materials that are durable, light and easy to install.

FRP decks that are commercially available at the present time can be classified according to two types of
construction—sandwich and adhesively bonded pultruded shapes. In general, both types of FRP constructions
possess the inherent advantages in higher strength and stiffness per unit weight compared to conventional steel-
reinforced concrete decks and hence present the opportunity for rapid replacement of the existing deck and
reduction in overall dead load of the structure (Bakis et al, 2002).

6
Final Thesis Report 2009, SEIT, UNSW@ADFA
Sandwich structures have been widely used in aerospace
and marine applications. They usually consists of strong,
stiff face sheets that carry the flexural loads and a low-
density, bonded core material that separates the face sheets
and ensures a composite action of the deck. The design for
varied depths and deflection requirements allows for the
flexibility in the manufacturing of the face sheets and core
Figure 5. (a) Generic foamed core (b) materials (Bakis et al, 2002). Face sheets of sandwich decks
Kansas Structures Composites Inc. (KSCI) are primarily composed of E-glass mats rovings infused
corrugated core (Bakis et al., 2002) with a polyester resin while current core materials are rigid
foams or thin walled cellular FRP materials as shown in
Figure 5 (a).

Honeycomb core sandwich panels as shown in Figure 5(b) have been used to improve the structural
performance of FRP decks. The honeycomb core configuration is made up of a sinusoidal wave pattern, and it
extends vertically between the top and bottom face sheets. The geometry of this sandwich structure is designed
to improve the buckling response and stiffness of the deck. A few analytical studies based on the energy or
series method were conducted on the KSCI deck system but these analytical solutions are too complicated for
routine design office applications (Davalos et al, 2001; Salim et al, 2006). Other researchers have also adopted
simplified finite element modelling methods to analyse the KSCI bridge deck system (Machado et al, 2008; Cai
et al, 2009). For example, in the study conducted by Cai et al (2009), finite element modelling of a small FRP
honeycomb panel is used to develop equivalent orthotropic properties of the panel. The results from his study
will provide a more convenient procedure to investigate the KSCI bridge deck performance. However, the KSCI
deck system will not be considered in this paper.

The second type of bridge deck


construction uses assemblies of adhesively
bonded pultruded shapes. Such profiles or
sections can be economically produced in
high volume through the pultrusion process.
The pultrusion process is a highly
automated process, consisting of pulling
resin impregnated unidirectional fibres,
various continuous strand mats and
stitched fibre fabrics through a heated
curing die at speeds of up to 3 m/min
depending on the size and complexity of Figure 6. Schematic diagram of the pultrusion process
the structural shape (Bakis et al, 2002). A (Wu, 2006)
schematic diagram of the pultrusion
process is shown in Figure 6. The pultruded profile’s constituents (type of fibres and resin used) and their
orientation can be altered to achieve design flexibility, and the desired section properties. The cross sectional
shapes are varied to a lesser extent due to the potentially high cost of the pultrusion dies. The die can be
designed to produce various shapes, such as I-, C- and box
sections. In addition, relatively large and complex thin
walled-profiles can be manufactured by the pultrusion
process since the equipment can be built to accommodate
large sections. However, the pultrusion process has some
constraints regarding fibre volume fraction and maximum
thickness of the FRP profile. Fibre volume fraction is
seldom over 45% and the maximum thickness of the FRP
profile sections is limited to about 12.7 mm (Wu, 2006).

Several decks designs have been developed by various


manufacturing companies and each of them has advantages
in terms of stiffness, strength and field implementation.
Figure 7. FRP decks produced from adhesively The four current commercially available pultruded decks
bonded pultruded shapes: (a) EZSpan; (b) are shown in Figure 7. They are (a) EZSpan (Atlantic
Superdeck; (c) DuraSpan; (d) Strongwell (Bakis Research Corp); (b) Superdeck (Creative Pultrusion Inc.);
et al, 2002) (c) DuraSpan (Martin Marietta Composites) and (d)

7
Final Thesis Report 2009, SEIT, UNSW@ADFA
Strongwell Deck (Strongwell). Extensive full scale laboratory and field experimental tests have been conducted
on these composite deck systems and their performance, such as displacement, strain response, load carrying
capacity and failure modes are discussed in the state of art reviews by Cheng & Karbhari (2006) and Bakis et al
(2002).

Although FRP decks promises a huge potential to solve the existing problems of deteriorating bridge decks, the
success of FRP decks depends on the accuracy and simplicity of the analytical tools which are developed to
design their applications. Bridge engineers have to adhere to design guidelines established within design
standards and codes when designing steel, concrete or timber bridges. In contrast, engineers have to use
advanced tools such as finite element methods to analyse the bridge deck in a bridge superstructure when using
FRP decks. Therefore, a simpler and accurate analysis procedure is still needed to predict the response of FRP
decks in a bridge superstructure (Salim et al, 2006).

3. Whole of bridge superstructure


The use of FRP materials has so far been well accepted in the areas of rehabilitation and replacement of
bridge decks as mentioned earlier. Figure 3 shows about less than 5% of the IBRC bridge program consists of
‘all-FRP superstructure’ and this represents the current trend in bridge engineering in most developed countries.
The development of new structural concepts and systems entirely out of FRP materials or a combination of FRP
materials and other conventional materials provides great potential for huge advances in civil infrastructures.
However, the application of FRP materials in this area is still within the realm of demonstration projects or field
testing rather than in the commercial area (Karbhari and Seible, 2000). There are a number of factors which are
limiting the use of FRP materials in new bridge construction projects. Some of these factors are: (1) high initial
costs in comparison with components fabricated from conventional materials, (2) an absence of appropriate
design guidelines, codes and standards and (3) insufficient field data on the long term structure performance and
durability of FRP materials (Karbhari, 2004).

The proposed design of a FRP bridge in this paper will fall within this category, demonstrating the combination
of superior mechanical characteristic of tension in FRP materials with the dominant characteristic of
compression in concrete.

II. Proposed Bridge Structure


Legais is a fictitious island country state located 30 nautical miles east of Bateman’s Bay, NSW, Australia.
Recently, the Road Authority of Legais has approached UNSW@ADFA for assistance in the construction of a
bridge as part of its future highway network. As neither concreting facilities nor steel fabrication plants are
readily available in Legais, the only method of constructing this bridge is to transport prefabricated bridge
components from Australia, via land and sea. However, this is not economically feasible as using conventional
materials will incur high transportation and labour costs.

In fact, this is a true scenario to most of the current timber bridges within Australia which are located in remote
regions without concreting facilities. Transportation cost of prefabricated concrete structures to the outback
regions will be very high and hence, pultruded FRP profiles can be a viable option. Hence, UNSW@ADFA has
proposed the design of a FRP bridge that consists of an ‘off-the shelf’ FRP bridge deck and hybrid FRP-
concrete beams as girders.

z The FRP bridge will be designed


over a span of 8 m and a width of 7
m with two traffic lanes. The width
of the bridge was chosen based on
8m y the design traffic lane specified in the
AS 5100 where the width of a design
lane is taken to be 3.2 m. An
additional 300 mm is allocated to
x each traffic lane to cater for space
7m
required to install the road safety
barriers and road shoulders. The FPR
bridge is modelled using CATIA as
shown in Figure 8. In this paper, the
following Cartesian coordinates are
used: (1) x-direction represents the
Figure 8. Model of FRP bridge using CATIA
8
Final Thesis Report 2009, SEIT, UNSW@ADFA
direction along the span of the bridge (also known as the traffic direction, (2) y-direction represents the
transverse direction along the width of the bridge and (3) z-direction represents the vertical direction along the
thickness of the bridge deck. It is very important that the above sign conventions are adhered to throughout the
design process of this FRP bridge. This will be more evident in the subsequent chapters.

The ‘off-the shelf’ bridge deck selected will be a Strongwell deck system produced by Strongwell™. The
girders will be designed based on a hybrid beam concept that combines high compressive strength concrete with
FRP materials. The design and analysis of each component will be explained in the following sections.
Adhesive bonding will be used as connections between the bridge deck and girder.

A. Bridge Deck
The ‘off-the shelf’ bridge deck chosen in this paper is
manufactured by Strongwell™. Strongwell™ is one of
the world’s leading pultruders of FRP structural shapes.
It has been pultruding FRP structural products since 1956.
It has more than 66 pultrusion machines and 647 000
square feet of manufacturing space in three plant
locations in the US. Strongwell™ manufactures and
fabricates FRP products for a wide variety of markets and
specific applications such as pultruded gratings on
offshore platforms, driveshaft components for trucks and
innovative solutions for bridge applications. Strongwell's
customers include Fortune 500 industrial and commercial Figure 9. Strongwell bridge deck on Tangier
firms, major architectural and engineering firms, leading Island Bridge, Virginia Department of
contractors and distributors, and many other companies - Transportation (Strongwell, 2009)
both large and small - in a variety of local and overseas
markets (Strongwell, 2009). Figure 9 shows one of the
Strongwell’s product applications in bridge decking.

Researchers at Virginia Tech have been evaluating the


performance of a multicellular modular FRP deck
system produced using components pultruded by
Strongwell. The proposed bridge deck used in this Figure 10. Cross Section of Strongwell Bridge
paper is adopted from their design. It consist of Deck
Extren® 625 series pultruded shapes: 9.5 mm flat plates
for the top and bottom layers and 152.4 mm x 152.4 mm x 9.5 mm square tubes for the centre core as shown in
Figure 10. The section properties of the flat plate and the square tube are easily available on the Strongwell’s
Design Manual provided by the company on their website. The section properties are shown in Table 1 and
Table 2. The pultruded shapes are made of E-glass roving and continuous strand mat embedded in polyester
resin. The top and bottom plates and square tubes are adhesively bonded to form full bridge-width preassembled
deck panels. Steel rods, with diameter 25.4 mm, are used to provide the necessary binding forces during the
curing process of the deck panels. Bonding is then accomplished in a vacuum bag to produce uniform pressure
and continuous bonding (Liu et al, 2008).

The reasons for the choice of this particular bridge deck design include: (1) the section properties of each
pultruded profile are readily available in the design manual provided by Strongwell, (2) an ‘off-the shelf’ bridge
deck design using pultruded shapes will reduce the overall cost of construction and (3) the orthotropic properties
of the square tube assembly and an equivalent deck have been calculated and published by Liu et al (2008). This
information is used for calculations in Chapter 3 and set up of preliminary parameters within ANSYS in Chapter
4.

Table 1. Section properties of square tube (Strongwell, 2009)

9
Final Thesis Report 2009, SEIT, UNSW@ADFA
Table 2. Section properties of flat plate (Strongwell, 2009)

B. Girder
The proposed bridge girders used in this paper will be based on a hybrid beam concept that combines the
superior mechanical properties of FRP materials and the high compressive strength of concrete. In order to
explain the concept of a hybrid beam design approach, the conventional reinforced concrete beam design will be
examined and be used as a starting reference.

A conventional reinforced concrete beam cross section is shown in Figure 11(a) with the steel bars encased
within the concrete. When the beam is under flexural loading, the strain (assuming linear strain variation over
depth of beam) across the beam profile is shown in Figure 11(b), with compressive strains in the top and tensile
strains at the bottom of the beam. However, the effective load carrying elements of the reinforced beam is
shown in Figure 11(c), with the concrete above the neutral axis taking the compressive forces, Fc and the steel
bars below the neutral axis taking the tensile forces, Ft. The tensile strength of concrete is very low and the
portion of concrete below the neutral axis is often ignored when we determine the ultimate load carrying
capacity of the beam. Thus, it can be concluded that the concrete in the tensile zone does not contribute directly
to the overall carrying capacity of the beam, and its main functions are to hold the steel bars in place and protect
the steel bars from corrosion.

The two main disadvantages of the reinforced concrete beam are the possibility of corrosion of the steel bars
and the high self weight of the beam. Prevention methods such as replacement of steel bars with epoxy coated
galvanised or stainless steel bars can be very costly and of limited use. The concrete in the tensile zone can be
viewed as inefficient and wasteful from a natural resources perspective since it does not contribute to the overall
load carrying capacity of the beam and yet contributes about 75-80% of the self weight of the beam (van Erp et
al, 2002).
The proposed hybrid beam section will eliminate both of the inherent disadvantages. The main components of

Fc
NA

Ft

(a) Reinforced (b) Strain diagram (c) Effective beam section (d) Hybrid beam section
concrete beam

Figure 11: Hybrid beam concept


the hybrid beam consist of high strength concrete, a pultruded FRP section and additional carbon FRP
underneath the tensile flange to increase the stiffness of the beam as shown in Figure 11(d). The hybrid beam
concept combines both conventional and high performance materials to create a highly optimised beam that is
able to save up to 2/3 the weight of the reinforced concrete beam and eliminates the problem of corrosion (van
Erp et al, 2002). This concept of combining FRP composites with concrete is not new and has been trialled by

10
Final Thesis Report 2009, SEIT, UNSW@ADFA
Deskovic et al. (1995), Correia et al. (2007), Hulatt et al. (2003), van Erp et al. (2005), Canning et al. (1999) and
Ribeiro (2002). These hybrid beam designs are shown in Figure 12.

Deskovic et al. (1995), van Erp


et al. (2005) and Ribeiro et al.
(2002) used almost similar
designs, where the hybrid
beam consisted of a glass fibre
reinforced plastic (GFRP) box
Mechanical section, a concrete block cast
onto the top flange and a thin
connector
carbon fibre reinforced plastic
(CFRP) laminate strip bonded
Deskovic et al. Correia et al. Hulatt et al. at the bottom flange. In their
designs, the CFRP strip
mimics the role of the yielding
of tensile steel in reinforced
concrete beams, providing a
warning of imminent collapse.
However, one of the problems
of such a design is that the
thickness of the CFRP
laminate is limited since it is
Ribeiro et al. designed to fail first. This
van Erp et al. Canning et al. means that the CFRP strip is
usually very thin and
GRFP CFRP Concrete Foam
contributes very little to the
Figure 12. Existing hybrid FRP-concrete beam designs overall stiffness of the hybrid
beam. Hence, the existing
problem of low stiffness of
GFRP profiles is not solved.

As a result of their inherent lack of stiffness, these hybrid beams have to be put side by side when they are used
in a bridge. This is shown in Figure 13. van Erp et al (2002) adhesively bonded two sections of seven hybrid
beam sections together with a strong CFRP
laminate in the transverse direction in order
to achieve adequate stiffness. The first fibre
composite bridge built in Toowoomba,
Australia, uses such a design (van Erp et al,
2002). However, this design is not very
economical and efficient as a large amount Figure 13. Cross section of 2.5m bridge width section (van
of concrete is still being used. Erp et al, 2002)

In order to address the above problem, a new hybrid concrete-FRP beam design is proposed by UNSW@ADFA
(Khennane, 2009). The new proposed beam design will consist of a GFRP pultruded profile, a high strength
concrete block, CFRP strip laminate and an outer GFRP laminate wrap with fibres orientated at ±45°. The
CFRP laminate is no longer used as a warning of imminent failure and hence it can contribute significantly to
the overall stiffness of the hybrid beam. The much sought after ‘pseudo-ductility’ is now obtained through the
cracking of the high strength concrete instead of the CFRP laminate (Khennane, 2009).

According to Khennane (2009), there are four features of this hybrid beam design that will allow for longer
spans of bridges, increased girder to girder spacing and shallower sections. These four features of the hybrid
beams are as follows: (1) the GFRP pultruded profile offers high resistance to lateral torsional buckling; (2) the
use of HSC provides the warning signs of failure; (3) the use of CFRP laminate to increase the stiffness of the
beam and (4) the use of an outer GFRP wrap to provide some form of confinement to the concrete block-
pultruded profile system and also to increase the shear strength of the beam section. The detailed design of this
hybrid beam will be discussed in Chapter 3.

11
Final Thesis Report 2009, SEIT, UNSW@ADFA
III. Design Methodology
As mentioned earlier in Chapter 1, there are a few factors that limit the choice of using FRP materials as a
construction material. One of the factors is the lack of design standards and codes. Standards and codes are
necessary to control risk in matters of public safety and without them, it is unlikely that FRP materials will be
used as a construction material beyond the scope of research and demonstration projects (Bakis et al, 2002).
Conventional construction materials such as steel are successful in this manner in the sense that steel is well
specified by grade. There are well established design standards, such as the AS 3600: Concrete structures and
AS 4100: Steel structures, that specifies the design of such structures. This assists the bridge designer to choose
a suitable steel section for the bridge superstructure structure based on the properties of that steel section
(Karbhari and Seible, 2000).

Currently, the “Canadian Highway Bridge Design Code 2006” has a section on the use of FRP materials for
rehabilitation applications only. For example, standards and guidelines have been developed by the Canadian
Department of Transportation for the use of FRP jackets in seismic retrofitting of columns. Similar documents
are either published or being written in Japan, the United States and Europe but none of these documents
provide a sound methodology for designing a whole of FRP bridge superstructure.

Hence, this paper will aim to design a FRP bridge using a stepped and logical process. However, there will be
several references made to the current Australian Standards (AS) 5100: Bridge Code, which is applicable only
for bridges constructed out of steel, concrete and timber. It must be noted that the AS 5100 will be used as a
starting point since there are no standards or guidelines that outlines the use of FRP materials in construction of
a new bridge.

A. Methodology of Design – An Overview


The design procedure of the FRP bridge will be divided into three distinct steps with obtainable deliverables
after each step. Step 1 is to determine the design loads. Step 2 is to estimate the number of girders required. Step
3 is to design a highly optimised girder based on the hybrid beam concept proposed by Khennane (2009). The
bridge superstructure will then be modelled using ANSYS, and this will be discussed in Chapter 4.

B. Step 1 - Determine Design Loads

1. Type of loads
Step 1 is perhaps the most important step within the design procedure presented in this paper. There are
many types of loads that a bridge must support in addition to traffic loads. These design loads include: (1) the
self weight of the structure (also known as dead load), (2) vehicle weights, (3) horizontal vehicle loads due to
braking, centrifugal force or horizontal surging effects in some vehicles, (4) dynamic vertical loads caused by
dynamic interaction between primary service vehicles and the bridge (influenced by factors such as road
roughness and vehicle suspension characteristics), (5) wind loads, (6) collision forces caused by either a service
vehicle striking the structure or by some moving object beneath the bridge, (7) earthquake loads and (8) thermal
effects (Connor and Shaw, 2000).

AS 5100.2 has categorised the above design loads or loads effects into (1) permanent effects (PE), (2) thermal
effects and (3) transient effects. The PEs considered in this paper are those arising from the structure’s dead load
and additional permanent loads (otherwise known as superimposed dead loads). The thermal effects will not be
considered in this paper and only vehicular traffic loads will be considered as part of transient effects. The three
load effects considered are as follows:

• Dead loads shall be considered as the weight of the parts of the structure that are structural elements and
any non-structural elements that are considered unlikely to vary during construction and use of the structure.
The dead load is the self weight of the bridge deck and the hybrid beam girders.

• Superimposed dead load shall be considered as the weight of all materials forming loads on the structure,
which are not structural elements and will vary during construction and use of the structure. The
superimposed dead load will be the self weight of a 50 mm thick concrete asphalt overlay as a road wearing
surface.

• The three nominal traffic loads are taken from AS5100.2: Design loads (2004) and AS5100.7: Rating of
existing bridges (2004). They are: (1) S1600 stationary traffic load, (2) M1600 moving traffic load and (3)
T44 truck load.

12
Final Thesis Report 2009, SEIT, UNSW@ADFA
2. Evolution of traffic loads from an Australian
perspective
The loads applied to Australian bridges have been
increasing at a rate of 10% per decade since the
beginning of this century (Heywood et al, 2000). The
T44 truck load, based on a 44 tonne truck, is a
previous version of the Australian bridge design load
and has been in place since 1976 (AS 5100, 1992). A
T44 truck load model is shown in Figure 14. However,
weigh-in-motion studies have shown that the T44
loading approximates an average extreme daily event
(i.e., the average of the largest event of the day) for
short-span structures, and has certain short-comings in
addressing the future trend towards longer and heavier
vehicles. As a result, two heavier families of vehicles
have been developed and they are known as L58 (0.58
tonne/m3) and S73 (0.73 tonne/m3) vehicles
(Heywood et al, 2000). From the L58 and S73
families of heavy vehicles, two traffic load models are
Figure 14. T44 truck load (AS 5100.2, 2004) developed in the current AS 5100, the M1600 moving
traffic loading and S1600 stationary traffic loading.

The M1600 moving traffic model as shown


in Figure 15 presents a solution to simulate
the effects induced by the S73 and L58
vehicles in a moving traffic stream. The
traffic model features 4 x 360 kN tri-axle
groups and a 6 kN/m uniformly distributed
load (UDL) that continues under the vehicle.
The 6 kN/m UDL is designed to represent
the moving traffic stream that accompanies
a very heavy combination vehicle. There
can be a variable spacing between the front
Figure 15. M1600 moving traffic load (AS 5100.2, 2004)
and rear vehicle. The variation of this
distance will not be considered since the
span of the proposed bridge is only 8 m and the length of a single heavy vehicle is already 8.75 m. The S 1600
stationary traffic load model is shown in Figure 16. It consists of 66% of the truck components of the M 1600
and a UDL of 24 kN/m. Like the M 1600, the S 1600 have a total gross weight of 1600 kN at a loaded length of
25 m.

A summary of the magnitudes of the axle loading, P and UDL, w


Axle load, P UDL, w for all the 3 traffic loading models is included in Table 3.
(kN) (kN/m)
S1600
stationary
240 24
traffic
load
M1600
moving
360 6
traffic
load

T44 truck Figure 16. S1600 stationary traffic load (AS 5100.2, 2004)
96 12.5
load

Table 3. AS 5100 design load


conditions

13
Final Thesis Report 2009, SEIT, UNSW@ADFA
3. Load factors
AS 5100.2 specifies the various load factors which are applied to the nominal PEs and traffic loads. These
factors are summarised in Table 4.
Type of load factor Symbol ULS SLS
Dead load (DL) γD 1.2 1.0

Superimposed dead load (SDL) γSD 2.0 1.0

Traffic load γT 1.8 1.0


Table 4. Load factors for the various load cases for ULS and SLS

4. Load combinations for Ultimate Limit State (ULS) and Serviceability Limit State (SLS)
For a thorough and complete design process, all of the design loads mentioned by O’Connor and Shaw
(2000) and all the load combinations stated in AS 5100 should be considered. However, I have considered three
of the load effects, neglecting thermal effects, wind, earthquake and impact loads etc. Clause 22 in AS 5100.2
specifies the various load combinations for ULS and SLS designs. The load combinations which I have chosen
are as follows:

• ULS Load = PE (DL x γD +SDL x γSD) + ultimate traffic loads

• SLS Load = PE (DL x γD +SDL x γSD) + serviceability traffic loads

The ultimate traffic load is obtained by multiplying nominal traffic loads by γT = 1.8 and the serviceability
traffic loads is obtained by multiplying nominal traffic loads by γT = 1.0. The nominal traffic loads are the
values of P and w shown in Table 4.

5. Allowable deflections
Clause 6.11 in AS 5100.2 states that the deflection limit of a road bridge under traffic for SLS design shall
be appropriate to the structure and its intended use, the nature of the loading and the elements supported by it.
The deflection for SLS under live loading shall not be greater than L/600. This deflection limit will be compared
with throughout the design process but it must be noted that the L/600 deflection criterion is derived for the
global bending of the entire bridge superstructure and generally governs for steel structures only. It is also
known that reinforced concrete decks currently do not pose a stiffness problem and hence, forcing a FRP deck
or girder to match the stiffness of a reinforced concrete or a steel girder is extremely conservative (Cassity et al,
2002).

C. Step 2 - Determine Number of Girders Required


Step 2 involves finding out the number of girders required for this proposed bridge. A unit length of the
bridge deck across the width of the bridge is first considered. This is to ensure the selected bridge deck behaves
like a simple beam instead of a plate. The strip of the bridge deck is shown in Figure 17.

1m

Figure 17. A strip of bridge deck

I have selected a cross section of six square tube lengths for consideration since it is closest to a unit length and
the pultruded square tubes are of fixed dimensions. The equivalent elastic modulus of this section (six square
tubes with two flat plates) is obtained from the Strongwell Design Manual and second moment of area of the
bridge deck is calculated in order to determine the deflection of the bridge deck, ∆T under traffic loading. The
deflections are calculated using the direct stiffness method, also known as the matrix stiffness method. This
method of structural analysis is particularly suited for computer-automated analysis and a Matlab code was

14
Final Thesis Report 2009, SEIT, UNSW@ADFA
written to calculate the deflections for this simple beam. The output of the Matlab code, showing the values of
∆T at various chosen points along the deck is presented in Appendix B.

The direct stiffness method is a matrix method that makes use of the beam element members' stiffness relations
for computing member forces and displacements. In applying this method, the beam deck is modelled as a set of
simpler, idealised elements interconnected by nodes. The material stiffness properties such as the elastic
modulus and second moment of area of these elements are then compiled into a single matrix equation which
governs the behaviour of the entire idealised structure. The structure’s unknown displacements and forces can
then be determined by solving the matrix equation.

1. Input data parameters


The strip of bridge deck is idealised as a continuous beam with the hybrid beam girders considered as simple
supports as shown in Figure 18. A total of nine girders are first assumed and the spacing between each girder is
875 mm. The following steps are taken to ensure that the input parameters represent the true physical problem
(Appendix B(i)):
P, Axle load

w, UDL

∆T
Figure 18. Beam analysis using direct stiffness method

• The whole beam system is divided into 20 elements connected with 21 nodes. The nodes are selected at
the positions of the girders, positions where P is applied and the midpoint distances between each girder to
girder.
• The elastic modulus, Ey is calculated to be 7.52 GPa (Liu et al, 2008). The second moment of area, Iy (six
square tube assembly with top and bottom plates) is calculated to be 210 x 106 mm4. These mechanical
properties are assigned to all 20 elements.

• Constraints are applied to all girders which are treated as simple supports, i.e. the displacements of
bridge deck in the vertical direction at the girders are fixed at zero but the bridge deck is allowed to rotate.

• Four axle loads, P are applied at appropriate distances according to the design traffic models shown in
Figure 15 and Figure 16. The UDL, w is also applied across the whole beam.

2. Discussion of results
The largest deflections of the bridge deck, ∆T obtained from the two traffic load conditions, M 1600 and S
1600 are 0.97 mm and 0.66 mm respectively (see Appendix B (ii) and (iii)). The largest deflections are located
at nodes 9 and 13 which are the locations where the middle two axle loads, P, are applied. The calculated
deflections, ∆T for both traffic load conditions when compared with the allowable deflection L/600 ≈ 11.7 mm,
are very small. This indicates there can be a possibility of reducing the total number of girders required.

Since the obtainable deliverable for this step is to optimise the number of girders required for the proposed
bridge, a logical thing to do is to reduce the number of girder such that ∆T is closer to the allowable deflection.
However, it must be noted that the lesser the number of girders, the larger the load each girder will have to
support, assuming that the traffic load effect is distributed across all the supporting girders. This means that the
girder has to be designed with a deeper section and reinforced with a thicker CFRP laminate or concrete section
to achieve sufficient flexural stiffness. A deeper beam section or concrete thickness will increase the self weight
of the beam. A thicker layer of CFRP laminate will increase the overall cost of the girder due to the high cost of
CFRP laminates. In other words, a compromise has to be achieved between the number of girders used and the
design of the girder. The choice in the number of girders will affect the subsequent design process since the
design loads in Steps 2 and 3 will change. With the above consideration, a total of nine girders are selected for
this proposed bridge structure with a girder to girder spacing of 875 mm.

15
Final Thesis Report 2009, SEIT, UNSW@ADFA
D. Step 3 – Design of Hybrid Beam
In reinforced concrete beam design, the accepted practice is to first design the beam section based on ULS
principles (typically some form of load and resistance factor design) to ensure sufficient strength capacity of the
beam. SLS such as deflections and crack width under fatigue, serviceability or sustained loads are then checked.
Although the serviceability criteria are usually applied after the ULS in reinforced concrete beam design, the
relatively lower elastic modulus of GRFP materials means that serviceability criteria will usually control the
design of the hybrid beam. In other words, the design of a hybrid FRP-concrete beam will be deflection driven.
This is an important concept as majority of the design process is checked with the deflection limit established in
AS 5100.

Step 3 involves the design of the hybrid beam which will be done mainly using Maple. The principles behind
the equations are derived from Chakrabortty (2009). There are various checks on the adequacy of the hybrid
beam section within this step and these will be explained in the following sections. Only the important equations
are shown in the main report while the detailed equations and calculations are shown in Appendix C.

1. Initial dimensions
The hybrid beam consists of:

• High strength concrete (HSC) with thickness c and width b,


• GFRP pultruded box sectional profile of height h with uniform thickness tw,
• CFRP laminate strip of thickness t, and
• Outer GFRP wrap of thickness tg with fibres orientated at ±45°.

The cross section of this hybrid beam is shown in Figure 19. The overall height of the hybrid beam is H and
width is b2. As a starting point, certain initial dimensions of the hybrid beam are chosen based on
recommendations for a reinforced concrete beam while other dimensions are chosen via trial and error so as to
allow for the design process to continue.
• According to Warner et al (1998), the reinforced concrete beam’s span to depth ratio varies between 10
and 20, i.e. L/H = 10 ~ 20. The lower limit of 10 is chosen so that a deeper beam can be achieved and the
problem of a lack of stiffness of a thin walled GFRP box section can be addressed. Thus, for a span of 8 m,
the depth of the hybrid beam, H is 800 mm.

• The concrete block should not be too thin otherwise it might fail prematurely through punching failure. It
should be thick enough to be able to take compressive stresses and contribute to the overall stiffness of the
beam. However, it should not be too thick otherwise the self weight of the beam will increase greatly since
concrete is the heaviest constituent in this beam. By trial and error, the ratio of the thickness of the
concrete layer to the depth of the GFRP pultruded profile is chosen to be 4, i.e. c = h/4. Since c + h = H, c
= 160 mm and h = 640 mm. Therefore, the GFRP pultruded profile will have a depth, h = 640 mm and a
width, b = 300 mm. The thickness of the GFRP pultruded profile is set at, tw = 10 mm since this is limited
by the pultrusion process. The width of the pultruded profile is first trialled at 300 mm and this dimension
will be verified under the lateral stability check later.

• The CFRP laminates which are available at UNSW@ADFA have a thickness of 1.2 mm. As discussed
earlier in Chapter 2, the thickness of the CFRP should not be limited in the sense that the CFRP is
designed to fail first. Hence, the CFRP laminates should be used to increase the overall stiffness of the
section. However, it should be noted that the CFRP laminates are very expensive and to achieve a highly
optimised and economical hybrid beam design, a balance between cost and performance have to be met.
Therefore, an initial thickness CFRP laminate layer, t = 6 mm is trialled.

16
Final Thesis Report 2009, SEIT, UNSW@ADFA
• An outer GFRP laminate wrap of thickness, tg = 0.9 mm is achieved by filament winding of the beam
with GFRP laminates orientated at ±45°. The role of this outer laminate wrap is to provide some form of
confinement to the FRP-concrete system and ensure enhanced composite action. With the fibres orientated
at ±45°, the outer laminate wrap can also increase the overall shear capacity of the hybrid beam.

b
tg

c GFRP Wrap
y1

tw

NA GFRP Pultruded
H Profile

HSC
h
d YN

CFRP Laminate

b2

Figure 19. Cross section of hybrid beam (Type A)

The properties of the different materials used in the above hybrid beam are as follows (Chakrabrotty, 2009):
• Ec = Young’s modulus of concrete = 36640 MPa
• Ep = Longitudinal (x-direction) Young’s modulus of pultruded profile = 22400 MPa
• Eg = Longitudinal (x-direction) Young’s modulus of ±45° GFRP laminate = 22400 MPa
• El = Longitudinal (x-direction) Young’s modulus of CFRP laminate = 140000 MPa
• Gc = Shear modulus of concrete = 15267 MPa
• Gp = Shear modulus of pultruded profile = 2890 MPa
• Gg = Shear modulus of ±45° GFRP laminate = 6940 MPa

2. Deflections
In order to calculate the deflections of the hybrid beam under loading, certain section properties of the
hybrid beam need to be determined. The flexural stiffness, D of the above hybrid beam can be calculated from
Equation (1).

bc 3 c
D = ∑ EI = E c [ + bc (t g + t + h + − YN) 2 ]
12 2
3
bt t
+ E p [ w + bt w (t g + t + h − w − YN) 2 ]
12 2
3
bt t
+ E p [ w + bt w (YN − t g − t + h − w ) 2 ]
12 2

17
Final Thesis Report 2009, SEIT, UNSW@ADFA
t wd3 d
+ 2E p [ + dt w (YN − − t w − t − t g ) 2 ]
12 2
3
b2t g tg
+ Eg[ + b 2 t g (H − − YN) 2 ]
12 2
3
b2tg tg
+ Eg[ + b 2 t g (YN − ) 2 ]
12 2
t g (H − 2t g ) 3
H − 2t g
+ 2E g [ + t g (H − 2t g )( + t g − YN) 2 ]
12 2
bt 3 t
+ EL[ + bt (YN − t g − )] (1)
12 2
where YN is the position of the neutral axis measured from the bottom of the hybrid beam,
1
YN = *
E c bc + 2E P bt w + 2E P d t w + 2E g b 2 t g + 2E g t g (H − 2t g ) + E l bt
c
[E c bc ( + h + t + t g )
2
t t d
+ E p bt w (h + t + t g − w ) + E p bt w (t + t g + w ) + 2E p t w d ( + t w + t + t g )
2 2 c
tg tg H
+ E g b 2 t g (H − ) + E g b 2 t g ( ) + 2E g (H − 2t g )t g ( )
2 2 2
t
+ E L bt (t g + )] (2)
2
There are two types of beam theories, namely the Timoshenko and the Euler-Bernoulli models. In Euler-
Bernouli beams, transverse shear stress is neglected as it is assumed the beam in bending behave in such a way
that the cross section normal to the neutral axis will remain normal to the neutral axis after bending. However,
in the case of Timoshenko beams, the cross section of beam which is normal to the neutral axis does not remain
normal after bending and transverse stresses are taken into account. This is especially important when analysing
composite beams with deep sections. FRP composite beams with relatively low shear modulus undergo
appreciable amount of shear deformation under load. Therefore, the shear-deformable Timoshenko beam model
is considered in this step when calculating the overall deflection of the hybrid girder. The overall deflection of
the hybrid beam can be obtained by Equation (3).

δ total = δ bend + δ shear (3)

where δ bend is the deflection due to bending in a four point loading case and δ shear is the deflection due to
transverse shear effects.

δ bend and δ shear are obtained by the Equations (4) and Equation (5) respectively.

23 PL3 Pa (3L2 − 4a 2 )
δ bend = or (equation applies only for four point loading case) (4)
648 EI 24 EI

Pa
δ shear = ( )f s (5)
∑ GA
where GA is the shear stiffness of the beam, ‘a’ is the distance of P from the support of the beam and f s is the
shear or form factor, which is a variable dependent on the shape and size of the beam.

18
Final Thesis Report 2009, SEIT, UNSW@ADFA
In order to consider the effects of shear deformation, Timoshenko made use of the differential equation of the
deflection curve which included a term containing the shear coefficient, αs. This shear coefficient is the ratio of
the shear stress at the neutral axis of the beam to the average shear stress. However, shear deflections obtained
by multiplying αs to the average shear stress is based upon shear strains at the neutral axis which does not take
into account the variation of in-shear strains throughout the height of the beam.

Timoshenko used the principle of virtual work and the unit-load method as a more accurate approach to find the
beam deflections caused by the effects of shear. Through this approach, a form factor for shear, f s is introduced
to replace αs. f s is dependent only upon the cross-sectional dimensions of the beam and it is known for certain
cross sectional shapes, for e.g. f s = 6/5 for rectangular beams. However, since the hybrid beam cross section is
not a standard shape, its form factor has to be obtained from Equation (6).

A Q2
I 2 A∫ b i2
fs = dA (6)

The hybrid beam is broken down into various 8 components – (1) concrete, (2) GFRP top flange, (3) GFRP
webs, (4) GFRP bottom flange, (5) GFRP wrap (top flange), (6) GFRP wrap (webs), (7) GFRP wrap (bottom
flange) and (8) CFRP laminate. The first moment of area, Q and area, A are calculated for each of the 8
components.

The general equation to calculate the form factor for this cross section is given by Equation (7). A more detailed
expansion of Equation (7) is provided in Appendix C (i).

2 2 2 2
A Qc Q p1 Q p2 Q p3
2 ∫
fs = [ 2
dA + ∫ 2 dA + ∫ 2 dA + ∫ 2 dA
I A b A
b A
b A
b
2 2 2 2
Q g1 Q g2 Q g3 Q
+∫ dA + ∫ dA + ∫ dA + ∫ L2 dA] (7)
A
b2 A
b2 A
b2 A
b

3. Lateral stability
In general, pultruded box sections have high resistance to lateral torsional buckling. However, limits are
often imposed on the section’s height to width ratio to prevent lateral instability and also to satisfy other design
requirements (Deskovic, 1995). The beam’s height to width ratio is often checked by the following equation, (c
+ h) / b < k, where k is a constant to be chosen by the designer. Deskovic (1995) has recommended a value of k
= 3 for composite sections and thin walled hybrid sections. I will adopt the same number as Deskovic (1995).
The initial trailed width, b = 300 mm gives a value of k = 2.67 which is well within the accepted value.

From the above relation, if a lower value of k is chosen by the designer, the width of the beam has to increase
and this will again contribute to the overall stiffness of the beam. Therefore, a study on the optimum value of k
can be conducted since the above relation governs the initial dimensions of the beam design. However, the
relation between the choice of k and width of the beam will not be discussed further.

4. Web buckling load


Another possible failure of the beam is through buckling of the webs of the pultruded box section. This type
of failure mechanism has been reported by Charonenphan et al (2004), where the beam undergoes large global
flexural deformation and local cell deformations, and the failure occurs along the corners of the box profile.

To ensure that the webs of the hybrid beam is stiff enough to take the load and not buckle, the theoretical web
shear buckling stress is calculated by Equation (8) and Equation (9) which are given by Holmes and Just (1983).

3
4K wb 4 D L D T
τ *
bw = , for θ > 1 (8)
t wd2

19
Final Thesis Report 2009, SEIT, UNSW@ADFA
4K wb D T H wb
τ *bw = , for θ < 1 (9)
t wd2
DLDT
where θ = (10)
H wb
3
1 Gwtw
H wb = (ν L D T + ν T D L ) + (11)
2 6(1 − ν L ν T )
3 3
E wL t w E wT t w
DL = and D T = (12)
12(1 − ν L ν T ) 12(1 − ν L ν T )
and the value of Kwb depends on θ, which is provided by Holmes and Just (1983) in Table 5. The values of EwL,
EwT, νL and νT are obtained from Chakrabortty (2009).

θ 0 0.2 0.5 1.0 2.0 3.0 5.0 10.0 20.0 40.0


Kwb 18.6 18.9 19.9 22.2 18.8 17.6 16.6 15.9 15.5 15.3
Table 5. Values of θ versus Kwb

Shear buckling occurs when the average shear stress equals the shear buckling stress. The average shear stress is
given by Equation (13).

Vu = 2 t w d τ *bw (13)

The theoretical web buckling load is obtained after calculating the average shear stress. The web buckling load
is given by Equation (14).

Vu L
Pwb = (14)
a
5. Strength capacity
The last check will be to calculate the strength capacity of the hybrid beam. This is done by ensuring that the
internal resisting moments is larger than the external moments generated in the beam as a result of loading. As
discussed in Chapter 2, the CFRP laminate in this hybrid beam is not designed to fail first to provide the
warning signs of imminent failure. Instead, the concrete layer is designed to fail first to have the ‘pseudo-
ductile’ failure behaviour. Thus, the ultimate failure strain of the HSC is considered as εc = 0.0026. Upon
loading up to this ultimate failure strain, Figure 20(a) shows the actual failure stress distribution. At the ultimate

αf c '

c γc Fc
y

Fp1

Fp2

t FL
(a) (b) (c)

Figure 20. (a) Original stress distribution, (b) Simplified stress distribution, (c) Location of forces

20
Final Thesis Report 2009, SEIT, UNSW@ADFA
failure strain, the stress distribution shape of the compressive concrete above the neutral axis is not linear and
adopts a curve. The height of the neutral axis as represented by the dotted line in Figure 20 is calculated in
Equation (2).

Warner et al (1998) explains how the rectangular stress block approach is used within AS 3600 to approximate
the ultimate stress curve in concrete at ultimate failure. A single parameter, γ is used to define both the
magnitude and the location of Fc. However, in the current AS 3600, the value of γ is limited to concretes with
compressive strengths, fc’ not in excess of 50MPa. The HSC achieved by UNSW@ADFA has a high
compressive strength of 85 MPa which is beyond the limit specified in AS 3600 (Khennane, 2009). Therefore, I
have adopted the approach taken in the New Zealand Standards for high compressive concrete. The depth of the
rectangular stress block is given by Equation (15) and the width of the uniform stress block is given by Equation
(16). The simplified rectangular stress block is shown in Figure 20(b). These equations are likely to be adopted
by the next version of the Australian standard.

γ = 0.85 − 0.008( f c '−30) (15)

α = 0.85 − 0.004( f c '−55) (16)

To calculate the internal resisting moments, moments can be taken about the point in which, force Fc is acting
upon. The individual force components as shown in Figure 20(c) are as follows,

2
1 Egb t w εc
Fp1 = (E g b t w ε c (h + t − YN − t w ) + + E g t w ε c (h + t − YN − t w ) 2 (17)
c + h + t − YN 2
2
1 Egb t w εc
Fp2 = (E g b t w ε c (YN − t − t w ) + + E g t w ε c (YN − t − t w ) 2 (18)
c + h + t − YN 2

1 E b t 2ε c
FL = (E L b t ε c (YN − t) + L (19)
c + h + t − YN 2
The internal resisting moment capacity of the beam is

h + t − YN) 2 t
M int = −Fp1 ( + y) + Fp2 ( (YN − t) + h + t − YN + y) + FL ( + h + y) (20)
3 3 2
6. Discussion of results
A few assumptions are made in Step 3 for calculation of deflections.

A four point loading case was considered instead of applying the axle loads, P at their exact positions as per the
S1600 and M1600 traffic load models (see Figure 15 and Figure 16 respectively). In other words, I have
combined all of the three axle loads and applied them as Ptriaxle with a distance of 3750 mm apart as shown in
Figure 21. This is to facilitate a quicker beam analysis since I can use an established deflection equation to
calculate the maximum deflection in the midpoint of the beam. Also, this approach is more conservative than
applying the axle loads, P at their actual positions since I have combined all the axle loads and apply them at the
nearest possible distance from each other. This will result in the worst case scenario in terms of deflection in the
midpoint of the beam for a four point loading case.
Ptriaxle Ptriaxle

3750 mm
a a

8000 mm

Figure 21. Four point loading case for Step 3


21
Final Thesis Report 2009, SEIT, UNSW@ADFA
Another assumption was made during the calculation of the dead load of the structure. Since the dimensions of
the hybrid beam is not yet finalised, I have used the self weight of a Type B beam in the calculation of the dead
load of the structure. A Type B beam design is shown in Figure 22.

Deflections are obtained using Equations (3), (4) and (5) for both P and w. The deflections, web buckling load
and internal resisting moments for Type A beam are summarised in Table 6.

δ total for P (mm) δ total for w (mm) Web buckling load Internal resisting
(kN) moment (Nmm)
S 1600 5.43 8.63
1430 1.86 x 109
M 1600 11 4.07
Table 6. Theoretical results for Type A Beam
The deflections, δ total for P and w are smaller than the allowable the allowable deflection, L/600= 13.3 mm.
Hence a Type A beam will satisfy the deflections limit specified by AS 5100.2. The calculated web buckling
load is 1430 kN which is above the applied load, Ptriaxle for both S1600 and M 1600 loading conditions.
However, the buckling of the web is one of the possible failures and for a pultruded box section which is 640
mm deep, a web thickness of 10 mm is considered very slender. Hence, I have decided to proceed on with the
design of Type B beam. This will be discussed in the next section.

The external moment resulting from the applied force can be obtained from the following equation, Mext = Ptriaxle
x a. The largest Ptriaxle obtained from the M1600 load case is considered and Mext is calculated to be 2.6 x 108
Nmm. The internal resisting moment, Mint is calculated to be 1.86 x 109 Nmm which is larger than the external
moment, Mext. This shows that the hybrid beam has a high amount of strength since it has a larger moment
resisting capacity compared with the external moments generated from the applied loads. This further confirms
the design philosophy of FRP materials being driven by the serviceability criteria.

7. Final design of hybrid beam


Another beam design was considered in Step 3. The single pultruded box section, used in a Type A beam is
segmented into three sections as shown in Figure 22. This design will now be known as Type B beam.

b
tg

c
y1 GFRP Wrap

tw
NA
d1 GFRP Pultruded
Profile
H

d2 h HSC
YN

d3 CFRP Laminate

b2
Figure 22. Cross section of hybrid beam (Type B)

22
Final Thesis Report 2009, SEIT, UNSW@ADFA
Segmenting the pultruded box section into 3 sections is analogous to the addition of web stiffeners in I-beams in
order to reduce the possibility of web buckling.

The neutral axis, YN, flexural stiffness, D and form factor, f s of Type B beam has to be re-calculated using the
same method of analysis as Type A beam. Since the shape of the pultruded profile in a Type B beam is more
complicated than the Type A beam, the GFRP pultruded profile has to be divided into smaller components
during the analysis in Step 3. This results in longer equations and more complicated calculations. The set of
equations and calculations for Type B beam is presented in Appendix C (ii).

The deflections, web buckling load and internal resisting moments for Type B beam are summarised in Table 7.

δ total for P (mm) δ total for w (mm) Web buckling load Internal resisting
(kN) moment (Nmm)
S 1600 5.11 8.23
4432 1.88 x 109
M 1600 10.36 3.62

Table 7. Theoretical results for Type B Beam

Since the deflections, δ total in Type B beam are smaller than the deflections in Type A beam, Type B beam will
satisfy the deflections limit specified by AS 5100.2. The web buckling load is 4432 kN which is well above the
applied load, Ptriaxle for both S1600 and M 1600 loading conditions. The web buckling load has increased by
approximately 3 times just by segmenting the single box section into 3 segments. This can be expected because
according to Equations (8) or (9), the web shear buckling stress, τ bw
*
will increase when the value of d, the clear
distance of the web, decreases. In addition, this relationship between τ bw
*
and ‘d’ is not linear since the term ‘d’
is squared. The three-fold increase in web buckling will reduce the possibility of web buckling failure in the
hybrid beam but there is a limitation to the number of segments the pultruded profile can be divided into. The
limitation of segmenting the box sections into smaller sections is caused by the increased difficulty in producing
a suitable die to pultrude the profile.

The Type B beam has adequate strength capacity since the internal resisting moment is higher than the external
moments generated from the applied loads.

Through the various checks of the beam in Step 3, the initial trialled dimensions of c and t are found to be
adequate. The hybrid beam design is finalised after Step 3 with the Type B beam design and the modelling of
the proposed bridge superstructure within ANSYS explained in Chapter 4.

IV. Finite Element Modelling of Bridge Using ANSYS

A. Physical Modelling
The whole of the bridge superstructure is modelled within ANSYS by creating a 2D-model made up of lines
and areas. To model the deck, eight rectangles are created. One such rectangle, 875 mm by 8000 mm is shaded
in grey as shown in Figure 23. These eight rectangles will now be known as deck rectangles. The 10 other blue
rectangles, measuring 400 mm by 250 mm, represent the actual positions of the axle loads, P in accordance with
the M 1600 and S 1600 traffic load models. These 10 rectangles will now be known as axle load rectangles. The
nine parallel lines running in the x-direction represent the hybrid beam girders and they are spaced 875 mm
apart from each other.

23
Final Thesis Report 2009, SEIT, UNSW@ADFA
z

8000 mm

7000 mm

Figure 23. Physical modelling of bridge structure

The deck and axle load rectangles can be created using a Boolean, ‘copy’ function with the appropriate offset
distances for the newly generated rectangles. The additional keypoints and lines generated within ANSYS by
the ‘Copy’ function are suppressed. This can be done using the ‘Merge item’ function under ‘Numbering
controls’ after all eight rectangles are generated. This will allow the user easier computational handling of the
model in ANSYS since there will be no overlapping lines or keypoints in the same dimensional space.

One of the major advantages of setting out the physical model as such is that the nine parallel lines which will
be meshed with beam elements, are also part of the deck rectangles. These deck rectangles will then be meshed
with shell elements. This will address the issue of setting nodal constraint equations to both the beam and shell
elements. However, this type of modelling technique assumes a rigid connection between the soffit of the deck
and the top of the girders which is highly dependent on the quality of the adhesive bonds or mechanical
connectors that are used to connect both structural elements.

B. Element Types and Real Constants


1. Bridge deck
I have selected to use the SHELL 91 element in ANSYS to model the bridge deck. SHELL 91 is a nonlinear
layered structural shell that can be used for layered applications of a structural shell model or modelling thick
sandwich structures. The element has six degrees of freedom at each node - translations in the nodal x, y and z
directions and rotations about the nodal x, y and z axes. The geometry, node locations and coordinate systems
for a SHELL 91 element are shown in Figure 24. Although SHELL 99 element is usually more efficient than
SHELL 91, SHELL 99 element is a linear layered structural shell that does not have some of the nonlinear
capabilities of SHELL 91. Also, when building a model with less than three layers, SHELL 91 is more efficient
than SHELL 99 (ANSYS, 2009).

The SHELL 91 element is defined by eight nodes, layer thicknesses, layer material direction angles and
orthotropic properties. Real constant model 1 is established for the SHELL 91 element. This allows me to define
the number of layers, the material type of each layer, the thickness of each layer and the orientation of the layers
for the bridge deck. Figure 25 shows a screenshot taken from ANSYS. Material models 5, 6 and 5 are applied to
the top plate, middle square tube assembly and bottom plate respectively. The choice of material models will be
explained in the next section. The thickness of each layer is also applied to each layer. The top and bottom
plates have a thickness of 9.5 mm each while the square tube assembly has a thickness of 152.4 mm. Since all
the layers have a constant thickness in the z-direction, only TK(I) needs to be input.

24
Final Thesis Report 2009, SEIT, UNSW@ADFA
The orientation of all layers and hence, theta is set to zero. This is true only if the sign convention stated earlier
in Chapter 2 is used consistently throughout the analysis and the appropriate material properties in the x, y and z
directions are correct for the plates and the tube assemblies.

Figure 24. SHELL 91 Geometry


(ANSYS, 2009)
Figure 25. Setting up real constant for SHELL 91
Element
2. Girder
The hybrid girders are modelled using BEAM 189 element in ANSYS. BEAM 189 is a 3D quadratic finite
strain beam element suitable for analysing slender to moderately thick beam structures. This element is also
based on Timoshenko beam theory and shear deformations are included. The element has six degrees of
freedom in each node, translations in the nodal x, y and z directions and rotations about the nodal x, y and z axes.
Beginning with ANSYS Release 6.0, newer versions of ANSYS will ignore any real constant data for the
BEAM 189 element. Instead, all real constant data are being replaced by the ‘SECTYPE’, ‘SECDATA’,
‘SECOFFSET’, ‘SECWRITE’ and ‘SECREAD’ ANSYS commands.

One important function that the BEAM 189


element provides is that it can be used with
any user defined beam cross section. This
allows the user to create any cross sectional
HSC profile of a beam and assign different material
models to represent the different constituents
of a composite beam. In other words, the user
will have greater flexibility in the choice of
materials and in designing the shape of the
composite beam. Another important function
GFRP which BEAM 189 element provides is the
Pultruded offset function from a chosen location. This
Profile function allows the beam to be offset from a
line when the line is meshed using BEAM
189 elements. This offset function will be
demonstrated later when the lines are meshed
with beam elements.

A 2-D cross sectional profile of the beam is


created using rectangles and the Boolean,
‘Subtract’ function. The cross sectional
CFRP profile of type B beam generated in ANSYS
Laminate is shown in Figure 26. Again, additional
keypoints, lines and areas generated from
using Boolean function should be merged and
Figure 26. Type B beam cross sectional profile suppressed for easier computational handling.

25
Final Thesis Report 2009, SEIT, UNSW@ADFA
Before applying the ‘SECWRITE’ command or using the equivalent ‘Write from area’ function in the ANSYS
GUI interface to save the 2-D model as a custom beam, the size of the mesh can be manually controlled. If the
size of the mesh is not user defined, ANSYS will automatically specify the size of the mesh. The advantage of
controlling the mesh size is the computational ease of handling the graphics generated after meshing a line with
BEAM 189 elements. The finer the mesh, the longer the mesh takes to appear on screen and the harder it is to
rotate or pan the view on screen. This problem becomes more evident when all nine lines are meshed in the final
model.

The element sizes of the lines highlighted in yellow as shown in Figure 26 are set at 75 mm while the rest of the
other lines are set at 30 mm. Although a coarser mesh is preferred, there is limit to the maximum element size
length. This is due to the small thicknesses of the pultruded profile shape (10 mm thick webs and flanges) and
the CFRP laminate in comparison with the dimensions of the concrete layer. Thin layers and profiles are more
problematic to mesh with large mesh sizes. Sometimes, a ‘shape violation limits’ warning sign will be output by
ANSYS if the mesh sizes are unsuitable. The mesh sizes have to be manually defined again with a suitable mesh
size.

HSC

GFRP
Pulrtuded
Profile

CFRP
Laminate (b)
(a)

Figure 27. Cross section mesh (a) Fine mesh (b) Coarse Mesh

Figure 27 shows a (a) fine mesh and (b) coarse mesh of the Type B beam cross sectional profile after applying
the ‘SECWRITE’ command to the 2-D solid model. The fine mesh is made up of 10 mm x 10 mm elements
while the coarse mesh is made up of elements ranging from 30 mm to 75 mm. The coarse mesh has been chosen
over the fine mesh for the subsequent analysis since it allows for easier computational handling. Also, same
deflections in a single beam analysis are obtained for both the fine and coarse meshes. This shows that the
fineness of the mesh when using the ‘SECWRITE’ command will not affect how ANSYS computes the section
properties of the beam. The user created cross sectional profile of the Type B beam is then saved as a custom
beam section which is further edited by assigning material properties.

Next, specific material models are assigned for each element in the section mesh to represent the actual material
in the Type B beam. This is done using the ‘Edit/Built-up’ function in the ANSYS GUI interface. If the user
does not define the material model for the elements in the section mesh, ANSYS will assign Material Model 1
to all the elements by default. I have assigned Material Model 1 to the concrete layer, Material Model 2 to the
GFRP pultruded profile and Material Model 3 to the CFRP laminate layer. This can be checked by plotting the
material numbers which will be colour coded as shown in Figure 27. The choice and assigning of the material
models will be discussed in the next section. The edited composite beam profile is then saved as a custom beam
profile again.

Lastly, the Section_ID of a beam can now be created using the ‘SECREAD’ command or the equivalent ‘Read
section mesh’ function in the GUI interface. This allows the user to specify the name, the positional offsets and
recall the saved custom beam profile of the beam. The created Section_ID of the beam is recalled again during
the meshing of the model.

26
Final Thesis Report 2009, SEIT, UNSW@ADFA
C. Material Properties
A total of six material models are used in this computational analysis and they are summarised in the
following tables and figures. Densities of the materials are included in all six materials models so as to account
for the effects of the structure’s self-weight in the analysis.

1. Girder
In order to build up the section mesh and assign different material types to the relevant elements of the Type
B beam using the ‘SECWRITE’ command as discussed earlier, Material Models 1,2 and 3 are used to represent
the HSC, GFRP pultruded profile and CFRP laminate respectively as shown in Table 8.

Material Component E, Young’s Modulus ν, Poisson’s Ratio Density (kg/m3)


Model (MPa)
1 HSC 36 640 0.2 2400
2 GFRP Pultruded Profile 24 400 0.3 1960
3 CFRP Laminate 140 000 0.3 1570
Table 8. Material models 1, 2 and 3

The computed section properties of the Type B


beam are shown in Figure 28. The equivalent
flexural stiffness, D of the Type B beam can be
computed using ‘SECREAD’ command and this
can be checked against the theoretical D
calculated from Step 3 in Chapter 3. The flexural
stiffness of the composite beam will determine the
extent of deflections in a beam according to
Equation (4). As explained earlier, the design of
FRP structural shapes are driven by deflections. I
have assumed elastic, isotropic material models to
represent these materials. This assumption can be
valid since the computed flexural stiffness of 1.83
x 1014 Nmm2 is similar to the theoretical flexural
stiffness up to 3 significant figures. However, the
GFRP ±45° wrap is not modelled within ANSYS
and this will affect the computed section
properties of the beam. The reason why the GFRP
wrap is not modelled in ANSYS will be explained
in Chapter 7.

2. Bridge deck
The layers within the decks are each modelled as
orthotropic plates. Both approaches of analysing
the Strongwell bridge deck by modelling it as
three layers and an equivalent elastic orthotropic
deck have been used by Liu et al (2008). The
latter approach has also shown accurate results
while retaining simplicity for practical use since
only a single layer is used. Figure 29 shows the
Figure 28. Section properties of Type B beam
material properties for the top and bottom plate,
computed using ANSYS
the calculated equivalent material properties of the
square tube assembly and the equivalent
orthotropic plate properties of the deck. I have adopted the values obtained by Liu et al (2008) when assigning
materials properties to the bridge deck. The materials properties of the equivalent deck approach is assigned as
Material Model 4, the top and bottom plates as Material Model 5 and the square tube assembly as Material
Model 6. These material models representing their respective layers are recalled when defining the real
constants of a SHELL 91 element as shown earlier in Figure 25. Special attention must be given to the use of
consistent sign convention throughout the whole analysis to ensure that the correct Young’s modulus, shear
modulus and Poisson’s ratio values are assigned.

27
Final Thesis Report 2009, SEIT, UNSW@ADFA
In Liu et al (2008) computational analysis,
the deck was modelled with SHELL 93
elements and the steel girders are modelled
with BEAM 188 elements as shown in
Figure 30. In the initial physical model, the
line which is to be meshed with BEAM 188
elements is already offset in the vertical
direction below the deck. Additional
constraint equations are then applied to the
beam element nodes and the shell element
nodes directly above them in order to model
the full composite action between the deck
and the girder. In this manner, both the
beam and shell nodes will experience
similar translations and rotations in all three
directions. Similarly, I will adopt both the
Figure 29. Strongwell FRP Deck material properties (Liu three-layered and the equivalent deck
et al, 2008) approaches for my analysis. However, the
differences in my analysis are: (1) I have
used FRP-concrete composite girders
instead of steel girders and (2) the constraint
equations governing the behaviour of the
nodes of the beam elements (girder) and
shell elements (bridge deck) are replaced by
the initial physical modelling of the lines as
part of the deck and girder. The results
obtained from both approaches will be
compared and discussed in Chapter 5.
Figure 30. Finite element model (Liu et al, 2008)

D. Meshing
Before meshing the model, the element attributes y
and mesh sizes must be assigned to the physical model.
Element attributes define the type of element for the x
particular meshed volume, area or line. Mesh sizes can
be controlled manually to ensure that the elements
generated from meshing will aid in the analysis. This
will become more evident in Chapter 5.

Figure 31 shows a line plot of the bridge layout in plan


view. I have selected a mesh size of 125 mm for the
nine blue dotted lines in the x-direction and 100 mm for
the 16 red dotted lines in the y-direction. All the deck
rectangles are assigned the SHELL 91 element and real
constant model 1. All the blue dotted lines are assigned
the BEAM 189 element with the correct Section_ID to
each line. The purpose of assigning the correct
Section_ID to each line is to ensure that the correct
offset of the beam from the centreline of the bridge deck Figure 31. Line plot of bridge after meshing
is applied. Both the deck rectangles and blue dotted (Plan view)
lines are now ready to be meshed after specifying their
element attributes and mesh sizes.

Figure 32 shows part of the cross sectional profile of the bridge deck and three girders after meshing the model.
Beam 2 represents a beam without any offsets assigned to it. Hence, the centreline of the deck coincides with
the bottom of the Type B beam. Solving such a meshed model without the correct offset applied will result in
erroneous results. Beam 1 represents a beam with the correct vertical offset applied. However, Beam 1 requires

28
Final Thesis Report 2009, SEIT, UNSW@ADFA
further horizontal offset in the y-direction to
ensure that the entire beam is beneath the
bridge deck. Horizontal offsets are applied to
the two end girders to ensure the entire girder
is below the bridge deck. Beam 3 represents a
beam with the correct vertical and horizontal
offsets. The top of the beam touches the soffit
of the bridge deck. The midpoint of Beam 3
2 coincides with the blue dotted lines to ensure
that the centre to centre distances of all the
girders in the bridge, excluding the two end
girders, are 875 mm apart.

1 3
Figure 32. Offsets of beams from centreline of deck

E. Load

1. Displacements

x x z
x x
y
x x
x

x
x x
8000 mm

Figure 33. Constraints on a simply supported beam

Boundary conditions or constraints are applied as structural displacement loads within ANSYS. The
displacements are applied on the nodes of the BEAM 189 elements at the end spans of the bridge. I will use an
example for a simply supported beam to explain the constraints for the whole bridge. The nodes at both ends of
the beam are free to rotate about the y-axis. One of the end nodes is also free to translate in the x-axis. The
unconstrained translation and rotations are indicated by the green ticks while the constrained translations and
rotations are indicated by the red crosses as shown in Figure 33.

The constraints as applied in the above simply supported


beam are then extended to all the girders in the bridge
structure. Special attention has to be paid to the selection
of the correct BEAM 189 element end nodes when
applying the constraints. An easy method of doing this is
to execute a ‘Multi-plot’ function such that all elements,
nodes, keypoints and lines are plotted in the same plot.
The correct end nodes are then selected using the ‘zoom’
and ‘pan’ functions. As a rule of thumb, the correct end
nodes must be located at the mid-point of the girder and
they are connected by dotted lines representing the lines
meshed with beam elements, running in the x-direction. A
screenshot of the degree of freedom (DOF) constraints on
the end nodes of the BEAM 189 element is shown in
Figure 34. A diagram of the FRP bridge when fully
Figure 34. Constraints of ANSYS FRP meshed is shown in Figure 35. The view of the meshed
bridge model girders can be toggled on and off in order to facilitate
easier graphical handling of the model.

29
Final Thesis Report 2009, SEIT, UNSW@ADFA
2. Force and pressure
The effects of self-weight of all the structural
elements are considered by assigning densities to all
the material models. An acceleration of 9.81 ms-2
due to gravity is then applied as a global inertia load.

The axle loads, P and UDL, w are applied as pressure


on selected elements. One of the problems faced in
applying axles loads, P is that they have to be applied
directly onto the elements and not onto the area of
the axle load rectangles. This is because only the
deck rectangles are meshed with SHELL 91 elements
and not the axle rectangles. However, the axle
rectangles can assist in the application of axle loads
P with the selection of the correct elements. The
accuracy of applying P depends on the fineness of Figure 35. Model of FRP bridge fully meshed
the mesh and this will be discussed in Chapter 5.
There are no general problems in applying the UDL since w can be applied directly onto the deck rectangles
which are meshed with shell elements.

V. Discussion of Results

A. Computational Results for a Single Type B Beam


The theoretical deflection results obtained for Type B beam in Step 3 is compared with the computational
results in ANSYS. The theoretical deflections due to Ptriaxle and w are 5.11 mm and 8.23 mm respectively as
shown in Table 7. The computational deflections are obtained as 6.33 mm and 10 mm. These results are
displayed in Figure 36. The differences between the theoretical and computational results are approximately
1.22 mm or 23.9% for the 4 point loading test and 1.75 mm or 21.3% for the UDL test. One of the possible
explanations for this is that the ±45° GFRP laminate wrap was not modelled in ANSYS. If included in the
model, the flexural stiffness of the beam will increase and hence reduce the deflections obtained in ANSYS.

(a) (b)
Figure 36. Type B Beam deflections from (a) Ptriaxle loading (b) w loading using a coarse mesh

30
Final Thesis Report 2009, SEIT, UNSW@ADFA
The computational results obtained also
demonstrated that when analysing the deflections
in the beam, there is no significant advantage in
using a finer mesh. The mesh size is reduced from
the previous 200 mm element length size to 50
mm, resulting in an increase of the number of
elements from 40 to 160. However, the beam
deflections obtained using a finer mesh is 6.39 mm
as shown in Figure 37 and when compared with
the results obtained using a coarser mesh, the
difference is approximately 0.06 mm or 1%, which
is not a significant difference.

The only advantage is that with a finer mesh, there


are more beam element nodes along the line and
this allows the user to apply the loads accurately.
Figure 37. Type B Beam deflections from Ptriaxle loading This is demonstrated in the following section
using a fine mesh where the mesh sizes for the bridge deck are
discussed.

P P P P P
3.75 m
1.25 m 1.25 m 1.25 m

8m

Figure 38. S 1600 loading model on a Type B Beam in ANSYS

At the start of Step 3, I have stated my consideration of combining three axle loads and applying them as Ptriaxle
in order to facilitate a quicker beam analysis with the use of an established deflection equation. The exact axle
loads, P are now applied in accordance with the S 1600 traffic load model as shown in Figure 38. Five axle
loads of 24 kN each are applied at their correct positions on the nodes of the beam elements and a deflection of
3.87 mm is obtained near the middle of the
beam as shown in Figure 39. Compared to
the deflection of 6.39 mm obtained when
applying Ptriaxle, the deflection of 3.87 mm is
about 40 % smaller and this suggests that the
approach used earlier in Step 3 is over
conservative. This discrepancy can be
solved by using a different set of deflection
limit instead of L/600 in Step 3. The
advantage of using a different deflection
limit is that Step 3, being an initial step in
the design process, should be done quickly
and yet reasonably accurate. The final
dimensions of the beam after the end of Step
3 can then be modelled and tested under the
actual S 1600 loading model using ANSYS.

Figure 39. Type B Beam deflections from S 1600 load model

31
Final Thesis Report 2009, SEIT, UNSW@ADFA
B. Mesh Size Control
One of the advantages of having a finer mesh is the application of a more accurate load model. Due to the
geometry of the elements (as a result of user defined mesh sizes), there might be limitation in modelling the
exact location and size of the axle contact area after the S 1600 traffic load model. This limitation is
demonstrated in Figure 40.

• The axle contact areas of 250 mm by 400 mm are indicated by the blue lines and the element areas in
which pressure are applied on are indicated by the translucent yellow boxes.

• The elements sizes are as follows: (a) 200 mm by 175mm, (b) 125 mm by 97.2 mm and (c) 125 mm by
25 mm.

• The actual axle contact area is 100 000 mm2 and the element areas highlighted in yellow are (a) 140 000
mm2, (b) 97 200 mm2 and (c) 100 000 mm2.

• An axle load, P of 40 kN is applied to all three mesh models but the force has to be converted to the
correct pressure for each mesh. The corrected pressure is obtained by dividing P over the highlighted
element areas.

(a) (b) (c)


Figure 40. Comparison of deck element meshes

Although the same load is still applied in the form of a corrected pressure, meshes (a) and (b) are unable to
model the exact location the axle pressure is applied on. This is demonstrated by the fact that the blue boxes and
the highlighted yellow boxes do not match in Figure 40 (a) and (b). Only the mesh (c) is capable of replicating
the actual S 1600 traffic load model but some of the disadvantages of using such a fine mesh are the increased
difficulty in computational handling and the increase in computation time.

The deflections obtained for mesh models (a), (b) and (c) from the loading as shown in Figure 38 are (a) 3.16
mm, (b) 3.15 mm and (c) 3.14 mm respectively. The differences in deflections using different meshes (a), (b)
and (c) are very small and can be considered as insignificant. Hence, I have chosen to use mesh (b) for my
subsequent analysis as a compromise between computational time and accuracy of replicating the actual S 1600
traffic load model.

32
Final Thesis Report 2009, SEIT, UNSW@ADFA
C. Deflections

1. Three-layered approach
As discussed in Step 3, the design of FRP structures is driven by deflection limits. Hence, the computational
analysis will involve deflection checks only. Three different load conditions based on the S 1600 traffic load
model are analysed. For load condition 1 (LC1), axle loads, P are applied on the left road lane only. For load
condition 2 (LC2), UDL, w is applied on the right road lane only. Load condition 3 (LC3) is a combination of
LC1 and LC 2 whereby P is applied on the left road lane and w is applied on the right load lane.

(a) (b)

The 3-layered approach is first examined. Figure 41


shows the deflections of the FRP bridge
superstructure due to the three S 1600 traffic load
conditions mentioned above. The deflections are
3.15 mm for LC1 (Figure 41(a)), 8.28 mm for LC2
(Figure 41(b)) and 7.96 mm for LC 3 (Figure 41(b)).
It may be reasonable to assume that LC3 will cause
the largest deflection since the overall loading for
LC3 is the largest amongst the three load conditions.
However, the worse case loading scenario is LC2.
From this, I have decided to investigate the effects of
applying w on both road lanes. This load condition
will now be known as load condition 4 (LC4). The
largest deflection obtained from LC4 is 9.47 mm.
This deflection is located near the middle of the
(c) bridge.
Figure 41. Deflections of FRP bridge from S 1600
loading using a 3 layered approach

2. Equivalent deck approach


The equivalent deck approach was also analysed.
This approach is easier to handle as there is only one
layer to be considered and hence, one material model
is required to define the real constant for the bridge
deck. The rest of the inputs such as mesh size controls,
constraints of end nodes, element types and loads are
the same as the previous three-layered approach The
deflection from LC4 for the equivalent deck approach
is 9.45 mm, which is very similar to the deflection
obtained using the three-layered approach. In other Figure 42. Deflections from equivalent deck
approach
33
Final Thesis Report 2009, SEIT, UNSW@ADFA
words, the equivalent deck approach is sufficiently accurate for future analysis done on this particular
Strongwell bridge deck.

VI. Conclusion
In Chapter 1, this paper has introduced the main problems facing existing bridges which are constructed out of
conventional materials such as timber, concrete, cast iron and steel. FRP materials have certain superior material
properties over conventional materials and pose the potential of being a greener construction material. Hence,
FRP materials are beginning to gain popularity in the area of rehabilitation of bridges and replacement of bridge
deck. However, the lack of design standards and the high initial cost of FRP materials are limiting their usage in
the area of construction of new bridges.

The proposed FRP bridge structure is introduced in Chapter 2. An ‘off-the shelf’ bridge deck on FRP-concrete
composite girder design is proposed to solve the existing problems facing many of the timber bridges within
Australia. The concept behind the hybrid beam design is explained in this chapter, and a literature review of the
types of hybrid beam is provided as well. The design methodology adopted in this paper is explained in Chapter
3. A stepped and logical process is outlined to determine the (1) number of girders required for the bridge
structure, (2) design loads based on SLS and ULS and (3) design of the hybrid beam. Each step will achieve a
deliverable to be used in the later stages of the design.

The FRP bridge is then analysed computationally using ANSYS in Chapters 4 and 5. Firstly, the hybrid beam is
modelled in ANSYS. The computational result obtained from a four point loading test is verified against the
theoretical result obtained in Chapter 3. Next, the whole of bridge superstructure is modelled. Both the three-
layered and equivalent deck approaches are analysed. All computational deflection results are achieved using
the S 1600 load model and SLS design adapted from AS 5100. They are then compared against the deflection
limit set in AS 5100.

It can be concluded that:

• The bridge deck deflections obtained in Step 2 of Chapter 3 are much smaller than the allowable
deflection stated in AS 5100 and this perhaps suggests lesser girders are required for the proposed FRP
bridge. However, a larger load will act on each girder if fewer girders are used to support the bridge deck
and traffic loads. The deflection of a single beam with a larger load needs to be rechecked again in Step 3 if
lesser girders are used.

• The web buckling load in a GFRP pultruded profile increases significantly just by segmenting the
pultruded box sections into smaller sections as demonstrated in a Type B beam. However, more
complicated profiles would result in higher cost of producing the pultrusion dies.

• The largest deflection of 9.47 mm is obtained from LC4. This is less than the deflection limit of 13.3 mm
specified in AS 5100 and hence the FRP bridge passes the serviceability criteria. Also, the equivalent deck
approach is sufficiently accurate for future analysis on this particular Strongwell bridge deck design.

VII. Future Work and Recommendations


The ultimate purpose of designing the FRP bridge is such that the FRP bridge can be constructed and performs
its role of providing a means to cross a gap. The design of the proposed FRP bridge is not yet complete as the
design should include all structural elements, such as the connection between the deck and girders, and non-
structural elements such as the safety railings along the bridge. Also, only static loads adapted from the AS 5100,
are applied in this study. The following recommendations are made for future research in order to achieve the
ultimate purpose of design.

• The other types of load effects such as thermal effects, earth pressure loads, pedestrian traffic loads and
wind loads (relevant to the location of the bridge) should be considered in their contribution to the overall
design loads. The fatigue study of the bridge is vital since traffic loads are cyclic in reality. A long term
fatigue study with cyclic loading should be conducted on the proposed FRP bridge.

• The various load factors adapted from the AS 5100 are applicable for steel, concrete and timber bridges
only. These factors of safety for each load effects have been derived from numerous experimental and field

34
Final Thesis Report 2009, SEIT, UNSW@ADFA
data done on these bridges. An independent study on FRP composite bridges should be conducted to obtain
factors of safety that are perhaps applicable to composites only.

• The GRFP winding of the hybrid beam has been tested experimentally and demonstrated good results in
increasing the flexural stiffness of the beam. The GFRP wrap can be modelled with its appropriate
behaviour of confining the concrete layer to the pultruded profile in order to achieve better composite
action between the two layers. Work was done to model the GFRP wrap but problems with the meshing of
the GFRP wrap were encountered since the thickness of the GFRP wrap is small compared to the other
dimensions of the beam profile. This resulted in a fine mesh being used and an increased in computational
time. A more efficient method of modelling the hybrid beam with GFRP winding should be trialled.

• There are two types of deck to girder connections currently – use of mechanical connectors and adhesive
bonding. Each method of connection has its own advantages and disadvantages and a study should be
conducted to investigate which method is best suited for this proposed FRP bridge. The design of special
connectors should be done to maximise the contact area for adhesive bonds between the soffit of the deck
and the girders. After all, the FRP bridge is modelled assuming rigid connections between the bridge deck
and girders.

Acknowledgements
This report details the work done by the author for his final-year thesis in the School of Engineering and
Information Technology, University of New South Wales at the University College, Australian Defence Force
Academy. The author would like to convey his gratitude to his thesis supervisors, Dr. Amar Khennane and Mr.
Gary Barker for their guidance and support. The author is also thankful to Mr. Anup Chakrabortty and Dr.
Murat Tahtali whose inputs have assisted this work. Lastly, special mention must be given to family and friends
who have continually supported this author in his academic endeavour.

35
Final Thesis Report 2009, SEIT, UNSW@ADFA
References

Alnahhal, W. & Aref, A. 2008, “Structural performance of hybrid fiber reinforced polymer-concrete bridge
suerpstructure systems”, Composite Structures, vol. 84, pp. 319-336.

ANSYS Release 11.0, Help Documentation, 2009.

Aref, A. & Parsons, I.D. 2000, “Design and performance of a modular fiber reinforced plastic bridge”,
Composites: Part B, vol. 31, pp. 619-628.

AS 5100, Australian Standards, Bridge Design Code, 1992.

AS 5100, Australian Standards, Bridge Design Code, 2004.

Bakis, C.E., Bank, L.C., Brown, V.L., Cosenza, E., Davalos, J.F., Lesko, J.J., Machida, A., Rizkalla, S.H. &
Triantafillou, T.C. 2002, “Fiber-reinforced polymer composites for construction- state-of-the-art review”,
Journal of Composites for Construction, vol. 6, no. 2, pp. 73-87.

Cai, C.S., Oghumu, S.O., Meggers, D.A. 2009, “Finite-element modeling and development of equivalent
properties of FRP bridge panels”, Journal of Bridge Engineering, vol. 14, no. 2, pp. 112-121.

Canning, L., Hodgson, J., Karuna, R., Luke, S. & Brown, P. 2007, “Progress of advanced composites for civil
infrastructure”, Structures and Buildings, vol. 160, pp. 307-315.

Canning, L., Hollaway, L. and Thorne A.M. 1999, “Manufacture, testing and numerical analysis of an
innovative polymer composite/concrete structural unit”, Proceedings – Institution of Civil Engineers, Structures
and Buildings, 134, pp. 231-241.

Cassity, P., Richards, D., Gillespie, J. 2002, “Compositely acting FRP deck and girder system”, Structural
Engineering International, vol. 2, pp. 71-75.

Chakrabrotty, A. 2009, Postgraduate student, SEIT, UNSW@ADFA (Personal communication).

Charoenphan, S., Bank, L.C. & Plesha, M.E. 2004, “Progressive tearing failure in pultruded composite material
tubes”, Composite Structures, vol. 63, pp. 45-52.

Cheng, L. & Karbhari, V.M. 2006, “New bridge systems using FRP composites and concrete: a state-of-the-art
review”, Progress in Structural Engineering and Materials, vol. 8, pp. 143-154.

Correia, J.R., Branco, F.A. & Ferreira, J.G., 2007, “Flexural behavior of GFRP-concrete hybrid beams with
interconnection slip”, Composite Structures, vol. 77, pp. 66-78.

Daniel, R.A. 2003, “Environmental considerations to structural material selections for a bridge”, European
Bridge Engineering Conference, pp. 1-10.

Davalos, J.F., Qiao, P., Xu, X., Robinson, J. & Barth, K.E. 2001, “Modeling and characterization of fiber
reinforced plastic honeycomb sandwich panels for highway bridge applications”, Composite Structures, vol. 52,
pp. 441-452.

Deskovic, N., Triantafillou, T.C. & Meier, U. 1995, “Innovative design of FRP combined with concrete: short
term behavior”, Journal of Structural Engineering, vol. 121, no. 7, pp. 1069-1078.

Garden, H.N. 2004, “Use of advanced composites in civil engineering infrastructure”, Strucutures & Buildings,
vol. 157, pp. 357-368.

36
Final Thesis Report 2009, SEIT, UNSW@ADFA
Harries, K.A. 2008, “FRP components for bridge superstructures – US perspective”, Proceedings of the 4th
International Conference on FRP Composites in Civil Engineering, Powerpoint presentation.

Heldt, T., Oates, R., van Erp, G. & Marsh, R. 2004, “Australia's first reinforced polymer bridge deck in a road
network - The anatomy of innovation”.

Heywood, R., Gordon, R. & Boully, G. 2000, “Australia’s Bridge Design Load Model”, Fifth International
Bridge Engineering Conference, vol. 2, pp. 1-7.

Hollaway, L.C. 2003, “The evolution of and the way forward for advanced polymer composites in the civil
infrastructure”, Composites and Building Materials, vol. 17, pp. 365-378.

Holmes, M. & Just, D.J. 1983, GRP in Structural Engineering, Applied Science Publishers, UK.

Hulatt, J., Hollaway, L. & Thorne, A. 2003, “Short term testing of hybrid T-beam made of new prepeg
material”, Journal of Composites for Construction, vol. 7, no. 2, pp. 135-144.

Karbhari, V.M. 2004, “Fiber reinforced composite bridge systems - transition from the laboratory to the field”,
Composite Structures, vol. 66, pp. 5-16.

Karbhari, V.M. 2003, “Durability of FRP composites for civil infrastructure - myth, mystery or reality”,
Advances in Structural Engineering, vol. 6, no. 3, pp. 243-255.

Karbhari, V.M. & Seible, F. 2000, “Fiber reinforced composites - advanced materials for the renewal of civil
infrastructure”, Applied Composite Materials, vol. 7, pp. 95-124.

Karbhari, V.M. & Zhao, L. 2000, “Use of composites for 21st century civil infrastructure”, Computer Methods
in Applied Mechanics and Engineering, no. 185, pp. 433-454.

Khennane, A. 2009, “Manufacture and testing of a hybrid beam using a pultruded profile and high strength
concrete”, Australian Structural Journal.

Liu, Z., Cousins, T.E., Lesko, J.J. & Sotelino, E.D. 2008, “Design recommendations for a FRP bridge deck
supported on steel superstructure”, Journal of Composites for Construction, vol. 12, no. 6, pp. 660-668.

Machado, M.A.S., Sotelino, E.D., Liu, J. 2008, “Modelling technique for honeycomb FRP deck bridges via
finite elements”, Journal of Structural Engineering, vol. 134, no. 4, pp. 572-580.

McCormick, F.C. 1993, “First traffic-bearing composite was built in China”, Civil Engineering, vol. 63, no. 10,
pg. 35.

Meier, U. 2000, “Composite materials in bridge repair”, Applied Composite Materials, vol. 7, pp. 75-94.

Naik, T.R. 2008, “Sustainability of concrete construction”, Practice Periodical on Structural Design and
Construction, vol. 13, no. 2, pp.98-103.

O’Connor, C. & Shaw, P. 2000, Bridge Loads, Spon Press.

Ribeiro, M.C.S., Tavares, C.M.L., Ferreira, A.J.M. & Marques, A.T. 2002, “Static flexural performance of
GFRP-polymer concrete hybrid beams”, Key Engineering Materials, vol. 230, pp. 148-151.

Salim, H.A., Barker, M., Davalos, J.F. 2006, “Approximate series solution for analysis of FRP highway
bridges”, Journal of Composites for Construction, vol. 10, no. 4, pp. 357-366.

Scott, I. & Wheeler, K. 2001, “Application of fibre reinforce polymer composites in bridge construction”,
Institution of Public Works Engineering Australia, Second Conference.

37
Final Thesis Report 2009, SEIT, UNSW@ADFA
Strongwell. 2009, Accessed 15 October 2009 from World Wide Web: http://www.strongwell.com.

Teng, J.G., Chen, J.F., Simth, S.T. & Lam, L. 2003, “Behaviour and strength of FRP-strengthened RC structure:
a state-of-the-art review”, Structures & Buildings, vol. 156, no. 1, pp. 51-62.

van den Einde, L., Zhao, L. & Seible, F. 2003, “Use of FRP composites in civil structural applications”,
Construction and Building Materials, vol. 17, pp. 389-403.

van Erp, G., Cattell, C.L., & Ayers, S. 2005, “The Australian approach to composites in civil engineering”,
Reinforced Plastics, vol. 49, no.6, pp.20-26.

van Erp, G., Heldt, T., McCormick, L., Carter, D. & Tranberg, C. 2002, “An Australian approach to fibre
composite bridges”, Composite Systems - Macrocomposites, Microcomposites, Nanocomposites, , pp. 145-153.

Warner, R.F., Rangan, B.V., Hall, A.S. & Faulkes, K.A. 1998, Concrete Structures, Longman.

Wu, H.C. 2006, Advanced civil infrastructure materials, 1st edition, Woodhead Publishing, pp. 118-200.

Appendices
A. Project Management Documentation
(i) Client brief
(ii) Task list
(iii) Milestone chart

B. Matlab (ver. 2008) beam analysis results


(i) Input parameters for S1600 stationary traffic load
(ii) M1600 moving traffic load
(iii) S1600 stationary traffic load

C. Maple (ver. 10) calculations for Step 3


(i) Type A Beam
(ii) Type B Beam

38
Final Thesis Report 2009, SEIT, UNSW@ADFA

You might also like