University of Southern Queensland

Faculty of Health, Engineering and Sciences

TOPOLOGY OPTIMISATION OF LARGE

REINFORCED CONCRETE BOX CULVERTS
UNDER SM1600 LOADS

A dissertation submitted by

Mrs. Juliana Sweeney

In fulfilment of the requirements for the

degree of

Bachelor of Engineering Hons. (Civil)

November 2014
 ABSTRACT

Topology optimisation of large reinforced concrete box culverts under

SM1600 loads

by Juliana Sweeney

Supervisor: Dr. Sourish Banerjee

School of Civil Engineering and Surveying

This research project is concerned with finding the optimum three-sided large box
culvert through topology optimisation using finite element analysis. The objective
function is to minimise the total strain energy while the design constraints include
minimising volume as a fraction of the initial volume and geometric restrictions to
ensure symmetry and appropriate cover to reinforcement. The optimised culvert
must also comply with the latest Australian specifications, must be subjected to
standard SM1600 loads for main roads and must be feasible and constructible to
be useful and practical to the Australian industry.

i
 LIMITATIONS OF USE

The Council of the University of Southern Queensland, its Faculty of Health,

Engineering & Sciences, and the staff of the University of Southern Queensland,
do not accept any responsibility for the truth, accuracy or completeness of material
contained within or associated with this dissertation.

Persons using all or any part of this material do so at their own risk, and not at the
risk of the Council of the University of Southern Queensland, its Faculty of Health,
Engineering & Sciences or the staff of the University of Southern Queensland.

This dissertation reports an educational exercise and has no purpose or validity

beyond this exercise. The sole purpose of the course pair entitled “Research
Project” is to contribute to the overall education within the student’s chosen degree
program. This document, the associated hardware, software, drawings, and other
material set out in the associated appendices should not be used for any other
purpose: if they are so used, it is entirely at the risk of the user.

ii
 CERTIFICATION OF DISSERTATION

I certify that the ideas, designs and experimental work, results, analyses and
conclusions set out in this dissertation are entirely my own effort, except where
otherwise indicated and acknowledged.

I further certify that the work is original and has not been previously submitted for
assessment in any other course or institution, except where specifically stated.

Juliana Sweeney
0061027945

iii
 ACKNOWLEDGMENTS

The author wishes to express sincere appreciation to Dr. Sourish Banerjee for his
help and level of commitment to supervising this project.

Special thanks is also due to Mr. Wayne Sue Tin, an excellent teacher without
whom this entire project would have been impossible. Another great teacher who
lent a hand was Mr. Mark Edwards, who was always willing to help and discuss
design issues.

Another extremely important person in the process of writing this thesis is Mr.
John Sweeney, husband and best friend, who was source of help, encouragement
and comfort at all times.

iv
 TABLE OF CONTENTS

Abstract .................................................................................................................................. i
Limitations of use ................................................................................................................ ii
Certification of dissertation .............................................................................................. iii
Acknowledgments .............................................................................................................. iv
List of Figures ................................................................................................................... viii
List of Tables........................................................................................................................ x
List of Appendices ............................................................................................................. xi
Definitions .......................................................................................................................... xii
Standards and Technical Specifications ....................................................................... xiii
Chapter 1 - Introduction ............................................................................................. 1
1.1 Background ..................................................................................................... 1
1.2 Research scope and objectives .................................................................... 1
Chapter 2 - Literature Review .................................................................................... 4
2.1 Introduction .................................................................................................... 4
2.2 Optimisation ................................................................................................... 4
2.3 Topology optimisation.................................................................................. 5
2.3.1 Solid Isotropic Material with Penalization (SIMP) ........................... 5
2.3.2 Homogenization based approach ........................................................ 6
2.3.3 Evolutionary approaches....................................................................... 7
2.3.4 Level-set method .................................................................................... 7
2.4 Cost optimisation: .......................................................................................... 8
2.5 Size optimisation ............................................................................................ 9
2.6 Shape optimisation ........................................................................................ 9
2.7 Reinforcement optimisation ........................................................................ 9
2.8 Optimisation and Constructability ........................................................... 11
2.9 Using finite element analysis (FEA) to model RCBC behaviour ....... 12
2.10 Conclusions................................................................................................... 16
Chapter 3 - Methodology.......................................................................................... 17
3.1 Introduction .................................................................................................. 17
3.2 Analysis of SM1600 loads .......................................................................... 18
3.2.1 W80 load ................................................................................................ 18
3.2.2 A160 load ............................................................................................... 19
3.2.3 M1600 load ............................................................................................ 20
3.2.4 Load factors ........................................................................................... 23
3.2.5 Self-weight WDC..................................................................................... 24
3.2.6 Vertical earth pressure due to fill WFV .............................................. 24
3.2.7 Construction live load induced vertical earth pressure WCV ......... 25
3.2.8 Vertical loads due to road traffic loadings WLV............................... 25

v
 3.2.9 Horizontal loads due to fill and compaction WFH and WAH ......... 26
3.3 Designing the standard RCBC .................................................................. 27
3.3.1 Materials ................................................................................................. 29
3.3.2 Durability Design.................................................................................. 29
3.3.3 Stability and Strength Design ............................................................. 31
3.3.4 Serviceability Design ............................................................................ 31
3.3.5 Non-optimised Culverts ...................................................................... 32
3.3.6 Reinforcement ....................................................................................... 32
3.4 Finding the optimum topology using the SIMP method ..................... 37
3.4.1 Matlab script: top_rcbc4.m ......................................................... 39
3.5 Feasibility analysis ........................................................................................ 42
3.5.1 Production Costs of an RCBC ........................................................... 42
3.5.2 Estimating Procedure .......................................................................... 43
3.6 Conclusion .................................................................................................... 46
Chapter 4 - RCBC Design ........................................................................................ 47
4.1 Introduction .................................................................................................. 47
4.2 Load Combination Results ........................................................................ 47
4.2.1 Symmetric loading ................................................................................ 48
4.2.2 Asymmetric loading.............................................................................. 50
4.3 Design Loads ................................................................................................ 51
4.4 1815 RCBC Design ..................................................................................... 51
4.5 1818 RCBC Design ..................................................................................... 65
4.6 2412 RCBC Design ..................................................................................... 75
4.7 2415 RCBC Design ..................................................................................... 86
4.8 2418 RCBC Design ..................................................................................... 96
4.9 Conclusion ..................................................................................................106
Chapter 5 - Topology Optimisation .....................................................................108
5.1 Introduction ................................................................................................108
5.2 1815 RCBC .................................................................................................108
5.3 1818 RCBC .................................................................................................112
5.4 2412 RCBC .................................................................................................116
5.5 2415 RCBC .................................................................................................121
5.6 2418 RCBC .................................................................................................125
5.7 Conclusion ..................................................................................................129
Chapter 6 - Feasibility Analysis..............................................................................131
6.1 Introduction ................................................................................................131
6.2 Labour and design components of total cost .......................................131
6.3 1815 RCBC .................................................................................................132
6.3.1 Materials ...............................................................................................132
6.3.2 Conclusion ...........................................................................................132
6.4 1818 RCBC .................................................................................................133
6.4.1 Materials ...............................................................................................133
6.4.2 Conclusion ...........................................................................................134

vi
 6.5 2412 RCBC .................................................................................................135
6.5.1 Materials ...............................................................................................135
6.5.2 Conclusion ...........................................................................................135
6.6 2415 RCBC .................................................................................................136
6.6.1 Materials ...............................................................................................136
6.6.2 Conclusion ...........................................................................................137
6.7 2418 RCBC .................................................................................................137
6.7.1 Materials ...............................................................................................137
6.7.2 Conclusion ...........................................................................................138
6.8 Conclusions.................................................................................................139
Chapter 7 - Conclusions..........................................................................................141
7.1 Achievements .............................................................................................141
7.2 Conclusions.................................................................................................141
7.3 Possible further work ................................................................................142
References.........................................................................................................................144
Appendices .......................................................................................................................149

vii
 LIST OF FIGURES

Number Page
Figure 2-1 - Four-sided box culvert (Foley Products 2014) ...................................... 12
Figure 2-2 - Three-sided box culvert (Rocla 2014) ..................................................... 13
Figure 3-1 - W80 wheel load ........................................................................................... 19
Figure 3-2 - A160 axle load ............................................................................................. 20
Figure 3-3 - M1600 tri-axle load ..................................................................................... 20
Figure 3-4 - M1600 according to AS1597.2 ................................................................. 21
Figure 3-5 - Compressive stress block .......................................................................... 33
Figure 3-6 - Ultimate bending capacity conditions ..................................................... 34
Figure 4-1 - Vertical load combinations........................................................................ 48
Figure 4-2 - Horizontal load combinations .................................................................. 48
Figure 4-3 – Asymmetric horizontal load combinations – one side of culvert ..... 50
Figure 4-4 – Asymmetric horizontal load combinations – other side of culvert .. 50
Figure 4-5 – Loads on top of 1815 RCBC ................................................................... 53
Figure 4-6 - Vertical loads................................................................................................ 55
Figure 4-7 - Horizontal live load .................................................................................... 55
Figure 4-8 - Compaction load ......................................................................................... 55
Figure 4-9 - Horizontal fill load ...................................................................................... 56
Figure 4-10 - Loads on the side of 1815 RCBC .......................................................... 56
Figure 4-11 – Bending moment diagram for 1815 RCBC ........................................ 57
Figure 4-12 - Shear force diagram for 1815 RCBC .................................................... 58
Figure 4-13 - 1815 RCBC reinforcement ..................................................................... 64
Figure 4-14 – Loads on top of 1818 RCBC ................................................................. 66
Figure 4-15 - Loads on the side of 1818 RCBC .......................................................... 67
Figure 4-16 – Bending moment diagram for 1818 RCBC ........................................ 68
Figure 4-17 - Shear force diagram for 1818 RCBC .................................................... 68
Figure 4-18 - 1818 RCBC reinforcement ..................................................................... 74
Figure 4-19 – Loads on top of RCBC 2412 ................................................................. 76
Figure 4-20 - Loads on the side of the 2412 RCBC ................................................... 77
Figure 4-21 - Bending moment diagram for 2412 RCBC ......................................... 78
Figure 4-22 - Shear force diagram for 2412 RCBC .................................................... 78
Figure 4-23 - Bending moment and shear force for the 2412 RCBC leg ............... 83
Figure 4-24 - 2412 RCBC reinforcement ..................................................................... 85
Figure 4-25 – Loads on top of 2415 RCBC ................................................................. 87
Figure 4-26 - Loads on the side of the 2415 RCBC ................................................... 88
Figure 4-27 - Bending moment diagram for 2415 RCBC ......................................... 89
Figure 4-28 - Shear force diagram for 2415 RCBC .................................................... 89
Figure 4-29 - 2415 RCBC reinforcement ..................................................................... 95

viii
Figure 4-30 – Loads on top of 2418 RCBC ................................................................. 97
Figure 4-31 - Loads on the side of the 2418 RCBC ................................................... 98
Figure 4-32 - Bending moment diagram for 2418 RCBC ......................................... 99
Figure 4-33 - Shear force diagram for 2418 RCBC .................................................... 99
Figure 4-34 - 2418 RCBC reinforcement ...................................................................105
Figure 5-1 - Topology optimisation result for 1815 RCBC ....................................109
Figure 5-2 – Optimised 1815 RCBC section elevation ............................................109
Figure 5-3 - Optimised 1815 with trimmed voids ....................................................110
Figure 5-4 - Optimised 1815 with trimmed voids and reinforcement ..................111
Figure 5-5 - Final 1815 optimised RCBC ...................................................................111
Figure 5-6 - Topology optimisation result for 1818 RCBC ....................................112
Figure 5-7 – Optimised 1815 RCBC section elevation ............................................113
Figure 5-8 - Optimised 1815 RCBC with trimmed voids........................................114
Figure 5-9 - Optimised 1815 RCBC with trimmed voids and reinforcement .....115
Figure 5-10 - Final 1818 optimised RCBC .................................................................116
Figure 5-11 - Topology optimisation result for 2412 RCBC ..................................117
Figure 5-12 – Optimised 2412 RCBC section elevation..........................................118
Figure 5-13 - Optimised 2412 RCBC with trimmed voids .....................................119
Figure 5-14 - Optimised 2412 RCBC with trimmed voids and reinforcement ...120
Figure 5-15 - Final 2412 optimised RCBC .................................................................120
Figure 5-16 - Topology optimisation result for 2415 RCBC ..................................121
Figure 5-17 – Optimised 2415 RCBC section elevation..........................................122
Figure 5-18 - Optimised 2415 RCBC with trimmed voids .....................................123
Figure 5-19 - Optimised 2415 RCBC with trimmed voids and reinforcement ...124
Figure 5-20 - Final 2418 optimised RCBC .................................................................124
Figure 5-21 - Topology optimisation result for 2418 RCBC ..................................125
Figure 5-22 – Optimised 2418 RCBC section elevation..........................................126
Figure 5-23 - Optimised 2418 with trimmed voids ..................................................127
Figure 5-24 - Optimised 2418 RCBC with trimmed voids and reinforcement ...128
Figure 5-25 - Final 2418 optimised RCBC .................................................................129
Figure H-1 - Loads on top of 1815 RCBC................................................................. 166
Figure H-2 – Loads on the side of 1815 RCBC....................................................... 167
Figure H-3 – Bending moment diagram for 1815 RCBC....................................... 168
Figure H-4 – Shear force diagram for 1815 RCBC.................................................. 169
Figure H-5 – Topology optimisation result for of 1815 RCBC............................. 175
Figure H-6 – Optimised 1815 RCBC.......................................................................... 176

ix
 LIST OF TABLES

Number Page
Table 3-1 - Stability and strength limit state load factors for vertical loads ........... 23
Table 3-2 - Stability and strength limit state load factors for horizontal loads ...... 24
Table 3-3 - Preferred RCBC internal dimensions ....................................................... 28
Table 4-1 - Vertical loads on 1815 RCBC .................................................................... 54
Table 4-2 - 1815 RCBC reinforcement schedule ........................................................ 65
Table 4-3 - 1818 RCBC reinforcement schedule ........................................................ 75
Table 4-4 - 2412 RCBC reinforcement schedule ........................................................ 86
Table 4-5 - 2415 RCBC reinforcement schedule ........................................................ 96
Table 4-6 - 2418 RCBC reinforcement schedule ......................................................106
Table 6-1 - Cost summary of non-optimised 1815 RCBC......................................133
Table 6-2 - Cost summary of optimised 1815 RCBC ..............................................133
Table 6-3 - Cost summary of non-optimised 1818 RCBC......................................134
Table 6-4 - Cost summary of optimised 1818 RCBC ..............................................134
Table 6-5 - Cost summary of non-optimised 2412 RCBC......................................136
Table 6-6 - Cost summary of optimised 2412 RCBC ..............................................136
Table 6-7 - Cost summary of non-optimised 2415 RCBC......................................137
Table 6-8 - Cost summary of optimised 2415 RCBC ..............................................137
Table 6-9 - Cost summary of non-optimised 2418 RCBC......................................138
Table 6-10 - Cost summary of optimised 2418 RCBC ............................................139

x
 LIST OF APPENDICES

Name Page
Appendix A – Project Specification ............................................................................149
Appendix B – flexanalysis.m .........................................................................................151
Appendix C – checkshear.m .........................................................................................153
Appendix D – finalscript.m...........................................................................................155
Appendix E – devlength.m ...........................................................................................157
Appendix F – top_rcbc4.m ...........................................................................................158
Appendix G – load_comb.m ........................................................................................163
Appendix H – 1815 RCBC Design .............................................................................165

xi
 DEFINITIONS

RCBC. Reinforced concrete box culvert, an inverted U-shape type of culvert

structure

Large RCBC. An RCBC that exceeds 1200 mm in span or 1200 mm in height and
does not exceed 4200 mm in span and 4200 mm in height.

Three-sided box culvert. An RCBC that has two legs and a crown. If this
structure requires a base, it is normally supplied separately or poured insitu.

Four-sided box culvert: An RCBC that has a box format, that is, two legs, a
crown and a base all cast in as one structure.

SM1600. A representation of the W80, A160, M1600 and S1600 design loads.

ESO. Evolutionary Structural Optimisation; a topology optimisation method

BESO. Bi-directional Evolutionary Structural Optimisation; a topology

optimisation method.

xii
 STANDARDS AND TECHNICAL
SPECIFICATIONS

AS1597.2. Australian Standard: Precast reinforced concrete box culverts – Part 2:

Large culverts

AS3600 Australian Standard: Concrete Structures

AS5100 Australian Standard: Bridge Design

AS5100.2 Australian Standard: Bridge Design – Part 2: Design Loads

MRTS24 Transport and Main Roads Specification: Manufacture of Precast

Concrete Culverts

Technical Note 20a Transport and Main Roads Technical Note: Design Criteria
for Large Box Culverts to MRTS24

xiii
 CHAPTER 1 - INTRODUCTION

1.1 Background
Reinforced concrete box culverts are extensively used structural elements to
convey flow of stormwater or sewerage. The typical box culvert produced in
Australia has three sides and it is shaped like an inverted U. It is widely
manufactured by precast concrete manufacturers, normally in steel moulds with
fixed or variable sizes.

The Australian Standard that governs the production aspects of this product is
currently AS1597.2-2013, which supersedes the 1996 version. Part 2 deals with
large box culverts, those with span and height between 1.2 m and 4.2 m. This
Standard gives preferred internal sizes for the box culverts (span and leg
height), and lengths are normally 1.2 m or 2.4 m, with a few exceptions,
depending on the manufacturer. Also, the Department of Transport and Main
Roads Queensland (TMR) specifies a few criteria that are to be met should the
large RCBC be installed under a main road in Queensland. The relevant
document is the MRTS24 (Aug/11), with which the culverts in this study will
also comply.

This study is concerned with finding the optimum topology for reinforced
concrete large box culverts so that the final products are useful to the industry.
To achieve this objective, the RCBCs will be compliant with AS1597.2-2013
and MRST24 (Aug/11) since without this compliance these culverts could not
be sold or installed under main roads. In addition, the culverts will potentially
be cheaper since they will utilise less material.

1.2 Research scope and objectives

This research has four main broad objectives:

1. Find the worst case among the applicable load combinations dictated by
AS1597.2-2013

Page | 1
 Firstly an analysis of all the possible load cases will be carried out to
determine the worst case to be used in design. The RCBCs in this study will
be assumed to be subjected to SM1600 loads, which is the standard traffic
load for large box culverts according to Standards Australia (2013, p. 27).
The SM1600 loads are a combination of the single wheel W80 load, the
single axle A160 load and the moving tri-axle M1600 load according to
AS1597.2-2013.

2. Design the standard RCBC complying with AS1597.2-2013 and MRTS24

(Aug/11)

The design loads found in the previous analysis will then be utilised to
design the RCBC utilising standard practices dictated by AS1597.2-2013
and MRTS24 (Aug/11). Finite element analysis using the software Strand7
will be employed to find design moment and shear capacity. The
reinforcement will then be designed with the application of the concept of
equivalent concrete compressive stress block for flexure analysis. The
standard culvert will later be compared with the optimised culvert to
evaluate its feasibility.

3. Optimise the RCBC using the SIMP method and finite element analysis

The topology optimisation procedure will be implemented using finite

element analysis based on the Solid Isotropic Material with Penalisation
(SIMP) method. The analysis starts with a design domain with finite
elements with relative material densities of 1, and in each iteration the
topology optimisation equation is solved for each element until the
objective function has been reached. In this case, the objective function will
be to diminish strain energy and therefore to increase stiffness, with a
minimum volume constraint. The SIMP method models material
properties as the relative density of each finite element raised to a power,
called the penalisation power, in order to diminish the occurrence of
intermediate densities.

Page | 2
4. Analyse the feasibility of the optimum RCBC

Once the optimum culvert has been found, a feasibility analysis will be
carried out to evaluate the possible advantages of using this method to
design and manufacture RCBCs over utilising standard methods. The cost
of labour and materials, the time to produce the reinforcement cages, the
extra time and cost to construct/prepare moulds, cast and install the units,
among others, will be taken into consideration and a conclusion will be
reached regarding feasibility of the optimum culvert.

This research will investigate 5 of the most commonly sold sizes of RCBC that
are required to comply with Main Roads specifications as well as Australian
Standards. The loads applied to the culvert will be SM1600 loads as specified
by Standards Australia (2013), excluding heavy load platform loads (HLP) and
railway loads. The desired outcome of this project is to find an optimum RCBC
that is compliant with Australian Standards, useful to the industry, cheaper to
manufacture and consequently possibly cheaper to the final customer and that
is environmentally responsible because it will utilise less material.

Page | 3
 CHAPTER 2 - LITERATURE REVIEW

2.1 Introduction
A review of literature for this research has identified a gap in the research for
optimum box culverts that can be fabricated and utilised by the industry in
Australia and this study endeavours to make a contribution to filling this gap.

There are not many FEA models to describe three-sided large box culverts,
which is the type of RCBC commonly sold in Australia. The majority of papers
describe procedures and results that apply to 4-sided culverts. In addition, there
is not much research about topology optimisation of box culverts specifically.
The published work mentions beams or multi-frames.

2.2 Optimisation
The word optimum comes from the Latin optimus, meaning ‘best’ or ‘very
good’. Optimisation is about generating the best possible design. However,
there are different ways to define what is best in terms of design. Some may
believe the cheapest product to be the best, while the design that uses the less
amount of material and therefore provokes the lowest impact on the
environment might be considered best by others.

During the design, manufacture and installation phases of construction of a

structure, many parameters could be optimised. During design, one could aim
for the shortest design time which would translate into savings for the company
in terms of less hours required from the engineering team. It could also bring
the company a competitive advantage if they are able to submit their design
proposal before their competitors. This could mean the structure does not
utilise the least amount of material and it may not have the most efficient size
and shape but in that situation, it may be the only way the company is going to
be hired to do the job.

Page | 4
As for manufacture, the most efficient reinforcement may be composed of
various bar diameters and lengths. This normally means it would take longer
for the reinforcement cage to be assembled since there is more measuring and
cutting involved, incurring extra labour costs if compared with a cage
composed of the same bar diameter of same or length at constant spacing.

In the installation phase, if the optimum structure has to be transported in a

different position, for example, there could be higher installation costs
including crane time and labour to rotate units. Another possibility is when the
optimum structure is thinner that the standard structure and would not be able
to be lifted the same way. The optimum structure could require spreader beams
to be utilised, more rotations or changing of lifting anchors during installation,
which also adds to crane hire and labour costs.

For these reasons, optimisation must be a trade-off between what is desirable,

the optimum structure, and what is feasible, both structurally and commercially.

The literature contains numerous approaches devised to achieve various levels

of optimisation. The most common ones are now discussed.

2.3 Topology optimisation

There is a myriad of approaches in literature to solve topology optimisation
problems, i.e. the homogenization based approach, the Solid Isotropic Material
with Penalization (SIMP) approach, evolutionary design methods and level set
methods, to name a few.

2.3.1 Solid Isotropic Material with Penalization (SIMP)

The SIMP method was first described by Martin Bendsøe in 1989. (Rozvany
2001). This method models material properties as relative densities where a
relative density of 1 indicates the solid material, a density of 0 models a void
and a density between 0 and 1 means the material has voids at a microlevel. To
ensure the material can be realized in practice as composites of the original

Page | 5
material, Bendsøe and Sigmund (1999) have stated that p must satisfy the
 2 4 
condition p ≥ max  ,  where ν is the Poisson ratio of the solid
1 −ν 1 +ν 
material.

One problem that can affect the SIMP results is mesh-dependence, which
causes different solutions to be obtained depending on mesh sizes or
discretization, instead of a more detailed solution of the same optimal structure
(Sigmund & Petersson 1998). One way of preventing this from happening is to
introduce a mesh-independence filtering scheme, which works by modifying
the element sensitivities and is very simple to implement. (Sigmund 2001)

Sigmund (2001) distributed online a 99 line MATLAB code based on the SIMP
method that solves the optimisation problem by applying Optimality Criteria
(OC) methods. These are indirect methods developed in an attempt to diminish
the number of design variables in the optimisation process (Hassani & Hinton
1998).

2.3.2 Homogenization based approach

This method uses a density of a composite with voids. When the density
variable is 0, there is no material (void). When it is 1, there is material (solid). If
the density is between 0 and 1 there is a porous component with voids at
microlevel. The difference between the homogenization based approach and
the SIMP method is that in the homogenization method, the material property
of each finite element is obtained using the homogenization theory, and the
optimal topology is achieved by solving a material distribution problem, while
in the SIMP method the intermediate densities are penalised using the power-
law approach, requiring no homogenization. One disadvantage of this method
is that it may produce infinitesimal pores in the materials that impede
construction. (Zhao, Long & Ma 2010).

Page | 6
2.3.3 Evolutionary approaches
A popular topology optimisation approach is the Evolutionary Structural
Optimisation method. The basis of the ESO method is to remove inefficient
material from the initial structure until a target condition is reached. The
efficiency of the material is evaluated by the level of stress or strain energy in
each element. However, ESO work published in the 1990s disregarded key
aspects of topology optimisation i.e. existence of a solution, checker-board,
mesh dependency and local optimum (Huang 2010).

To overcome these faults, a method called Bi-directional Evolutionary

Structural Optimisation was developed by Huang and Xie (2007). This method
allows elements to be added at the same time as they are removed and it also
deals with the shortcomings mentioned above in the ESO method by utilising
a mesh-independency filter and by including historical information of the
sensitivity numbers of each element to improve their accuracy (Huang & Xie
2007).

2.3.4 Level-set method

Level-set methods were first introduced by Osher and Sethian to model
moving boundaries (Van Dijk et al. 2013).

Shojaee and Mohammadian (2012) explain that this method depicts the
transformation of an interface between two domains. It utilises a level-set
function to describe the boundary as the zero level set, while nonzero level sets
are used in the domain. While the optimization iterations are occurring, the
level set surface may move causing the boundary to suffer considerable
changes.

Wang, Wang and Guo (2004) utilised the level-set method in a boundary
optimisation problem. The domain is represented by a level-set model
embedded in a scalar function, governed by a Hamilton-Jacobi convection
equation. This yielded a 3D structural optimisation technique which gives
results comparable to other established optimisation techniques.

Page | 7
Yamada et al. (2010) proposed a new optimisation technique utilising the level
set method and incorporating a fictitious interface energy (Chan-Hilliard
energy) to overcome numerical instability problems such as mesh-dependency,
checkerboard patterns and greyscales. Their results showed, through various
numerical examples, minimal dependency on the finite element size or initial
configurations.

2.4 Cost optimisation:

Sarma and Adeli (1998) present a review on different approaches used and cost
savings achieved in different reinforced concrete structure optimisation papers
from 1970 to 1996. These structures include beams, slabs, frames, plates and
water tanks, among others, based on standard codes from the USA, Britain,
Canada, India, Europe and Australia. No mention is made of box culverts in
this literature review. The authors claim that optimising the weight of the
structure will not necessarily produce the optimum design, since three
parameters greatly influence the final cost of the concrete structure: concrete,
steel and formwork. Therefore, the authors conclude that it is necessary to take
a more general approach when considering cost optimisation and that this
practice can result in significant savings.

Ignacio Martin has brilliantly stated in his discussion of the paper by Sarma and
Adeli (1998): “An experienced builder can erect a safe structure, but only
engineers can design economical safe structures” (Martín, Adeli & Sarma 1999).
The discusser also states that defining cost optimisation is not an easy task and
it should take into consideration parameters like function, availability of space,
life cycle, construction time and marketability, among others.

Stanton and Javadi (2014) developed a finite-element based least cost

optimisation Excel spreadsheet, ResOpt, that models optimum reservoirs using
genetic algorithm as a basis for the optimisation process. The authors show
how ResOpt produced cost savings of over 21% when utilised to model a 13Ml
reservoir in Cornwall, UK.

Page | 8
Stanton and Javadi (2014) allege there to be a recent trend toward optimisation
of structures that encompasses the life-cycle of a building, including the design
phase, construction, maintenance and demolition.

2.5 Size optimisation

Zhu et al. (2012) demonstrated that finite element analysis was successfully
utilised to optimise the four-sided box culvert structures built under a highway
in the Tuanbo Reservoir area, in Tianjin, China. The authors analysed dozens
of combination of sidewall thicknesses and baseplate thicknesses in ABAQUS
under vehicle loads as per the General Code for Design of Highway Bridges
and Culverts (China Ministry of Transport 2004). Using FEM analysis to
simulate stresses and deformations, the authors obtained the optimised culvert.
In this case, the optimum culvert was the one that met the stress and deflection
requirements of current bridge specifications in China and had minimum
weight. The reduced self-weight also resulted in decreased soil bearing capacity
requirements. The final structure was cast insitu and was composed of two 22
metres long sections with a width of 24.2 metres.

2.6 Shape optimisation

Rath, Ahlawat and Ramaswamy (1999) developed a design procedure that
optimises the shape of flexural members’ cross sections made of anisotropic
materials like reinforced concrete. The aim is to minimise total cost, which in
this case is made up of material, manufacture and placement costs. The authors
assert that if there is more material in high stress zones, the use of materials will
be more efficient and will result in savings. The procedure is exemplified in the
design of three types of beams: simply supported, cantilever and 2-span
continuous beams. Finite element modelling, natural velocity field method and
genetic algorithms were utilised.

2.7 Reinforcement optimisation

Aschheim, Hernandez-Montes and Gil-Martin (2008) propose simpler
approach to design optimum reinforced concrete beams, walls and columns

Page | 9
that does not require tables or interaction charts. The authors show the design
procedure of a reinforced concrete section under an axial force and a moment,
using nonlinear conjugate gradient search technique. A proposed single model
can then be used for beams, walls and columns. The authors’ approach can also
be integrated in widely available spreadsheet programs. By finding the optimum
design, which is this case means the design with minimum reinforcement and
minimum concrete, the authors claim to improve the sustainability of
reinforced concrete construction.

Gil-Martin et al. (2011) presented and proved a theorem they called TORS –
theorem of optimal section reinforcement. This theorem establishes which
cases of bottom and top reinforcement will result in minimum reinforcement,
using ACI-318-08 assumptions. The theorem states that the minimum total
reinforcement area occurs for one of the four following cases:

1. The bottom reinforcement area and/or the top reinforcement area is

zero

2. The strain at the bottom reinforcement (εS) is equal to or slightly

greater than the yield strain of the reinforcement (- εY)

3. The strain at the top reinforcement and bottom reinforcement are

equal to the maximum concrete strain of 0.003 (ε= εS= ε’S=
εC.max=0.003)

4. The strain at the top reinforcement (ε’S) is equal to the yield strain of
the reinforcement (- εY)

The authors also state that there is an infinite number of admissible

reinforcement solutions for each problem, but their proposed theorem enables
a quicker solution using optimum quantity of reinforcement without the need
for reinforcement sizing diagrams by evaluating the four cases above. The study
indicated that the optimum solution, the one that uses a minimum amount of

Page | 10
reinforcement, is significantly different than the typical symmetric
reinforcement solution shown in standards and textbooks.

2.8 Optimisation and Constructability

Guest and Moen (2010) employed topology optimisation methods to truss
analysis and developed an optimisation routine aimed at reducing crack widths
and enhancing member performance compared to traditional strut and tie
models. It allows engineers to visually examine designs and enables them to
identify the stiffest truss which describes the flow of forces in a general concrete
member with general loading and support conditions.

However, if constructability is not taken into consideration there is a great

chance the resulting design will not be achievable in practice. To correct that,
Zhu et al. (2014) propose that constructability measures should be inserted into
free-form topology optimisation as constraints and/or objective functions.
This will ensure that constructability, which according to the authors is typically
the primary governing cost in building a structure, is taken into consideration
and that the optimisation process yields results that can be applied in practice.

Guest et al. (2012) also defend that although topology optimisation can yield
valid design ideas, it is often prohibitively difficult to build these optimum
structures. To help mitigate the negative effects of difficult constructability, the
authors developed algorithms to: influence the constructability of systems and
manufacturability of components; utilise nonlinear material models to optimise
design and improve optimisation by considering fabrication or construction
errors or damage.

One way to improve constructability is to restrict the geometric design space.

(Guest et al. 2012) This has been done by Stromberg et al. (2011) using pattern
gradation and repetition. This means that restrictions are placed regarding
number and variable size of repeating patterns along any direction on the design
domain, which results in enhanced constructability.

Page | 11
Guest, Prévost and Belytschko (2004) also managed to enhance constructability
by restricting the diameter of the designed members. The authors used nodal
volume fractions as a design variable, making element volume fractions a
function of the nodal volume fractions.

It is also possible to regulate the maximum length scale of members. Guest

(2009) showed that by searching the design domain and applying local
constraints that will impede the development of features that are larger than a
required maximum, it is possible to improve constructability.

2.9 Using finite element analysis (FEA) to model RCBC behaviour

There are mainly two types of box culverts: three-sided, which do not have a
base slab, and four-sided, which have a box format. In Australia, the tree-sided
culvert is the type used in the great majority of construction projects.

Figure 2-1 - Four-sided box culvert (Foley Products 2014)

Page | 12
Figure 2-2 - Three-sided box culvert (Rocla 2014)

Many of the studies available that use FEA to model RCBC behaviour utilise
the four-sided culvert, since this type is vastly used overseas. Some of these
studies are now presented.

Awwad et al. (2000) performed a three-dimensional finite element analysis of

four-sided large reinforced concrete box culverts using the software SAP 2000.
The culverts analysed had spans in excess of 3.6 metres fill heights between 0
and 3 metres. Live loads such as AASHTO H20 truck were applied, as well as
overburden pressure, lateral pressure and bearing pressure.

McGrath, Liepins and Beaver (2005) performed three-dimensional analyses on

four-sided reinforced concrete box culverts with depths of fill up to 0.600m
subjected to live loads according to the American Association of State Highway
and Transportation Officials (AASHTO) load and resistance factor design
(LRFD) introduced in 1994.

Page | 13
A PhD thesis published in the University of Texas (Garg 2006) simulated
experimental tests done in four-sided RCBCs using the finite element
modelling software ABAQUS. The author used three-dimensional shell and
solid elements as well as welded wire fabrics to reproduce the behaviour of the
RCBC and its reinforcement in order to draw conclusions regarding the
appropriateness of the ASSHTO 2005 shear provisions across the culvert joint.
It was also concluded that the results obtained by the 3D FEM analysis in
relation to deflections corresponded with the experimental results.

Ahmed and Amanat (2008) argue that a two-dimensional analysis of four-sided

reinforced concrete box culverts, which is the basis of the design procedure in
Canada, is unable to realistically model the interaction between the buried box
culvert and the soil above it in deeply buried RCBCs. They affirm that a detailed
three-dimensional finite element analysis is required to enable an evaluation of
the stresses developed in the flow direction and consequently to realistically
model the box culvert interrelationship with the soil.

However, Awwad et al. (2008) contend that for four-sided box culverts with
spans of 3.6m, a plane frame analysis outputs less conservative moment and
deflection results than a three-dimensional finite element analysis. The authors
performed a parametric study on three sizes of culverts with spans of 3.6m,
5.4m and 7.2m. For fill depths under 0.9m, the wheel loading was found to be
dominant. However, for fill depths between 2.1m and 3m, the position of the
wheel along the midspan of the culvert slab was found not to yield considerably
different results with respect to earth loading. As for fill depths over 3.0m, it
was found that these results did not differ at all.

Kang et al. (2008) used the software programmes CANDE (Culvert ANalysis
and DEsign), ABAQUS and MSC/NASTRAN to investigate the effects of
frictional forces on the sidewalls of four-sided RCBCs.

Garg and Abolmaali (2009) used ABAQUS to simulate four-sided RCBC

behaviour and compared them with previously done experimental tests. The

Page | 14
finite element models used showed cracking propagation patterns were very
similar to those found in the experimental results. The box culverts used in this
study had the standard sizes according to ASTM-C-1433-04.

Chen, Zheng and Han (2010) used the commercially available geotechnical
finite element software PLAXIS to investigate factors that influence vertical
earth pressures onto four-sided culverts, including height of fill and dimensions
of the culvert. The reinforced concrete culverts were modelled as an elastic
material and the study concluded that the Chinese General Code for Design of
Highway Bridges and Culverts provide conservative methods to estimate earth
pressures on culverts.

PLAXIS was also used by Kim et al. (2011) to perform a finite element analysis
of a four-sided 1.8m x 1.8m reinforced concrete box culvert with an inlet
opening at the top. This culvert presented severe cracking and was about to
collapse in Georgia, USA. The results of the analysis were used to provide
repair alternatives to prevent complete failure.

Das (2013) utilised 3D-FEA to perform a refined load rating procedure on four
four-sided box culverts, three of which were built before 1940 while the fourth
was built in 1985. The author concluded that the results between the
conventional rating analysis, based on ASSHTO’s Allowable Stress and Load
Factor rating method, and his refined 3D-FEA could vary by more than 250%,
depending on the physical conditions of the culverts, field measurements and
load test data. Das (2013) states that the improvement brought by the 3D-FEA
method is due to appropriate use of a few factors including realistic live load
distribution obtained from 3D-FEA.

In contrast with the great amount of studies about four-sided box culverts,
there appears to be very few studies on three-sided box culverts. Frederick and
Tarhini (2000) pointed out that the American Society for Testing and Materials
(ASTM) did not discuss three-sided culverts. To fill this gap, the authors used
three-dimensional finite element analysis to analyse and design three-sided box

Page | 15
culverts with spans between 4m and 11m with less than 0.6m of fill and
subjected to live load, impact load, dead load and lateral earth pressure.
Frederick and Tarhini (2000) concluded that the requirements established by
ASSHTO and ASTM were met when analysing these structures using plane
frame analysis or 3D FEA, with the latter having the advantage that is gives
values for the transverse bending moments and shear forces, which in the case
of this study were very low.

FEA is also applicable to various other structures. Regarding pipes, for

example, Kitane and McGrath (2006) stated that although two dimensional
analysis was suitable for situations when the pipe culverts are deeply buried, it
may lead to conservative designs should the culvert be buried closer to the
surface and subjected to live loads.

2.10 Conclusions
The literature revealed a multitude of possible uses of FEA to model structural
behaviour. There is, however, a lack of research specifically on three-sided
reinforced concrete box culverts, which is the most common type of box
culvert found in Australia. By applying FEA to large box culverts in search for
the optimised structure, there will be great gain to the industry, to the end
customer and the environment.

This research will utilise the SIMP method due to its mathematical simplicity
since it does not require derivations including higher mathematics; its
computational efficiency due to the utilization of a single free variable per finite
element; and the fact that it does not require homogenization, only adjustment
of a suitable penalization factor.

Page | 16
 CHAPTER 3 - METHODOLOGY

3.1 Introduction
Various topology optimisation methods have been extensively studied.
However, the most popular manner to introduce the concept of topology into
structural analysis is via the Solid Isotropic Material with Penalisation (SIMP)
method. (Bruns 2005). This method assigns a value to the relative density of
each finite element in the domain, and penalizes the intermediate values
between 0 (void) and 1 (solid material) more heavily in order to generate solid-
void structural designs. Some advantages of this method are that it is
computationally efficient, it can be used for any combination of design
constraints and it is conceptually simple without requiring derivations involving
higher mathematics (Rozvany 2001). Due to these characteristics, the SIMP
method will be utilised in this project.

RCBCs which need to comply with MRTS24 comprise of approximately 40%

of the large box culvert sales where Roome (2014) currently works. The other
60% of culverts are for subdivision works, but they generate less sales volume
per each project. This is one of the reasons why this project focuses on Main
Roads culverts. Also, the most common box culvert sizes sold according to
Roome (2014) are between 1.8 m span by 1.5m leg (1815 RCBC) to 2.4 m span
by 1.8 m leg (2418 RCBC). That interval comprises a total of 5 box culvert sizes
out of the 24 possible sizes. Roome (2014) estimates these sizes make up
between 60% of all large box culvert sales. Due to the commercial significance
of these Main Roads RCBC sizes, they will each be investigated in detail.

To find the optimum RCBC, four main steps will be required. Firstly, the
SM1600 loads will be analysed and the worst case for each part of the culvert
will be found and taken as design load. Then, the standard RCBC will be
designed according to AS1597.2-2013 and MRTS24 (Aug/11) to be later
compared with the optimum RCBC. The next step will be to find the optimum

Page | 17
topology for the RCBC under the design loads utilising the SIMP method and
finite element analysis. Lastly, a feasibility analysis will be carried out to outline
the benefits and drawbacks of utilising this optimisation procedure in the
industry. These four steps are discussed in the next sections.

3.2 Analysis of SM1600 loads

The live loads for road bridge design in AS5100.2 (Bridge Design code) are
referred to as the SM1600, which are made up of the W80, A160, M1600, and
S1600 and M1600 design loads. (Standards Australia 2013, p. 58). These loads
represent road traffic design loads for main and secondary roads, which are
commonly specified in the industry. According to AS1597.2 (Large RCBC
code), large reinforced concrete box culverts are to be designed for W80, A160
and M1600 as per AS5100.2 but excluding the uniformly distributed load
component from the M1600 load (Standards Australia 2013, p. 27). Heavy load
platform loads (HLP320 and HLP400) and railway loads (300LA) are not
included in this study.

3.2.1 W80 load

The W80 load represents an individual heavy wheel load that uniformly
imposes 80kN distributed over a contact area of 400 mm x 500 mm for the
strength limit state and 200 mm x 500 mm for the serviceability limit state
(Standards Australia 2013, pp. 28-31) (Standards Australia 2004a, p. 12).

Page | 18
 Figure 3-1 - W80 wheel load

Figure 3-1 shows the truncated prism representation for a W80 wheel load
distributed through fill. The top blue rectangle represents the wheel contact
area. The dashed red lines represent the truncated prism and the bottom
hatched rectangles represent the area over which the pressure is distributed on
top of the RCBC.

3.2.2 A160 load

The A160 load represents an individual heavy axle load that uniformly imposes
160kN distributed over a contact area of 400 mm x 500 mm for the strength
limit state and 200 mm x 500 mm for the serviceability limit state The standard
design lane size is 3200 mm and the distance between the two wheels in the
axle is 2000 mm. (Standards Australia 2013, pp. 28-31) (Standards Australia
2004a, pp. 12-3).

Page | 19
 Figure 3-2 - A160 axle load

Figure 3-2 shows the truncated prism representation for an A160 axle load
distributed through fill.

3.2.3 M1600 load

The M1600 load models two heavy vehicles in the same lane together with an
accompanying stream of general traffic (Standards Australia 2007). Each heavy
vehicle of M1600 has two tri-axles, one of which is represented in Figure 3-3,
with the red lines representing the truncated prism load distribution through
fill. Each axle is 1.25m from the other.

Figure 3-3 - M1600 tri-axle load

Figure 3-4 shows the M1600 moving traffic loads, with all dimensions in mm.

Page | 20
 Figure 3-4 - M1600 according to AS1597.2

According to AS1597.2-2013, the minimum roadway load class to be

considered is Class 2-A, which means the amount of fill considered is to be
from 0m to 2m of fill inclusive. (Standards Australia 2013, p. 13). The road
traffic loads are to be applied for the entire range of fill from 0m to 2m inclusive
(Standards Australia 2013, p. 27). However, AS1597.2 also dictates the
minimum fill directly above the top of the culvert (overlay) is to be at least 150
mm (Standards Australia 2013, p. 9). To be conservative, the calculations in this
project will go from 0.1 m to 2.0 m of fill.

The distribution of loads through fill is calculated by using a truncated prism

type approximation, in accordance to AS1597.2-2013 Clause 3.3.5.5.2
(Standards Australia 2013, p. 58). Graphic examples of the truncated prism can
be seen in Figure 3-1, Figure 3-2 and Figure 3-3.

For the serviceability limit state, the W80 single wheel load pressure area is
given by A = L1 L2 = (b + 1.45H )(a + 1.45H ) where

a=0.2 m

b=0.5 m

Page | 21
H = height of fill over RCBC in metres

It can be seen in Figure 3-3 that for the tri-axle case, the loads will overlap
(shown in green). In this case, AS1597.2 stipulates the distribution area will still
be even but will be given by A = L1 L2 = (G + b + 1.45H )(J + a + 1.45H ) where
all parameters remain the same and

G=distance between wheels in metres

J=distance between axles in metres

For the ultimate limit state, the W80 single wheel load pressure area is given by
A = L1 L2 = (b + 1.15H )(a + 1.15H ) where

a=0.4 m for W80 and A160 and a=0.3 for M1600

b=0.5 m

H = height of fill over RCBC in metres

Where the loads overlap, the distribution area will still be even but will be given
by A = L1 L2 = (G + b + 1.15H )(J + a + 1.15H ) where all parameters remain
the same and

G=distance between wheels in metres

J=distance between axles in metres

When designing the culvert according to AS1597.2-2013, the maximum

superimposed load case should be used as the design basis. (Standards Australia
2013, p. 57). The critical load combination will be determined according to
AS1597.2-2013 Clause 3.4, which dictates culverts are to be designed to resist
loads imposed onto them during intermediate and final stages of construction.
(Standards Australia 2013, p. 32). The culverts will be subjected to both
construction loads and in-service loads and both in the horizontal and the
vertical directions.

Page | 22
The vertical loads applicable to this study are:

• WDC: self-weight
• WFV: vertical earth pressure due to fill
• WCV: construction live load induced vertical earth pressure
• WLV: roadway live load induced vertical earth pressure

The horizontal loads applicable to this study are:

• WFH: horizontal earth pressure due to fill

• WAH: horizontal pressure due to compaction
• WCH: construction live load induced horizontal earth pressure
• WLH: roadway live load induced horizontal earth pressure

As mentioned earlier, heavy load platform loads (HLP320 and HLP400) and
railway loads (300LA) are not included in this study.

3.2.4 Load factors

The load factors dictated by AS1597.2-2013 for stability and strength limit
states are shown in Table 3-1 and Table 3-2.

Table 3-1 - Stability and strength limit state load factors for vertical loads

Alternative
Load
Load Load
Factor
Factor

WDC self-weight 1 -

WFV vertical earth pressure due to fill 1.4 0.9

construction live load induced

WCV 1.5 0
vertical earth pressure
roadway live load induced
WLV 1.8 0
vertical earth pressure

Page | 23
 Table 3-2 - Stability and strength limit state load factors for horizontal loads

Symmetric loading Asymmetric loading

Load Alternative On one On opposite

Load
Factor Load Factor side side
horizontal earth
WFH 0.7 1.4 1.4 0.7
pressure due to fill
horizontal pressure
WAH 0.7 1.4 1.4 0.7
due to compaction
construction live load
WCH induced horizontal 1.5 0 - -
earth pressure
roadway live load
WLH induced horizontal 1.8 0 - -
earth pressure

Some loads have two possible load factors: one higher than unity and one lower
than unity. When a change in the situation being analysed (like an increase in
the load) decreases safety, a load factor greater than unity is used and when it
increases safety (like a decrease in the load), the load factor is smaller than one.

3.2.5 Self-weight WDC

The vertical loads to be considered will obviously always include the culvert
self-weight, which has a load factor of 1. According to section G2 of Standards
Australia (2013), in the absence of more specific material information, the self-
weight should be calculated assuming a reinforced concrete density of
2650kg/m3 and a gravity force per unit volume of 26.0kN/m3. The volume of
each culvert was calculated in AutoCAD and the mass was found utilising the
above density.

3.2.6 Vertical earth pressure due to fill WFV

In embankment installation conditions, the side zone material shall extend out
horizontally for a width equal to one-third of the height of the culvert or a
minimum width of 300 mm, whichever is greater (Standards Australia 2013, p.

Page | 24
42). The vertical earth pressure due to fill (WFV) for embankment installation is
H
measured in kPa and is given in AS1597.2-2013 by WFV = (1 + 0.2 )γ H
Bc

where H is the height of fill over the culvert, from 0.0 m to 2.0 m

Bc is the is the overall outside width of the culvert in metres

γ is the gravity force per unit volume of the fill material, assumed
20kN/m3

There are two possible load factors to be used. In terms of vertical earth
pressure due to fill, a situation with a small height of fill would act beneficially
to dissipating the live and construction loads on top of the culvert, therefore
the most appropriate load factor would be 0.9. However, as fill depths increase,
their beneficial action to dissipating loads on top of the culvert is countered by
the pressure the greater amount of fill actually puts on top of the culvert. In
this situation, the most appropriate load factor is 1.4.

3.2.7 Construction live load induced vertical earth pressure WCV

The induced vertical earth pressure caused by the construction live loads is to
be taken into consideration during the intermediate stages of construction and
it is to be applied at 0.4m of fill or at the final fill height, if less than 0.4m.
(Standards Australia 2013, p. 28) Depending on the depth of fill, the critical
case will vary between the W80, A160 and M1600 load cases. That is because
the shallower the depths, the less room there is for load distributions from
multiple axes based on the truncated prism to fully overlap.

3.2.8 Vertical loads due to road traffic loadings WLV

The effect of roadway live load induced vertical earth pressure is to be
calculated by dividing the sum of the wheel loads applying pressure to the
culverts by the area of application, based on the truncated prism method. A
dynamic load allowance (DLA) is also applied and it varies linearly from 0.4 at

Page | 25
0m of fill to 0.1 at 2.0m of fill. (Standards Australia 2013, p. 30). This translates
∑P
to WLV = (1 + DLA) ⋅
A

where DLA is dynamic load allowance from 0.4 to 0.1

∑ P is the is sum of the individual wheel loads applying pressure to

the culvert in kN

A is the area of the truncated prism base in m2

There are five options for the critical case of live loads due to traffic:

• W80 load on single lane

• A160 load on single lane
• M1600 load on single lane
• A160 load on dual lane
• M1600 load on dual lane

The case with W80 wheel load on a dual lane is not considered because it
produces localized effects and therefore is not appropriate for dual lane
(Standards Australia 2013, p. 58).

AS1597.2 brings the critical cases for SM1600 in table G3, which are used in
this study (Standards Australia 2013, p. 62). For fill depths up to 1.2 m the
critical case is the single lane W80. From fill depths of 1.3 m to 2.0 m, the
critical case is the dual lane A160.

3.2.9 Horizontal loads due to fill and compaction WFH and WAH
When it comes to horizontal loads due to fill and compaction, there are two
situations to be analysed: when the load on both sides of the culvert is the same
(symmetric loading) and when they differ one from the other (asymmetric
loading).

In the symmetric case, there are two possible load factors, one greater and one
lower than unity. This is due to the fact that higher horizontal forces generated

Page | 26
by fill and compaction act favourably to strengthening the culvert crown in
bending, since they counter the moment generated at the edge of the crown. In
this situation, the most appropriate load factor would be 0.7. In contrast, it is
possible that the fill will not act favourably to strengthening the culvert in
bending, for example in case of poor compaction, when the culvert crown
bends more freely without as much restraint from fill and compaction. Then,
the appropriate load factor would be 1.4.

Regarding asymmetric loading, it is also required to check for the worst

combination. A check is required with the lower load factor of 0.7 on one side
of the culvert and the higher load factor of 1.4 on the other side and also with
these reversed.

3.3 Designing the standard RCBC

The culverts were designed utilising linear structural analysis combined with
ultimate strength theory, as dictated in the Concrete Structures code AS3600
(Standards Australia 2009, p. 28). It is known that in practice when reinforced
concrete structures are subjected to loads they do not behave linearly. However,
the Australian Standard Codes followed in this study not only permit the use
of linear analysis but also impose the use of safety coefficients at several stages
of design. Since the objective of this study is not to analyse structural failure,
the use of linear analysis and safety coefficients is deemed sufficiently accurate.

According to the Main Roads Standard Specification MRTS 24 (06/09) and

AS1597.2-2013, the culverts are to be designed as portal frames and the
supports shall be modelled as pins at the base (Transport and Main Roads
2010a, p. 3) (Standards Australia 2013, p. 24). Also, sidesway does not need to
be taken into consideration if the culvert is installed according to AS1597.2-
2013 Section 5. (Transport and Main Roads 2010b) (Standards Australia 2013,
p. 34). It is therefore assumed in this study that culverts are installed according
to AS1597.2-2013 Section 5.

Page | 27
AS1597.2 describes the preferred internal dimensions of large RCBCs, which
are normally observed by manufacturers. As the intention of this project is to
generate results and conclusions that can be applied in practice, the culverts
were modelled to have the same internal dimensions as described in AS1597.2.

Table 3-3 - Preferred RCBC internal dimensions

Nominal Nominal
Size class
span height
mm mm mm
1500 x 900 1500 900
1500 x 1200 1500 1200
1500 x 1500 1500 1500

1800 x 1200 1800 1200

1800 x 1500 1800 1500
1800 x 1800 1800 1800

2400 x 1200 2400 1200

2400 x 1500 2400 1500
2400 x 1800 2400 1800
2400 x 2400 2400 2400

3000 x 1200 3000 1200

3000 x 1800 3000 1800
3000 x 2400 3000 2400
3000 x 3000 3000 3000

3600 x 1200 3600 1200

3600 x 1800 3600 1800
3600 x 2400 3600 2400
3600 x 3000 3600 3000
3600 x 3600 3600 3600

4200 x 1800 4200 1800

4200 x 2400 4200 2400
4200 x 3000 4200 3000
4200 x 3600 4200 3600
4200 x 4200 4200 4200

Page | 28
The general design requirements described by AS1597.2 are that culverts are to
be designed to satisfy stability, strength, serviceability and durability limit states
(Standards Australia 2013, p. 24).

3.3.1 Materials
The concrete utilised has an assumed Poisson’s ratio of 0.2, in accordance with
AS3600-2009 Clause 3.1.5. Also, clause 3.1.3 of the same standard stipulates
the density of normal-weight concrete is to be taken as 2400kg/m3, unless
specific laboratory results are available. (Loo 2010, pp. 13-4) The concrete is
assumed to have characteristic strength of 50MPa as per MRTS24 clause 10.7
and therefore have a Young’s Modulus of 34800MPa.

The steel reinforcement is assumed class N deformed bar (designation D500N)

with yield stress of 500 MPa. The elastic modulus of the steel reinforcement is
assumed to be 200 GPa for both tension and compression (Foster 2010, p.
532).

3.3.2 Durability Design

MRTS24 states that large box culverts shall be designed for a minimum
exposure classification of B2 in accordance with AS 5100. This standard caters
for reinforced concrete structures and members with a design life of 100 years.
The exposure classification B2 is appropriate for surfaces of members in above-
ground exterior environments in coastal areas in any climatic zones. This means
the culverts can be up to 1km from the coastline but not in tidal or splash zones.
Members can also be permanently submerged in sea water (Standards Australia
2004b, p. 29). Table 4.5 from AS5100.5 shows the requirement for at least 25
MPa compressive strength at the completion of accelerated curing, such as
steam curing, which is commonly used in the precast industry. Also, the
minimum strength of concrete to be utilised is to be 40 MPa. This means the
assumed 50 MPa concrete in this study complies as long as the product achieves
at least 25 MPa during accelerated curing.

Page | 29
Abrasion also needs to be taken into consideration since the culvert is being
designed for 0.0 m of fill. For medium or heavy pneumatic-tyred traffic, the
minimum compressive strength required by AS5100.5 is 32 MPa, and for non-
pneumatic-tyred traffic the minimum is 40 MPa. Again, the assumption made
in this study is compliant.

The cover to reinforcing steel must be suitable for both the placement of
concrete and for the protection of reinforcement against corrosion. For
concrete placement, the cover shall not be less than the maximum between 1.5
times the maximum nominal size of the aggregate and the diameter of the
reinforcing bar. In this study, it is assumed that the aggregate nominal size is 20
mm and the maximum bar diameter utilised in the culvert reinforcement is N28
(28 mm diameter), the cover should not be less than 30 mm (1.5 x 20 mm =
30 mm).

Because it is common in the precast industry to utilise rigid formwork such as

rigid steel forms and intense compaction obtained with vibrating tables or self-
compacting, super workable concrete, the nominal cover for 50 MPa concrete
subject to B2 exposure classification is 35 mm (Standards Australia 2004b, p.
34), with a tolerance of -5, +10 mm. This means the minimum cover to
reinforcement has to be 30 mm and the maximum cover has to be 45 mm,
which is in accordance with the stated conditions for cover for concrete
placement.

In summary, for the culvert to comply with durability requirements, it needs to:

• Have cover to reinforcement between 30 mm and 45 mm

• Utilise rigid formwork and intense compaction
• Have aggregate nominal size of no more than 20 mm
• Utilise reinforcement bars of the class D500N with a maximum
diameter of 28 mm

Page | 30
 • Be installed in conditions suitable for B2 exposure classification i.e. not
in tidal or splash zones

3.3.3 Stability and Strength Design

According to AS1597.2 clause 3.6.1, design for strength shall be in accordance
with AS3600, which dictates the allowable strength checks and methods of
structural analysis. In this study, a strength check procedure for linear elastic
methods of analysis and ultimate strength theory will be utilised. It is required
that the design capacity of the cross section being considered is greater than
the design action effects. When that concept is applied to a cross section of the
crown or leg or the box culvert in bending, it yields φ M u ≥ M * where

φ is the capacity reduction factor

Mu is the moment capacity of the section
M* is the design ultimate moment

The design for stability shall comply with AS1597.2-2013 clause 3.4, which
dictates the load combinations to be applied to culverts, as explained in Section
3.2.

3.3.4 Serviceability Design

According to AS1597.2 clause 3.6.2, design for durability shall be in accordance
with AS5100 or AS3600. In AS3600, the key aspects of serviceability design are
concerned with deflections and cracking of concrete (Foster 2010, p. 89).

Deflection was considered by using the simplified calculation for slab

deflection described in AS3600-2009 section 9.3.3. A prismatic beam of unit
width was the equivalent structure utilised. According to AS3600-2009 section
 ∆ 
2.3.2, the deflection limitation on the crown   , which is subject to
 Lef
 
vehicular traffic, is to be less than 1/800. For the legs, the lateral deflection shall
not exceed 1/500 of the leg height.

Page | 31
Shrinkage and temperature effects play an important role in concrete cracking
and to control these effects distribution reinforcement must be provided in box
culvert the crown and legs with a maximum bar spacing of 300 mm and a
minimum area of 150 mm2/m measured in the direction of the main flexural
reinforcement (Standards Australia 2013, p. 35).

3.3.5 Non-optimised Culverts

To design the non-optimised culverts, the moment and shear capacity are
determined using the finite element analysis software Strand7. The boundary
conditions are introduced by restraining the nodes at the bottom of the culvert
to model pins. In Strand7, the nodes have three translational and three
rotational degrees of freedom (Strand7 Pty. Ltd. 2010). By fixing all but the
rotational degree of freedom in the Z direction, which is the direction of the
length of the culvert of 2.4 m, the boundary conditions are implemented by
modelling pins as required by Transport and Main Roads (2010a, p. 3) and
Standards Australia (2013, p. 24).

The crown and legs of the RCBC are modelled in Strand7 as beams and the
vertical and horizontal loads are applied to them as distributed loads based on
the critical loads determined in Section 3.2. The linear static analysis then yields
the bending moment and shear force diagrams.

3.3.6 Reinforcement
The design process used to design the RCBC is iterative. Firstly an initial
assumption is made regarding the thickness of the crown and leg of the
structure. Then, the reinforcement is determined according to the procedures
in AS3600 (AS1597.2 clause 3.5.1), with the number and diameter of required
bars found iteratively with the Matlab program flexanalysis.m (see
Appendix B). If the section does not have enough capacity, it is thickened and
the process starts again.

Page | 32
The concept of equivalent concrete compressive stress block is utilised for
flexure analysis by the Matlab program flexanalysis.m, described by two
parameters:

Equation 3-1

γ = 1.05 − 0.007 f 'c within the limits of 0.67 ≤ γ ≤ 0.85

Equation 3-2

α 2 = 1.0 − 0.003 f 'c within the limits of 0.67 ≤ α 2 ≤ 0.85

These parameters dictated by AS3600-2009 ensure the total volume of the

stress block is the same as the total volume of the equivalent stress block and
that the centroid of the two blocks is also at the same height, as shown in Figure
3-5.

Figure 3-5 - Compressive stress block

The value for the extreme fibre concrete strain is adopted in AS3600-2009 as
ε cu = 0.003 and α 2 = 0.85 for f 'c ≤ 50MPa .

Page | 33
Therefore, the forces calculated in the Matlab program flexanalysis.m are
derived from the conditions at Mu (ultimate bending capacity) as shown in
Figure 3-6.

Figure 3-6 - Ultimate bending capacity conditions

These forces are:

Equation 3-3 - Forces at Mu

T = σ s Ast

Cc = γ d n bα 2 f 'c

Cs = Esε sc Asc

The Matlab program flexanalysis.m (see Appendix B) then calculates dn

so that Cc+Cs=T, which is a condition for equilibrium. It then checks that
dn
ku = < 0.36 to ensure the section is under-reinforced and calculates
d
M u = Cc ( d − d c ) + Cs ( d − d sc ) and φ M u , with φ = 0.8 for bending.

All details of reinforcement such as spacing, extensions and termination of

reinforcement shall comply with AS3600-2009.

The development length for deformed bars in tension utilised in this study is
the basic one described in AS3600-2009 Section 13:

Page | 34
 0.5k1k3 f sy d b
Lsy .tb = ≥ 29k1db
k 2 f 'c

where

k1= 1.3 for a horizontal bar with more than 300 mm of concrete cast
below the bar or

= 1.0 otherwise

k2= (132-db)/100 and

k3= 1.0 − 0.15(cd − db) / db (within the limits 0.7 ≤ k3 ≤ 1.0); where

cd = minimum between the distance between parallel bars and the

cover to the bar from the tension face

In addition, according to AS1597.2 section 3.7.2, the minimum flexural

reinforcement shall not be less than 0.002Ag in span direction, where Ag is the
gross concrete cross-sectional area.

A shear check is also carried out to check if shear reinforcement is required and
for shear design of RCBCs, Clause 8.2.7 of AS3600-2009 is normally applicable
(Standards Australia 2013, p. 35). It dictates that the ultimate shear strength
(Vuc) excluding the contribution of shear reinforcement is given by

1
 A 3
Vuc = β1β 2 β 3bv d 0 f cv  st 
 bv d 0 

where

β1= 1.1(1.6-d0/1000)≥0.8

β2= 1 for pure bending

β3= 1

fcv= f’c(1/3)≤ 4 MPa

Page | 35
 Ast = cross-sectional area of longitudinal reinforcement provided in the
tensile zone and fully anchored at the cross-section under
consideration

d0 = the distance from the extreme compressive fibre to the centroid

of the most tensile reinforcement

bv = effective width of the web

According to AS3600-2009, there are three possible cases in relation to shear

reinforcement:

1. clause 8.2.5 (a): if V * ≤ 0.5φVuc no shear reinforcement is required

except where the overall depth of the beam exceeds 750 mm, in which
case minimum shear reinforcement shall be provided.

2. clause 8.2.5 (b): if 0.5φVuc < V * ≤ φVu.min only minimum shear Asv.min
is required. Also according to the same standard, clause 8.2.5 (ii), if
V * ≤ φVuc the minimum shear reinforcement requirement may be
waived. According to clause 8.2.8:

Equation 3-4

Asv.min = 0.06 f 'c bv s / f sy. f ≥ 0.35bv s / f sy . f

where:

f’c is the concrete compressive strength, in this case 50 MPa

bv is the effective width of the web for shear
s is the centre-to-centre spacing of shear fitments
fsy.f is the characteristic yield strength of the reinforcement used as
fitments

3. clause 8.2.5 (b): if V * > φVu.min shear reinforcement is required

according to clause 8.2.10.

Page | 36
All these conditions are checked by the Matlab program checkshear.m (see
Appendix C for code).

3.4 Finding the optimum topology using the SIMP method

The aim of topology optimisation is to determine the optimum layout of a
structure subject to specific loads within a specific design domain. The topology
optimisation method utilised in this study is the Solid Isotropic Material with
Penalization, also known as SIMP or power-law method. The design variables
are the relative densities of each finite element, and they relate to the material
property via the power-law. The power law dictates that a relative change in
one quantity results in a proportional relative change in the other quantity risen
to a power.

The known quantities at the start of the optimisation will be the applied loads
as determined in Section 3.2; the support conditions, which are assumed pinned
(Transport and Main Roads 2010a, p. 3) (Standards Australia 2013, p. 24); the
final volume of the structure; and the location and size of prescribed openings.

The loads will be applied on the top and sides of the culvert, as described in
Section 3.2. However, to minimise computation time, only half of the culvert
was modelled in the symmetric case and the culvert crown was assumed to be
supported on rollers.

The final volume of the structure is one of the constraints of the topology
optimisation method utilised in this study. The topology optimisation script will
stop only when the total volume of the final structure is as required by the
volume constraint, and when the variance of the relative densities is smaller
than 0.5%, which is the chosen accuracy for the convergence criterion. Once
the topology optimisation process finishes, the culvert with voids is modelled
in Strand7 and the moment and shear capacity are determined. The
reinforcement is placed as required in AS3600 and the void size and shape can
be modified to allow for the placement of reinforcement within the specified
cover. In this case, because the culverts are to comply with MRTS24, the

Page | 37
minimum exposure classification is B2, which means the nominal cover is to
be 35 mm (Standards Australia 2013, p. 16). This means that if the bar diameter
is 20 mm, for example, the distance between the edge of the void and the edge
of the culvert needs to be at least 35+35+20=90 mm.

Within the domain, an area representing the culvert nominal opening was
assigned a relative density of 0.001, to ensure there is no material in the opening.
This represents the second constraint.

The objective function is to determine the minimum vector of relative densities.

The aim of the method is to find a topology with as many densities equal to 1
or 0, meaning no intermediate densities are desired since in this study only
concrete and steel reinforcement will be utilised. The intermediate densities are
penalised by being elevating to a power p. If p is too low or too high it can
cause too many finite elements with intermediate densities or too fast a
convergence to local minima (Sigmund & Maute 2013). Bendsøe and Sigmund
(1999) claim that a p>3 will give results that have a physical meaning. The
authors explain that if p can be modelled as a material if it complies with

Equation 3-5

 2 4 
p ≥ max  , 
1 − v 1 + v 

where v is the Poisson ratio of the solid material.

However, this project will only utilise steel reinforcement and concrete and it
will not attempt to model composite materials. Since the Poisson’s ratio of the
concrete utilised is 0.2 (as per AS3600-2009 Clause 3.1.5), Equation 3-5 would
then become

Equation 3-6

 2 4 
p ≥ max  ,  ∴ p ≥ max {2.5,3.33}
1 − 0.2 1 + 0.2 

Page | 38
Trying a penalisation factor of 4 resulted in non-convergence, so a penalisation
factor of 3 was chosen instead, which gave good results.

The SIMP method was implemented by utilising a Matlab script, which is

explained in detail in the next subsection.

3.4.1 Matlab script: top_rcbc4.m

To design the optimised culverts, a Matlab script called top_rcbc4.m, which
was adapted from Sigmund (2001), was utilised. The adapted program can be
found in Appendix F.

Firstly the material is uniformly distributed through the design domain, which
is assumed to be rectangular with the finite elements assumed square like a
quad4 element from Strand7. The scale of the real size to the modelled size
culvert can vary, but in the majority of cases a scale of 25 proved sufficient.
This means that each square finite element side represents 25 mm of the real
culvert size. When the scale was diminished, the computation time was greatly
increased to unpractical times without significantly improving the result,
proving ineffective.

Because the box culverts need to have a certain size opening, a range of finite
elements is made passive by changing their relative density to 1E-3. This means
there is a void, not an element. If these relative densities were to be changed to
zero it would result in a matrix singularity, hence the densities are changed to a
very small number ie 1E-3.

Then the finite element analysis is performed to find the displacement vector
U. To achieve that, the element stiffness matrix is generated, utilising the
Young’s modulus E and Poisson’s ratio ν previously input into the code, which
is this case translates into E=34800 MPa for our chosen 50 MPa concrete with
ν=0.2. The global stiffness matrix can then be assembled by looping through
all elements and by utilising element node numbers described as global element
numbers to ensure correct placement of elements in the global stiffness matrix.

Page | 39
The next step is to apply forces to the RCBC model. The forces applied are
those found by the load combination program, as described in more detail in
Section 3.2 - Analysis of SM1600 loads. The force applied in the program as
written by Sigmund (2001) was a unit concentrated force at the edge of the
design domain. This was changed in the adapted code to model the loads the
RCBCs were subjected to. The vertical load on top of the crown is always a
uniformly distributed load (UDL), so the point force was changed to a vector
to model this. The side load was approximated to a UDL to facilitate
implementation in Matlab.

Subsequently, the support conditions must be set up. Every element has two
degrees of freedom, namely horizontal and vertical. To implement a support,
these degrees of freedom are eliminated from the linear equations to model the
constrained degrees of freedom. The unconstrained degrees of freedom are the
difference between all degrees of freedom and the fixed ones.

After that the objective function, which is the minimum vector of relative
densities, is found by applying finite element analysis principles, yielding

Equation 3-7 - Optimisation objective function

N
min : c(x) = UT KU = ∑ ( xe ) p ueT k0ue
x
e =1

where

x is the vector of relative densities, the design variables

xmin is the minimum of the vector of relative densities, with non-zero values
U is the global displacement vector
K is the global stiffness matrix
ue is the element displacement vector and
ke is the element stiffness matrix

Page | 40
It can be seen that the objective function is found by multiplying the global
force matrix, which is F=KU, by the transposed global displacement matrix
UT. The summation displayed in the right side of Equation 3-7 is then
implemented in the Matlab script.

Sigmund (2001) claims that it is possible to improve the likelihood of the

existence of solutions by implementing a filtering scheme, which is the next
step in the Matlab script. The filter modifies the element sensitivities by using
∧
a convolution operator (weight factor) H f = rmin − dist (e, f ) in which rmin is
the filter size divided by the element size and dist(e,f) is the distance between the
centre of element e to the centre of element f. Sigmund (2001) warns that the
filter does not guarantee the existence of solutions, but it has been tested by
the author in various applications with positive results.

The design variables stored in the x vector are then updated using the optimality
criteria method. To do this, the value of the Lagrange multiplier that satisfies
the volume constraint chosen by the user when the function is called is found.
The bi-sectioning method is utilised to achieve this, since the material volume
is a monotonously decreasing function of the Lagrange multiplier (Sigmund
2001).

Each vector of design variables is then printed as an image in turn using a black-
white colour map, in which black means relative densities of 1 (presence of
material) and white means relative densities of 1E-3 (voids). Areas with grey
colour would indicate a composite material of intermediate density.

Each iteration summary is also printed on the screen with the iteration number,
the objective function value, the fraction of the initial volume, the convergence
criterion and the time the iteration was performed.

With this information it is possible to identify where the voids should be and
how the optimisation was performed.

Page | 41
The program is run in Matlab by calling it from the prompt line with:

[x,U]=top_rcbc4(span,legheight,volfrac,penal,rmin)

In square brackets are the program’s outputs, which will be the vector of
relative densities, x, and the global displacement matrix U. The inputs are in
parenthesis:

span is the RCBC span in mm

legheight is the RCBC leg height in mm
volfrac is the volume constraint ie 30% is entered as 0.30
penal is the penalisation factor
rmin is the filter size divided by the element size

3.5 Feasibility analysis

Once the optimum culvert has been found, an analysis was be carried out to
ascertain its constructability and commerciality by investigating the level of
efficiency and cost savings during design, manufacture and installation.

The cost information was gathered by interviewing Mr. Roome, an Engineered

Solutions Manager with over 20 years private industry experience and vast
knowledge of RCBCs.

3.5.1 Production Costs of an RCBC

When performing a cost estimation of a given RCBC, the factors taken into
consideration are:

• Concrete materials: cement, aggregate, water

• Reinforcement materials: steel
• Labour: preparing steel cage, casting procedures, loading procedures,
quality assurance (QA) checks
• Overhead costs: plant, asset depreciation, maintenance, staff rates
• Design costs

Page | 42
The customer selling price will then be this total production cost plus a profit
margin. According to Roome (2014), the private industry normally offers
delivery as a service to the customer and small margins are added to delivery to
cover the administration costs regarding its organisation. Alternatively, the
precast product can be picked up from the factory, which is called ‘ex-works’
and does not involve extra costs. Delivery will therefore not be included in this
feasibility analysis.

3.5.2 Estimating Procedure

There are various ways to measure costs. Materials, for example, are normally
measured in $/tonne or $/m3. Labour and overhead costs are normally
measured in man-hours/tonne. For instance, in the case of labour, if the cost
is 4 man-hours/tonne and the product weighs 1 tonne, there were 4 hours of
labour activity involved to produce it. This activity includes setting up the
moulds, producing the steel cages, casting the product, the curing procedure,
loading and checking. Design is normally charged in $/hour.

To estimate the cost of a product, all these costs need to be taken into
consideration and transformed into the same currency ie dollars. The following
sections will look at each cost component in detail.

3.5.2.1 Design
The cost of design will vary from company to company. In this study, it is
assumed a design engineer with a couple of years’ experience will design the
culverts, utilising software as it is common in the industry. To estimate the
design cost to be input into the RCBC cost estimate, let us assume this engineer
earns $80000 per year and works 38 hours per week. That would give the
company a cost of approximately $40.50 per hour to pay for this engineer’s
salary. However, for the engineer to design the box culverts, it needs an office,
computers, software and the cost to maintain all this and the depreciation of all
this needs to be taken into account. In this study, it is assumed the cost of an

Page | 43
engineer’s hour to design a box culvert, in total, is approximately $80/hour,
utilising design software.

A design is produced according to the required specifications and it yields the

product mass and reinforcement content, which are the input in the cost
analysis. More routine designs are done more quickly and special designs or
non-routine requirements will increase design time.

3.5.2.2 Materials
The cost of materials, according to Roome (2014), does not vary too much
since the production procedures are standard throughout the industry and a
supplier cannot generally get the same quality product for a very different price.
Mr. Roome believes the cost of concrete is around $140/m3 and the cost of
reinforcement steel is $1300/tonne. These values will be utilised to estimate the
cost of materials.

3.5.2.3 Labour costs

On average, one hour of labour costs around $40. The most significant
component of a large box culvert price is labour, since it is the one that can
vary the most and that is significantly large compared to other costs. This means
that, in the precast industry, one of the most effective ways to save on
production costs is to save on labour costs. This is achieved by simplifying
procedures like casting and reinforcement cage manufacture and augmenting
their level of repetitiveness.

One way to achieve this is by maximising the amount of units made in the same
size. That is because making a lot of units utilising the same mould setup will
spread the cost of setting up that mould onto more units, with the setup cost
for each unit decreasing. For example, a large RCBC job normally consists of
an average of 600 metres worth of box culverts. Large box culverts are normally
sold in 2.4 m lengths, since it is more efficient to produce them than the 1.2 m
lengths. This is due to the fact that the reinforcement cage, mould setup and
casting procedures have to be done once only to product 2.4 m of product,

Page | 44
while they would have to be done twice to produce the same length using a 1.2
m long mould.

Roome (2014) advises that in his experience, the private industry estimates
costs by calculating the quantity of man-hours required to product a tonne of
product. For example, the price of large box culverts in Mr Roome’s experience
is around 2.5 man-hours per tonne, if the design is standard, without additions,
voids or special requirements and if the culvert is transported legs down.

The loading and transportation procedure is different for culverts with legs up
to 2.1 metres and those with taller legs. If the product’s leg is up to 2.1 m long,
it can generally be transported legs down on the truck. This means there is no
rotation involved in demoulding, loading, unloading and installing.

If the culvert leg is taller than 2.1 m a design analysis will have to be carried out
that takes into consideration the fact that the centre of gravity of the product
will be higher in the truck and that creates a much higher risk for transportation.
The transportation design analysis generally yields one of two possible
solutions: either the culvert is transported upside down, with the crown on the
truck bed, or it is transported on its side. Either of these will incur extra costing
related to design, labour for the extra rotations and extra lifters setup required,
as well as longer loading times.

3.5.2.4 Overheads
The term overheads refers to the costs of operating a business. They include
plant depreciation and maintenance, rent, water, electricity, insurance,
employees’ salaries, payroll taxes, employee pension costs and other employee
benefits.

Roome (2014) estimates these costs to be around $70/tonne at present at his

place of employment. This will obviously vary depending on how a business is
run, how modern their plant are, how much maintenance everything needs,

Page | 45
where the factory is located, among many others. However, for this study, the
value used for overheads is $70/tonne.

3.6 Conclusion
The methodology for this project consists of four main steps: analysis of loads,
design of the standard RCBC, determination of optimum culvert and feasibility
analysis Firstly, the SM1600 loads were analysed and the worst case for each
part of the culvert were found and taken as design load. This was achieved by
implementing a couple of Matlab scripts, namely load_comb.m and
finalscript.m. Then, a flexibility and shear analysis was carried out to
design the reinforcement according to AS1597.2-2013 and MRTS24 (Aug/11).
This non-optimised culvert was the basis for a comparison with the optimum
RCBC. The Matlab scripts utilised to implement that were flexanalysis.m,
devlength.m and checkshear.m. Following that the optimum topology
for the RCBC under the design loads was found utilising the SIMP method and
finite element analysis. The Matlab script utilised in this step was
top_rcbc4.m, which was an adaptation from Sigmund (2001). Lastly, a
feasibility analysis was carried out to outline the benefits and drawbacks of
utilising this optimisation procedure in the industry. The information was
obtained by means of an interview with an experienced manager in the industry.

Page | 46
 CHAPTER 4 - RCBC DESIGN

4.1 Introduction
The structural design procedure is always iterative. Firstly the leg and crown
thicknesses are assumed. Then the design loads are found and applied to the
finite element model of the structure generated in Strand7. The bending
moment and shear force diagrams are generated. A flexibility and shear analysis
then follows to find a suitable reinforcement for the structure. If by any chance
the section is found to be too thin and fails in shear or bending, the assumed
values at the beginning of the procedure are changed and the process starts
again. There are many ways a compliant design can be achieved and different
designers could find different acceptable solutions. The designs found in this
study were kept as similar as feasibly possible to each other to allow for easy
comparison.

4.2 Load Combination Results

Four horizontal and four vertical loads are applicable to this study, as discussed
in Section 3.2. However, these loads may be combined in a variety of ways to
model different scenarios. That is why different load factors apply to each load,
and the aim is to find out which load combination is the worst so that it can be
used as the design load.

To achieve that, a Matlab script named load_comb.m was developed. It puts

together all possible combinations of loads with their applicable load factors,
in both the symmetric and asymmetric loading cases, which are discussed in
detail in the following sections. The Matlab script can be found in Appendix
G.

Page | 47
4.2.1 Symmetric loading
Following the principles outlined in Section 3.2, the only possible symmetric
load combinations for vertical and horizontal loads respectively are shown in
Figure 4-1 and Figure 4-2.

1.0 ⋅WDC + 1.4 ⋅WFV + 0 ⋅WCV + 1.8 ⋅WLV

1.
1.0 ⋅WDC + 1.4 ⋅WFV + 1.5 ⋅WCV + 0 ⋅WLV
2.

1.0 ⋅WDC + 0.9 ⋅WFV + 0 ⋅WCV + 1.8 ⋅WLV

3.
1.0 ⋅WDC + 0.9 ⋅WFV + 1.5 ⋅WCV + 0 ⋅WLV
4.

1.0 ⋅WDC + 0.9 ⋅WFV + 0 ⋅WCV + 0 ⋅WLV

5.
1.0 ⋅WDC + 1.4 ⋅WFV + 0 ⋅WCV + 0 ⋅WLV
6.
Figure 4-1 - Vertical load combinations

0.7 ⋅ WFH + 0.7 ⋅WAH + 0 ⋅WCH + 1.8 ⋅WLH

7.
0.7 ⋅WFH + 1.4 ⋅ WAH + 0 ⋅WCH + 1.8 ⋅WLH
8.
1.4 ⋅WFH + 0.7 ⋅ WAH + 0 ⋅WCH + 1.8 ⋅WLH
9.
1.4 ⋅WFH + 1.4 ⋅WAH + 0 ⋅WCH + 1.8 ⋅WLH
10.

0.7 ⋅ WFH + 0.7 ⋅WAH + 1.5 ⋅WCH + 0 ⋅WLH

11.
0.7 ⋅WFH + 1.4 ⋅ WAH + 1.5 ⋅WCH + 0 ⋅WLH
12.
13. 1.4⋅WFH + 0.7 ⋅WAH +1.5⋅WCH + 0⋅WLH
14. 1.4⋅WFH +1.4⋅WAH +1.5⋅WCH + 0⋅WLH

15.
0.7⋅WFH + 0.7⋅WAH + 0⋅WCH + 0⋅WLH

16.
0.7⋅WFH +1.4⋅WAH + 0⋅WCH + 0⋅WLH
17.
1.4⋅WFH + 0.7⋅WAH + 0⋅WCH + 0⋅WLH
18.
1.4⋅WFH +1.4⋅WAH + 0⋅WCH + 0⋅WLH
Figure 4-2 - Horizontal load combinations

It can be seen that live construction loads and live roadway loads are never
considered together. That is because unless a specific construction plant vehicle
is utilised, the culverts are to be designed to support construction traffic loads

Page | 48
and their effects defined by the wheel loadings of SM1600 (Standards Australia
2013, p. 28). This means the construction plant cannot be heavier than the
SM1600 traffic loads the culverts are being designed for. However, the load
factor for the construction loads and their effects is 1.5 while the one for live
roadway traffic is 1.8. This is to account for the fact that the construction load
will happen less often than the roadway live load.

To find all possible symmetric load combinations, we assemble each of the

vertical loads (from 1 to 6) with each of the horizontal loads (from 7 to 18),
keeping in mind that if there is no vertical construction load there cannot be a
horizontal construction load. The same applies to roadway live load.

Combination 1: 1V and 7H
Combination 2: 1V and 8H
Combination 3: 1V and 9H
Combination 4: 1V and 10H

Combination 5: 2V and 11H

Combination 6: 2V and 12H
Combination 7: 2V and 13H
Combination 8: 2V and 14H

Combination 9: 3V and 7H
Combination 10: 3V and 8H
Combination 11: 3V and 9H
Combination 12: 3V and 10H

Combination 13: 4V and 11H

Combination 14: 4V and 12H
Combination 15: 4V and 13H
Combination 16: 4V and 14H

Combination 17: 5V and 15H

Combination 18: 5V and 16H
Combination 19: 5V and 17H
Combination 20: 5V and 18H

Combination 21: 6V and 15H

Combination 22: 6V and 16H
Combination 23: 6V and 17H
Combination 24: 6V and 18H

Page | 49
4.2.2 Asymmetric loading
The only direction in which the symmetric and asymmetric loadings differ is
the horizontal, and the differences are only relevant regarding fill and
compaction, since the construction and roadway load depend only on the
vertical loads, which remain the same. Therefore, on one side of the culvert
there will be

17.
1.4⋅WFH + 0.7⋅WAH + 0⋅WCH + 0⋅WLH
18.
1.4⋅WFH +1.4⋅WAH + 0⋅WCH + 0⋅WLH
Figure 4-3 – Asymmetric horizontal load combinations – one
side of culvert

The correspondent loads on the other side of the culvert will be:

15.
0.7⋅WFH + 0.7⋅WAH + 0⋅WCH + 0⋅WLH
16.
0.7⋅WFH +1.4⋅WAH + 0⋅WCH + 0⋅WLH
Figure 4-4 – Asymmetric horizontal load combinations – other side of culvert

The only possible asymmetric load combinations will therefore be:

Combination 1: 1V and 16H/17H

Combination 2: 1V and 16H/17H
Combination 3: 1V and 15H/18H
Combination 4: 1V and 16H/18H

Combination 5: 2V and 15H/17H

Combination 6: 2V and 16H/17H
Combination 7: 2V and 15H/18H
Combination 8: 2V and 16H/18H

Combination 9: 3V and 15H/17H

Combination 10: 3V and 16H/17H
Combination 11: 3V and 15H/18H
Combination 12: 3V and 16H/18H

Combination 13: 4V and 15H/17H

Combination 14: 4V and 16H/17H
Combination 15: 4V and 15H/18H
Combination 16: 4V and 16H/18H

Page | 50
Combination 17: 5V and 15H/17H
Combination 18: 5V and 16H/17H
Combination 19: 5V and 15H/18H
Combination 20: 5V and 16H/18H

Combination 21: 6V and 15H/17H

Combination 22: 6V and 16H/17H
Combination 23: 6V and 15H/18H
Combination 24: 6V and 16H/18H

However, it is important to note that the objective of this study is to analyse

culverts that comply with MRTS24 and can be installed under main roads. If
the horizontal loading on the culvert due to fill and compaction is asymmetric,
it means there is a different amount of fill on either side of the culvert or that
one side is unsuitably compacted while the other is suitably compacted. Because
this situation is very rare in Main Roads projects and it would probably not be
compliant, the asymmetric loading is not going to be considered in this study.

A programme called finalscript.m (see Appendix D) was created in

MATLAB to reveal the critical symmetrical load combination, after the script
load_comb.m (see Appendix G) generates all possible combinations. The
critical load combination was found by analysing the various possible
combinations of vehicles, load distributions through fill and fill heights in
increments of 0.1 m.

4.3 Design Loads

As explained in Chapter 3 - Methodology, the culverts in this study are assumed
to be subjected to SM1600 loads, which represent the W80, A160, M1600 and
S1600 design loads. (Standards Australia 2013, p. 58). These loads model road
traffic design loads for main and secondary roads, which are commonly
specified in the industry. Heavy load platform loads (HLP320 and HLP400)
and railway loads (300LA) are not included in this study.

4.4 1815 RCBC Design

To design the RCBC with 1.8 m span and 1.5 m leg (1815 RCBC), different
crown thicknesses (200 mm, 300 mm and 400) and leg thicknesses (200mm,

Page | 51
220mm, 300 mm and 350mm) were trialled before the design could be finalised.
The detailed calculations for a trial design with a leg thickness of 200 mm and
crown thickness of 250 mm can be seen in Appendix H. The design called for
shear reinforcement with N12 bars at 33 mm centres, meaning the spacing
between the edge of the bars would actually be 19 mm. That is a problem since
most 50MPa concrete mixes would have maximum aggregate size of 20 mm,
and having those bars close together would impact with the casting procedure.

The most suitable design for the 1815 RCBC was achieved with a 350 mm leg
and 400 mm crown, meaning the overall culvert width was 2.5 m and the overall
height was 1.9 m. The results from the load combination Matlab script
finalscript.m (see appendix D) were as shown in Figure 4-5 and Figure
4-10. The loads on the top of the culvert are uniformly distributed over the
entire top of culvert, as expected since the truncated prism model (see Section
3.2 for details) distributes the vehicle and construction loads uniformly over the
top of the culvert. The other loads that act on top of the culvert are also
uniformly distributed, namely the fill and the self-weight. There are 20 straight
lines in Figure 4-5, each for a different fill height. The top line is the load when
there is only 0.1 m of fill over the culvert, giving a load of 654 kPa for all values
of x (width of culvert). The bottom lines are very close together, which makes
it difficult to distinguish one from the other. However, the last line represents
the load for the case when there is 2.0 m of fill over the culvert, giving 114.4
kPa.

Page | 52
 Symmetric load combinations - top of culvert
700

Fill height: 0.1m

600 Load: 654 kPa

500
Load (kPa)

Fill height: 0.2m

400 Load: 455.3 kPa

300 Fill height: 0.3 m

Load: 339.4 KPa

200

100
0 500 1000 1500 2000 2500 3000
Width (mm) Fill height: 2.0 m
Load: 111.4 KPa

Figure 4-5 – Loads on top of 1815 RCBC

The worst load case was that for the least amount of fill, conservatively chosen
to be 0.1 m in this study (refer Section 3.2.3). The worst load combination was
the vertical load number 1 (see Figure 4-1) and horizontal load number 10 (see
Figure 4-2). This is expected since these combinations have the highest load
factors. All vertical load values for the various fill heights as plotted in Figure
4-5 can be seen in Table 4-1, since the Matlab script finalscript.m
calculates all possible load combinations in 0.1 m fill increments:

Page | 53
Table 4-1 - Vertical loads on 1815 RCBC

Fill Height (m) Load (kPa)

0.1 654.039
0.2 455.3062
0.3 339.4182
0.4 266.775
0.5 217.5766
0.6 184.623
0.7 160.7142
0.8 142.2502
0.9 131.031
1 121.6566
1.1 115.927
1.2 110.2422
1.3 111.8022
1.4 111.607
1.5 111.4566
1.6 111.351
1.7 113.0902
1.8 113.0742
1.9 114.7966
2 114.4166

The horizontal loads have a different form to the vertical loads (Figure 4-6).
They increase from the top of culvert up to 0.5 m below the top of culvert,
from which they continue uniformly. That is because the only horizontal load
which is uniformly distributed is the horizontal live load WLH as shown in
Figure 4-7. The compaction load WAH is shown in Figure 4-8. As dictated in
AS1597.2-2013, it increases up to 0.5 m below the top of the culvert, then
remains constant up to 1.5 m below the top of the culvert, then decreases
linearly up to 2.0 m below the top of the culvert (see Section 3.2.9 for details).
The horizontal fill load WFH is also non-uniform since it varies with the depth
below the culvert.

Page | 54
 Figure 4-6 - Vertical loads

Figure 4-7 - Horizontal live load

Figure 4-8 - Compaction load

Page | 55
 Figure 4-9 - Horizontal fill load

Because of these non-uniformities, the shape of the horizontal load graphs

(Figure 4-10) differs from the vertical load graphs (Figure 4-5).

Symmetric load combinations - sides of culvert

350
X: 500
Y: 344.3
300 X: 1
Y: 316.4
X: 501
Y: 245
250 X: 1
Y: 217
X: 501
200 Y: 187
Load (kPa)

X: 1
Y: 159
150

100 X: 501
Y: 70.4
X: 1
Y: 42.4
50

0
0 200 400 600 800 1000 1200 1400 1600 1800
Height from top of culvert (mm)

Figure 4-10 - Loads on the side of 1815 RCBC

The 1815 RCBC with crown thickness of 400 mm and the leg thickness of 350
mm is then modelled in Strand7 and the worst load case (for 0.1 m fill) is

Page | 56
applied. The bending moment diagram and shear force diagram are then found
as per Figure 4-11 and Figure 4-12.

Figure 4-11 – Bending moment diagram for 1815 RCBC

Page | 57
 Figure 4-12 - Shear force diagram for 1815 RCBC

The next step is to perform a flexibility and shear analysis to determine the
suitable reinforcement for the unit, using the Matlab scripts
flexanalysis.m and checkshear.m, as described in Section 3.3.5 - Non-
optimised Culverts and Section 3.3.6 - Reinforcement.

Flexure and shear analysis for the middle of the crown

For the middle of the crown, M*=369.1 kNm (Figure 4-11) and by running
flexanalysis.m with 14-N16 bars as tensile reinforcement, the results are:

>> flexanalysis(400,14,16,0,12) ;
dn = 19.6200
ku = 0.0550
Cc = 1.4009e+03
Cs = 0
Mu = 490.4901

Page | 58
phiMu = 392.3921
Number of N12 bars required for compression: 0
Number of N16 bars required for tension: 14

This means that:

• the section is under-reinforced since ku<0.36;

• φ M u > M * is suitable and so is the chosen reinforcement

The development length of the tension bars is checked using the Matlab
function devlength.m (see Matlab code in Appendix E) as per described in
Section 3.3.6. The result for 14-N16 in tension, with cover=35 mm and k1=1
is Lsy.tb = 464 mm. This means this bar has to be at least 928 mm long.

The shear for the middle of the crown is 0 so a shear check does not need to
be carried out and no shear reinforcement is required.

The same procedure is utilised for the end of the crown, the top of the leg and
the bottom of the leg.

Flexure and shear analysis for end of the crown

The design moment at the end of the crown is M*=396.4 kNm (see Figure
4-11). By running flexanalysis.m for 15-N16 bars in tension, the results
are:

>> flexanalysis(400,15,16,0,12) ;
dn = 21.0200
ku = 0.0589
Cc = 1.5008e+03
Cs = 0
Mu = 524.7540
phiMu = 419.8032
Number of N12 bars required for compression: 0
Number of N16 bars required for tension: 15

Page | 59
This means that:

• the section is under-reinforced since ku<0.36;

• φ M u > M * is suitable and so is the chosen reinforcement

Therefore, a shear check is required to check if shear reinforcement is

necessary. The design shear force at the end of the crown is V*=1550.4 kN
(see Figure 4-12). By running checkshear.m for the 15-N16 bars in tension,
the results are:

>>checkshear(1550.4E3,16,15,400);
Vumax = 8568000
Vuc = 6.5536e+05
Vumin = 1.2612e+06
Vusmin = 1.5595e+06
s_vusmin = 88.1342
s = 66.6667
Shear reinforcement is required. Provide 7-N12 bars at
67 mm spacings

This means that (1550400 = V * ) > (φVu.min = 882840) and 7-N12 ligatures
will be provided at 67 mm spacing for an extent of D=400 mm. The maximum
spacing, the one relating to Vus.min noted in the Matlab script as s_vusmin,
would be 88 mm therefore 67 mm spacing is suitable. If 6-N12s were chosen
instead, that spacing would be more than the minimum and it would not be
suitable.

The development length of the tension bars is checked using the Matlab
function devlength.m (see Matlab code in Appendix E) as per described in
Section 3.3.6. The result for 15-N16 in tension, with cover=35 mm and k1=1
is Lsy.tb = 464 mm. This means that each leg of this L bar this bar has to be at
least 464 mm long.

Page | 60
Flexure and shear analysis for top of the leg

The design moment at the top of the leg is the same as the one at the end of
the crown, as expected, and is M*=396.4 kNm (see Figure 4-11). However, this
section is thinner since the leg is 350 mm and the crown is 400 mm. By running
flexanalysis.m for 18-N16 bars in tension, the results are:

>>flexanalysis(350,18,16,0,12) ;
dn = 25.2300
ku = 0.0822
Cc = 1.8014e+03
Cs = 0
Mu = 537.1291
phiMu = 429.7033
Number of N12 bars required for compression: 0
Number of N16 bars required for tension: 18

This means that:

• the section is under-reinforced since ku<0.36;

• φ M u > M * is suitable and so is the chosen reinforcement

A shear check is required to check if shear reinforcement is necessary. The

design shear force at the top of the leg is V*=908.8 kN (see Figure 4-12). By
running checkshear.m for the 18-N16 bars in tension, the results are:

>>checkshear(908.8E3,16,18,350)
Vumax = 7368000
Vuc = 6.5511e+05
Vumin = 1.1761e+06
Vusmin = 6.4317e+05
s_vusmin = 131.2630
s = 87.5000
Shear reinforcement is required. Provide 5-N12 bars at
88 mm spacings

Page | 61
This means that (908800 = V * ) > (φVu.min = 823270) and 5-N12 ligatures will
be provided at 88 mm spacing for an extent of D=350 mm.

The development length of the tension bars is checked using the Matlab
function devlength.m (see Matlab code in Appendix E) as per described in
Section 3.3.6. The result for 18-N16 in tension, with cover=35 mm and k1=1.3
is Lsy.tb = 603 mm. This means that each leg of this L bar this bar has to be at
least 603 mm long.

Flexure and shear analysis for bottom of the leg

The design moment at the bottom third of the leg is M*=103.37 kNm (see
Figure 4-11). By running flexanalysis.m for 8-N12 bars in tension, the
results are:

>> flexanalysis(350,8,12,0,12) ;
dn = 6.1600
ku = 0.0199
Cc = 439.8240
Cs = 0
Mu = 134.9574
phiMu = 107.9659
Number of N12 bars required for compression: 0
Number of N12 bars required for tension: 8

This means that:

• the section is under-reinforced since ku<0.36;

• φ M u > M * is suitable and so is the chosen reinforcement

A shear check is required to check if shear reinforcement is necessary. The

design shear force at the bottom of the leg is V*=413.32 kN (see Figure 4-12).
By running checkshear.m for the 8-N12 bars in tension, the results are:

Page | 62
>>checkshear(413.32E3,12,8,350)
Vumax = 7416000
Vuc = 4.1076e+05
Vumin = 9.3515e+05
ans =Minimum shear reinforcement is required.

This means that 0.5φVuc < V * ≤ φVu.min and Asv.min will be provided for an

extent of D=350 mm. In this case, V * > φVuc so the minimum shear
reinforcement requirements cannot be waived. For it to be waived, 24-N12s
would have to be provided and since there is a great difference between the
required 8-N12s and 24-N12s, shear ligatures will be provided and 8-N12 bars
will be installed.

Using the maximum spacing of s=0.5D=175 mm from AS3600-2009 clause

8.2.12.2 into Equation 3-4, the result is Asv.min = 356.3818 mm, which means
there will be 4-N12 ligatures required at 116 mm spacings.

The development length of the tension bars is checked using the Matlab
function devlength.m (see Matlab code in Appendix E) as per described in
Section 3.3.6. The result for 8-N12 bars in tension, with cover=35 mm and
k1=1 is Lsy.tb = 348 mm. This means this bar has to be at least 696 mm long.

The final 1815 RCBC reinforcement is shown diagrammatically in Figure 4-13.

Page | 63
 Figure 4-13 - 1815 RCBC reinforcement

In order to simplify the assembly of the reinforcement cages, especially with

relation to the ligatures, the L bars at the end of the crown / top of the leg
(N1601) were made the same length with the same spacing. The bar at the
bottom of the crown (N1602) was also changed from 14-N16 to 18-N16, to
follow the same spacing as the L bars and therefore facilitate the installation of
the ligatures. The quantity of the bar N1204 was changed from 8 to 9 to
facilitate the placement of ligatures, which should not be further apart than 600
mm through the width of the beam. The bars N1201 and N1204 were included
to enable the connection of the distribution bars for cracking and the shear
ligatures.

The reinforcement schedule for the 1815 RCBC is shown in Table 4-2.

Page | 64
 Table 4-2 - 1815 RCBC reinforcement schedule

Total mass
Bar Mark Grade & Size Qty Shape
(kg)
N1201 N12 10 15.77 straight
N1202 N12 18 17.19 straight
N1203 N12 37 76.56 straight
N1204 N12 18 21.90 straight
N1205 N12 70 113.82 Ligs
N1206 N12 72 94.18 Ligs
N1601 N16 72 136.18 L
N1602 N16 18 33.00 straight

Total reinforcement mass: 508.60 kg

4.5 1818 RCBC Design

To design the RCBC with 1.8 m span and 1.8 m leg (1818 RCBC), the same
cross section was utilised than that of the 1815 RCBC (see Section 4.4), that is,
a 350 mm leg and 400 mm crown. The overall culvert width was therefore 2.5
m and the overall height was 2.2 m.

The results from the load combination Matlab script finalscript.m (see
appendix D for code) were as shown in Figure 4-14 and Figure 4-15. The loads
on the top of the culvert are uniformly distributed over the entire top of culvert,
as expected since the truncated prism model (see Section 3.2 for details)
distributes the vehicle and construction loads uniformly over the top of the
culvert. The other loads that act on top of the culvert are also uniformly
distributed, namely the fill and the self-weight. There are 20 straight lines in
Figure 4-14, each for a different fill height. The top line is the load when there
is only 0.1 m of fill over the culvert, giving a load of 656.2 kPa for all values of
x (width of culvert). The bottom lines are very close together, which makes it
difficult to distinguish one from the other. However, the last line represents the
load for the case when there is 2.0 m of fill over the culvert, giving 112.4 kPa.

Page | 65
 Symmetric load combinations - top of culvert
700

600 Fill height: 0.1 m

Load: 656.2 kPa

500
Load (kPa)

Fill height: 0.2 m

400 Load: 457.5 kPa

300 Fill height: 0.3 m

Load: 341.6 kPa

200

100
0 500 1000 1500 2000 2500 3000
Width (mm) Fill height: 2.0 m
Load: 112.4 kPa

Figure 4-14 – Loads on top of 1818 RCBC

The worst load case was that for the least amount of fill, conservatively chosen
to be 0.1 m in this study (refer Section 3.2.3). The worst load combination was
the vertical load number 1 (see Figure 4-1) and horizontal load number 10 (see
Figure 4-2). This is expected since these combinations have the highest load
factors.

The horizontal loads have a different form to the vertical loads (see Section
4.4). Because of non-uniformities, the shape of the horizontal load graphs
differs from the vertical load graphs.

Page | 66
 Symmetric load combinations - sides of culvert
350
X: 501 X: 2001 X: 2201
Y: 344.4 Y: 344.4 Y: 347.2

300 X: 1
Y: 316.4

X: 501 X: 2001 X: 2201

Y: 245 Y: 245 Y: 247.8
250
X: 1
Y: 217

X: 501 X: 2001 X: 2201

200 Y: 187 Y: 187 Y: 189.8
Load (kPa)

X: 1
Y: 159

150

100 X: 501 X: 2001 X: 2201

Y: 70.4 Y: 70.4 Y: 73.2
X: 1
Y: 42.4
50

0
0 500 1000 1500 2000 2500
Height from top of culvert (mm)

Figure 4-15 - Loads on the side of 1818 RCBC

The 1818 RCBC with crown thickness of 400 mm and the leg thickness of 350
mm is then modelled in Strand7 and the worst load case (for 0.1 m fill) is
applied. The bending moment diagram and shear force diagram are then found
as per Figure 4-16 and Figure 4-17.

Page | 67
Figure 4-16 – Bending moment diagram for 1818 RCBC

Figure 4-17 - Shear force diagram for 1818 RCBC

Page | 68
The next step is to perform a flexibility and shear analysis to determine the
suitable reinforcement for the unit, using the Matlab scripts
flexanalysis.m and checkshear.m, as described in Section 3.3.5 - Non-
optimised Culverts.

Flexure and shear analysis for the middle of the crown

For the middle of the crown, M*=400.91 kNm (Figure 4-16) and by running
flexanalysis.m with 15-N16 bars as tensile reinforcement, the results are:

>> flexanalysis(400,15,16,0,12) ;
dn = 21.0200
ku = 0.0589
Cc = 1.5008e+03
Cs = 0
Mu = 524.7540
phiMu = 419.8032
Number of N12 bars required for compression: 0
Number of N16 bars required for tension: 15

This means that:

• the section is under-reinforced since ku<0.36;

• φ M u > M * is suitable and so is the chosen reinforcement

The development length of the tension bars is checked using the Matlab
function devlength.m (see Matlab code in Appendix E) as per described in
Section 3.3.6. The result for 15-N16 in tension, with cover=35 mm and k1=1
is Lsy.tb = 464 mm. This means this bar has to be at least 928 mm long.

The shear for the middle of the crown is 0 so a shear check does not need to
be carried out and no shear reinforcement is required.

Page | 69
The same procedure is utilised for the end of the crown, the top of the leg and
the bottom of the leg.

Flexure and shear analysis for end of the crown

The design moment at the end of the crown is M*=509.1 kNm (see Figure
4-16). By running flexanalysis.m for 19-N16 bars in tension, the results
are:

>> flexanalysis(400,19,16,0,12) ;
dn = 26.6300
ku = 0.0746
Cc = 1.9014e+03
Cs = 0
Mu = 661.0715
phiMu = 528.8572
Number of N12 bars required for compression: 0
Number of N16 bars required for tension: 19

This means that:

• the section is under-reinforced since ku<0.36;

• φ M u > M * is suitable and so is the chosen reinforcement

The design shear force at the end of the crown is V*=1693 kN (see Figure
4-17). By running checkshear.m for the 19-N16 bars in tension, the results
are:

>> checkshear(1693E3,16,19,400);
Vumax = 8568000
Vuc = 7.0909e+05
Vumin = 1.3149e+06
Vusmin = 1.7095e+06
s_vusmin = 80.4016
s = 66.6667

Page | 70
Shear reinforcement is required. Provide 7-N12 bars at
67 mm spacings

This means that (1693000 = V * ) > (φVu.min = 920430) and 7-N12 ligatures
will be provided at 67 mm spacing for an extent of D=400 mm.

The development length of the tension bars is checked using the Matlab
function devlength.m (see Matlab code in Appendix E) as per described in
Section 3.3.6. The result for 19-N16 in tension, with cover=35 mm and k1=1
is Lsy.tb = 464 mm. This means that each leg of this L bar this bar has to be at
least 464 mm long.

Flexure and shear analysis for top of the leg

The design moment at the top of the leg is the same as the one at the end of
the crown, as expected, and is M*=509.1 kNm (see Figure 4-16). However, this
section is thinner since the leg is 350 mm and the crown is 400 mm. By running
flexanalysis.m for 23-N16 bars in tension, the results are:

>>flexanalysis(350,23,16,0,12) ;
dn = 32.2400
ku = 0.1050
Cc = 2.3019e+03
Cs = 0
Mu = 680.7193
phiMu = 544.5754
Number of N12 bars required for compression: 0
Number of N16 bars required for tension: 23

This means that:

• the section is under-reinforced since ku<0.36;

• φ M u > M * is suitable and so is the chosen reinforcement

Page | 71
A shear check is required to check if shear reinforcement is necessary. The
design shear force at the top of the leg is V*=1693 kN (see Figure 4-17). By
running checkshear.m for the 23-N16 bars in tension, the results are:

>> checkshear(1693E3,16,23,350)
Vumax = 7368000
Vuc = 7.1089e+05
Vumin = 1.2319e+06
Vusmin = 1.7077e+06
s_vusmin = 69.2136
s = 58.3333
Shear reinforcement is required. Provide 7-N12 bars at
58 mm spacings

This means that (1693000 = V * ) > (φVu.min = 862330) and 7-N12 ligatures
will be provided at 58 mm spacing for an extent of D=350 mm.

The development length of the tension bars is checked using the Matlab
function devlength.m (see Matlab code in Appendix E) as per described in
Section 3.3.6. The result for 23-N16 in tension, with cover=35 mm and k1=1.3
is Lsy.tb = 603 mm. This means that each leg of this L bar this bar has to be at
least 603 mm long.

Flexure and shear analysis for bottom of the leg

The design moment at the bottom third of the leg is M*=197.15 kNm
(seeFigure 4-16). By running flexanalysis.m for 15-N12 bars in tension,
the results are:

>> flexanalysis(350,15,12,0,12) ;
dn = 11.5600
ku = 0.0374
Cc = 825.3840
Cs = 0
Mu = 251.7042

Page | 72
phiMu = 201.3633

Number of N12 bars required for compression: 0

Number of N12 bars required for tension: 15

This means that:

• the section is under-reinforced since ku<0.36;

• φ M u > M * is suitable and so is the chosen reinforcement

A shear check is required to check if shear reinforcement is necessary. The

design shear force at the bottom of the leg is V*=571.5 kN (see Figure 4-17).
By running checkshear.m for the 15-N12 bars in tension, the results are:

>> checkshear(571.5E3,12,15,350)
Vumax = 7416000
Vuc = 5.0650e+05
Vumin = 1.0309e+06
Asvmin = 356.3818
ans =Minimum shear reinforcement is required.

This means that 0.5φVuc < V * ≤ φVu.min and Asv.min will be provided for an

extent of D=350 mm. In this case, V * > φVuc so the minimum shear
reinforcement requirements cannot be waived. Using the maximum spacing of
s=0.5D=175 mm from AS3600-2009 clause 8.2.12.2 into Equation 3-4, the
result is Asv.min = 356.3818 mm, which means there will be 4-N12 ligatures
required at 116 mm spacings.

The development length of the tension bars is checked using the Matlab
function devlength.m (see Matlab code in Appendix E) as per described in
Section 3.3.6. The result for 15-N12 bars in tension, with cover=35 mm and
k1=1 is Lsy.tb = 348 mm. This means this bar has to be at least 696 mm long.

Page | 73
The final 1818 RCBC reinforcement is shown diagrammatically in Figure 4-18.

Figure 4-18 - 1818 RCBC reinforcement

In order to simplify the assembly of the reinforcement cages, especially with

relation to the ligatures, the L bars at the end of the crown / top of the leg
(N1601) were made the same length with the same spacing. The bars N1201
and N1204 were included to enable the connection of the distribution bars for
cracking and the shear ligatures.

The reinforcement schedule for the 1818 RCBC is shown in Table 4-2.

Page | 74
 Table 4-3 - 1818 RCBC reinforcement schedule

Total mass
Bar Mark Grade & Size Qty Shape
(kg)
N1201 N12 10 15.77 straight
N1202 N12 30 36.63 straight
N1203 N12 41 84.84 straight
N1204 N12 30 44.49 straight
N1205 N12 70 113.82 Ligs
N1206 N12 110 169.28 Ligs
N1601 N16 92 174.00 L
N1602 N16 15 27.50 straight

Total reinforcement mass: 663.33 kg

4.6 2412 RCBC Design

To design the RCBC with 2.4 m span and 1.2 m leg (2412 RCBC), the cross
section utilised comprised of a 350 mm leg and 400 mm crown. The overall
culvert width was therefore 3.1 m and the overall height was 1.6 m.

The results from the load combination Matlab script finalscript.m (see
appendix D for code) were as shown in Figure 4-19 and Figure 4-20. There are
20 straight lines in Figure 4-19, each for a different fill height. The top line is
the load when there is only 0.1 m of fill over the culvert, giving a load of 650.2
kPa for all values of x (width of culvert). The bottom lines are very close
together, which makes it difficult to distinguish one from the other. However,
the last line represents the load for the case when there is 2.0 m of fill over the
culvert, giving 105.8 kPa.

Page | 75
 Symmetric load combinations - top of culvert
700

600 Fill height: 0.1 m

Load: 650.2 kPa

500
Load (kPa)

400 Fill height: 0.2 m

Load: 451.4 kPa

300 Fill height: 0.3 m

Load: 335.5 kPa

200

100
0 500 1000 1500 2000 2500 3000 3500
Width (mm) Fill height: 2.0 m
Load: 105.8 kPa

Figure 4-19 – Loads on top of RCBC 2412

The worst load case was that for the least amount of fill, conservatively chosen
to be 0.1 m in this study (refer Section 3.2.3). The worst load combination was
the vertical load number 1 (see Figure 4-1) and horizontal load number 10 (see
Figure 4-2). This is expected since these combinations have the highest load
factors. The horizontal loads combinations were the same as for the 1818
RCBC, since the loads on the culvert are the same (SM1600) and the culvert
height is the same. They are shown in Figure 4-20.

Page | 76
 Symmetric load combinations - sides of culvert
350
X: 501 X: 1600
Y: 344.4 Y: 344.4

300 X: 1
Y: 316.4

X: 501 X: 1600
Y: 245 Y: 245
250
X: 1
Y: 217

X: 501 X: 1600
200 Y: 187 Y: 187
Load (kPa)

X: 1
Y: 159

150

100 X: 501 X: 1600

Y: 70.4 Y: 70.4
X: 1
Y: 42.4
50

0
0 200 400 600 800 1000 1200 1400 1600 1800
Height from top of culvert (mm)

Figure 4-20 - Loads on the side of the 2412 RCBC

The 2412 RCBC with crown thickness of 400 mm and the leg thickness of 350
mm is then modelled in Strand7 and the worst load case (for 0.1 m fill) is
applied. The bending moment diagram and shear force diagram are then found
as per Figure 4-21 and Figure 4-22.

Page | 77
 Figure 4-21 - Bending moment diagram for 2412 RCBC

Figure 4-22 - Shear force diagram for 2412 RCBC

The next step is to perform a flexibility and shear analysis to determine the
suitable reinforcement for the unit, using the Matlab scripts

Page | 78
flexanalysis.m and checkshear.m, as described in detail in Section
3.3.5 - Non-optimised Culverts and Section 3.3.6 - Reinforcement.

Flexure and shear analysis for the middle of the crown

For the middle of the crown, M*=761.7 kNm (Figure 4-21) and by running
flexanalysis.m with 19-N20 bars as tensile reinforcement, the results are:

>> flexanalysis(400,19,20,0,12);
dn = 41.2800
ku = 0.1163
Cc = 2.9474e+03
Cs = 0
Mu = 1.0037e+03
phiMu = 802.9922
Number of N12 bars required for compression: 0
Number of N20 bars required for tension: 19

This means that:

• the section is under-reinforced since ku<0.36;

• φ M u > M * is suitable and so is the chosen reinforcement

The development length of the tension bars is checked using the Matlab
function devlength.m (see Matlab code in Appendix E) as per described in
Section 3.3.6. The result for 19-N20 in tension, with cover=35 mm and k1=1
is Lsy.tb = 580 mm. This means this bar has to be at least 1060 mm long.

The shear for the middle of the crown is 0 so a shear check does not need to
be carried out and no shear reinforcement is required.

The same procedure is utilised for the end of the crown, the top of the leg and
the bottom of the leg.

Page | 79
Flexure and shear analysis for end of the crown

The design moment at the end of the crown is M*=713.5 kNm (see Figure
4-21). By running flexanalysis.m for 18-N20 bars in tension, the results
are:

>> flexanalysis(400,18,20,0,12);
dn = 39.1100
ku = 0.1102
Cc = 2.7925e+03
Cs = 0
Mu = 953.0967
phiMu = 762.4773
Number of N12 bars required for compression: 0
Number of N20 bars required for tension: 18

This means that:

• the section is under-reinforced since ku<0.36;

• φ M u > M * is suitable and so is the chosen reinforcement

The design shear force at the end of the crown is V*=2145.7 kN (see Figure
4-22). By running checkshear.m for the 18-N20 bars in tension, the results
are:

>> checkshear(2145.7E3,20,18,400);
Vumax = 8520000
Vuc = 8.0425e+05
Vumin = 1.4067e+06
Vusmin = 2.2610e+06
s_vusmin = 69.0833
s = 57.1429
Shear reinforcement is required. Provide 8-N12 bars at
57 mm spacings

Page | 80
This means that (2145700 = V * ) > (φVu .min = 984690) and 8-N12 ligatures
will be provided at 57 mm spacing for an extent of D=400 mm.

The development length of the tension bars is checked using the Matlab
function devlength.m (see Matlab code in Appendix E) as per described in
Section 3.3.6. The result for 18-N20 in tension, with cover=35 mm and k1=1
is Lsy.tb = 580 mm. This means that each leg of this L bar this bar has to be at

least 580 mm long.

Flexure and shear analysis for top of the leg

The design moment at the top of the leg is the same as the one at the end of
the crown, as expected, and is M*=713.5 kNm (see Figure 4-21). However, this
section is thinner since the leg is 350 mm and the crown is 400 mm. By running
flexanalysis.m for 22-N20 bars in tension, the results are:

>>flexanalysis(350,22,20,0,12) ;
dn = 47.8000
ku = 0.1567
Cc = 3.4129e+03
Cs = 0
Mu = 983.8424
phiMu = 787.0740
Number of N12 bars required for compression: 0
Number of N20 bars required for tension: 22

This means that:

• the section is under-reinforced since ku<0.36;

• φ M u > M * is suitable and so is the chosen reinforcement

A shear check is required to check if shear reinforcement is necessary. The

design shear force at the top of the leg is V*=1075 kN (see Figure 4-22). By
running checkshear.m for the 22-N20 bars in tension, the results are:

Page | 81
>> checkshear(1183E3,20,22,350)
Vumax = 7320000
Vuc = 8.0833e+05
Vumin = 1.3259e+06
Vusmin = 7.2739e+05
s_vusmin = 115.3102
s = 87.5000
Shear reinforcement is required. Provide 5-N12 bars at
88 mm spacings

This means that (1075000 = V * ) > (φVu.min = 928130) and 5-N12 ligatures
will be provided at 88 mm spacing for an extent of D=350 mm.

The development length of the tension bars is checked using the Matlab
function devlength.m (see Matlab code in Appendix E) as per described in
Section 3.3.6. The result for 22-N20 in tension, with cover=35 mm and k1=1.3
is Lsy.tb = 754 mm. This means that each leg of this L bar this bar has to be at
least 754 mm long.

Flexure and shear analysis for bottom of the leg

The design moment at 0.5 m from the bottom of the leg is M*=69.9 kNm (see
Figure 4-23).

Page | 82
 Figure 4-23 - Bending moment and shear force for the 2412 RCBC leg

By running flexanalysis.m for 6-N12 bars in tension, the results are:

>> flexanalysis(350,6,12,0,12) ;
dn = 4.6200
ku = 0.0150
Cc = 329.8680
Cs = 0
Mu = 101.3958
phiMu = 81.1167
Number of N12 bars required for compression: 0
Number of N12 bars required for tension: 6

This means that:

• the section is under-reinforced since ku<0.36;

• φ M u > M * is suitable and so is the chosen reinforcement

A shear check is required to check if shear reinforcement is necessary. At 0.5

m from the bottom of the leg is V*=346 kN (see Figure 4-23). By running
checkshear.m for the 6-N12 bars in tension, the results are:

Page | 83
>> checkshear(345E3,12,6,350)
Vumax = 7416000
Vuc = 3.7320e+05
Vumin = 8.9759e+05
Asvmin = 356.3818
ans = Minimum shear reinforcement is required.

This means that 0.5φVuc < V * ≤ φVu.min and Asv.min will be provided for an

extent of D=350 mm. In this case, V * > φVuc so the minimum shear
reinforcement requirements cannot be waived. However, if the number of N12
bars is increased to 14-N12 in tension, the results from flexanalysis.m
are:

>>flexanalysis(350,14,12,0,12)
dn = 10.7900
ku = 0.0349
Cc = 770.4060
Cs = 0
Mu = 235.1460
phiMu = 188.1168
Number of N12 bars required for compression: 0
Number of N12 bars required for tension: 14

That means this reinforcement would also be suitable, but in this case, the shear
reinforcement requirements could be waived. This is shown by running
checkshear.m:

checkshear(345.7E3,12,14,350)
Vumax = 7416000
Vuc = 4.9499e+05
Vumin = 1.0194e+06
ans = Minimum shear reinforcement is required but may
be waived.

Page | 84
In this case, V * ≤ φVuc so the minimum shear reinforcement requirements can
be waived. It is more feasible to provide the extra 8-N12 straight bars than it is
to provide the required amount of shear reinforcement. The 32 ligatures
required in total weigh approximately 50 kg while the extra 16-N12 bars
required in total weigh approximately 22 kg. Also, it simplifies the assembly of
the cage since it is simpler to install straight bars than ligatures.

The development length of the tension bars is checked using the Matlab
function devlength.m (see Matlab code in Appendix E) as per described in
Section 3.3.6. The result for 14-N12 bars in tension, with cover=35 mm and
k1=1 is Lsy.tb = 348 mm. This means this bar has to be at least 696 mm long.

The final 2412 RCBC reinforcement is shown diagrammatically in Figure 4-24.

Figure 4-24 - 2412 RCBC reinforcement

In order to simplify the assembly of the reinforcement cages, especially with

relation to the ligatures, the L bars at the end of the crown / top of the leg
(N2001) were made the same length with the same spacing. The bars N1201

Page | 85
and N1204 were included to enable the connection of the distribution bars for
cracking. The reinforcement schedule for the 2412 RCBC is shown in Table
4-4.

Table 4-4 - 2412 RCBC reinforcement schedule

Total mass
Bar Mark Grade & Size Qty Shape
(kg)
N1201 N12 10 15.37 straight
N1202 N12 28 17.41 straight
N1203 N12 37 76.56 straight
N1204 N12 28 22.38 straight
N1205 N12 80 130.08 Ligs
N1206 N12 50 76.95 Ligs
N2001 N20 88 327.13 L
N2002 N20 19 59.61 straight

Total reinforcement mass: 725.49 kg

4.7 2415 RCBC Design

To design the RCBC with 2.4 m span and 1.5 m leg (2415 RCBC), the cross
section utilised comprised of a 350 mm leg and 400 mm crown. The overall
culvert width was therefore 3.1 m and the overall height was 1.9 m.

The results from the load combination Matlab script finalscript.m (see
appendix D for code) were as shown in Figure 4-30 and Figure 4-31. The loads
on the top of the culvert are uniformly distributed over the entire top of culvert,
as expected since the truncated prism model (see Section 3.2 for details)
distributes the vehicle and construction loads uniformly over the top of the
culvert. The other loads that act on top of the culvert are also uniformly
distributed, namely the fill and the self-weight. There are 20 straight lines in
Figure 4-30, each for a different fill height. The top line is the load when there
is only 0.1 m of fill over the culvert, giving a load of 651.9 kPa for all values of
x (width of culvert). The bottom lines are very close together, which makes it

Page | 86
difficult to distinguish one from the other. However, the last line represents the
load for the case when there is 2.0 m of fill over the culvert, giving 107.5 kPa.

Symmetric load combinations - top of culvert

700

Fill height: 0.1 m

600
Load: 651.9 kPa

500
Load (kPa)

Fill height: 0.2 m

400 Load: 453.2 kPa

300 Fill height: 0.3 m

Load: 337.3 kPa

200

100
0 500 1000 1500 2000 2500 3000 3500
Width (mm) Fill height: 2.0 m
Load: 107.5 kPa

Figure 4-25 – Loads on top of 2415 RCBC

The worst load case was that for the least amount of fill, conservatively chosen
to be 0.1 m in this study (refer Section 3.2.3). The worst load combination was
the vertical load number 1 (see Figure 4-1) and horizontal load number 10 (see
Figure 4-2). This is expected since these combinations have the highest load
factors. The horizontal loads combinations were the same as for the 1815
RCBC, since the loads on the culvert are the same (SM1600) and the culvert
height is the same. They are shown in Figure 4-31.

Page | 87
 Symmetric load combinations - sides of culvert
350
X: 500
Y: 344.3
300 X: 1
Y: 316.4
X: 501
Y: 245
250 X: 1
Y: 217
X: 501
200 Y: 187
Load (kPa)

X: 1
Y: 159
150

100 X: 501
Y: 70.4
X: 1
Y: 42.4
50

0
0 200 400 600 800 1000 1200 1400 1600 1800
Height from top of culvert (mm)

Figure 4-26 - Loads on the side of the 2415 RCBC

The 2415 RCBC with crown thickness of 400 mm and the leg thickness of 350
mm is then modelled in Strand7 and the worst load case (for 0.1 m fill) is
applied. The bending moment diagram and shear force diagram are then found
as per Figure 4-32 and Figure 4-33.

Page | 88
Figure 4-27 - Bending moment diagram for 2415 RCBC

Figure 4-28 - Shear force diagram for 2415 RCBC

Page | 89
The next step is to perform a flexibility and shear analysis to determine the
suitable reinforcement for the unit, using the Matlab scripts
flexanalysis.m and checkshear.m, as described in detail in Section
3.3.5 - Non-optimised Culverts and Section 3.3.6 - Reinforcement.

Flexure and shear analysis for the middle of the crown

For the middle of the crown, M*=761.0 kNm (Figure 4-32) and by running
flexanalysis.m with 19-N20 bars as tensile reinforcement, the results are:

>> flexanalysis(400,19,20,0,12);
dn = 41.2800
ku = 0.1163
Cc = 2.9474e+03
Cs = 0
Mu = 1.0037e+03
phiMu = 802.9922
Number of N12 bars required for compression: 0
Number of N20 bars required for tension: 19

This means that:

• the section is under-reinforced since ku<0.36;

• φ M u > M * is suitable and so is the chosen reinforcement

The development length of the tension bars is checked using the Matlab
function devlength.m (see Matlab code in Appendix E) as per described in
Section 3.3.6. The result for 19-N20 in tension, with cover=35 mm and k1=1
is Lsy.tb = 580 mm. This means this bar has to be at least 1060 mm long.

The shear for the middle of the crown is 0 so a shear check does not need to
be carried out and no shear reinforcement is required.

Page | 90
The same procedure is utilised for the end of the crown, the top of the leg and
the bottom of the leg.

Flexure and shear analysis for end of the crown

The design moment at the end of the crown is M*=718.0 kNm (see Figure
4-32). By running flexanalysis.m for 18-N20 bars in tension, the results
are:

>> flexanalysis(400,18,20,0,12)
dn = 39.1100
ku = 0.1102
Cc = 2.7925e+03
Cs = 0
Mu = 953.0967
phiMu = 762.4773
Number of N12 bars required for compression: 0
Number of N20 bars required for tension: 18

This means that:

• the section is under-reinforced since ku<0.36;

• φ M u > M * is suitable and so is the chosen reinforcement

The design shear force at the end of the crown is V*=2151.2 kN (see Figure
4-33). By running checkshear.m for the 18-N20 bars in tension, the results
are:

>> checkshear(2151.2E3,20,18,400);
Vumax = 8520000
Vuc = 8.0425e+05
Vumin = 1.4067e+06
Vusmin = 2.2689e+06
s_vusmin = 68.8441
s = 57.1429

Page | 91
Shear reinforcement is required. Provide 8-N12 bars at
57 mm spacings

This means that (2151200 = V * ) > (φVu .min = 984690) and 8-N12 ligatures
will be provided at 57 mm spacing for an extent of D=400 mm.

The development length of the tension bars is checked using the Matlab
function devlength.m (see Matlab code in Appendix E) as per described in
Section 3.3.6. The result for 18-N20 in tension, with cover=35 mm and k1=1
is Lsy.tb = 580 mm. This means that each leg of this L bar this bar has to be at
least 580 mm long.

Flexure and shear analysis for top of the leg

The design moment at the top of the leg is the same as the one at the end of
the crown, as expected, and is M*=718.0 kNm (see Figure 4-32). However, this
section is thinner since the leg is 350 mm and the crown is 400 mm. By running
flexanalysis.m for 22-N20 bars in tension, the results are:

>>flexanalysis(350,22,20,0,12) ;
dn = 47.8000
ku = 0.1567
Cc = 3.4129e+03
Cs = 0
Mu = 983.8424
phiMu = 787.0740
Number of N12 bars required for compression: 0
Number of N20 bars required for tension: 22

This means that:

• the section is under-reinforced since ku<0.36;

• φ M u > M * is suitable and so is the chosen reinforcement

Page | 92
A shear check is required to check if shear reinforcement is necessary. The
design shear force at the top of the leg is V*=1110 kN (seeFigure 4-33). By
running checkshear.m for the 22-N20 bars in tension, the results are:

>> checkshear(1183E3,20,23,350)
Vumax = 7320000
Vuc = 8.0833e+05
Vumin = 1.3259e+06
Vusmin = 7.7739e+05
s_vusmin = 107.8937
s = 87.5000
Shear reinforcement is required. Provide 5-N12 bars at
88 mm spacings

This means that (1110000 = V * ) > (φVu.min = 928130) and 5-N12 ligatures
will be provided at 88 mm spacing for an extent of D=350 mm.

The development length of the tension bars is checked using the Matlab
function devlength.m (see Matlab code in Appendix E) as per described in
Section 3.3.6. The result for 22-N20 in tension, with cover=35 mm and k1=1.3
is Lsy.tb = 754 mm. This means that each leg of this L bar this bar has to be at
least 754 mm long.

Flexure and shear analysis for bottom of the leg

The maximum design moment at the bottom third of the leg is M*=46.9 kNm
(see Figure 4-32). By running flexanalysis.m for 8-N12 bars in tension,
the results are:

>> flexanalysis(350,8,12,0,12) ;
dn = 6.1600
ku = 0.0199
Cc = 439.8240
Cs = 0
Mu = 134.9574

Page | 93
phiMu = 107.9659
Number of N12 bars required for compression: 0
Number of N12 bars required for tension: 8

This means that:

• the section is under-reinforced since ku<0.36;

• φ M u > M * is suitable and so is the chosen reinforcement

The reason why this seems overdesigned is that if less bars than 8-N12s are
utilised, the minimum shear reinforcement requirement cannot be waived. 8-
N12 bars are then chosen to eliminate the need to provide shear reinforcement.
The design shear force at the bottom of the leg is V*=278.5 kN (see Figure
4-33). By running checkshear.m for the 8-N12 bars in tension, the results
are:

>> checkshear(278.5E3,12,8,350)
Vumax = 7416000
Vuc = 4.1076e+05
Vumin = 9.3515e+05
ans =Minimum shear reinforcement is required but may be
waived.

This means that 0.5φVuc < V * ≤ φVu.min and Asv.min should be provided for an

extent of D=350 mm. In this case, V * ≤ φVuc so the minimum shear

reinforcement requirements can be waived.

The development length of the tension bars is checked using the Matlab
function devlength.m (see Matlab code in Appendix E) as per described in
Section 3.3.6. The result for 8-N12 bars in tension, with cover=35 mm and
k1=1 is Lsy.tb = 348 mm. This means this bar has to be at least 696 mm long.

The final 2415 RCBC reinforcement is shown diagrammatically in Figure 4-34.

Page | 94
 Figure 4-29 - 2415 RCBC reinforcement

In order to simplify the assembly of the reinforcement cages, especially with

relation to the ligatures, the L bars at the end of the crown / top of the leg
(N2001) were made the same length with the same spacing. The bars N1201
and N1204 were included to enable the connection of the distribution bars for
cracking and the shear ligatures.

The reinforcement schedule for the 2415 RCBC is shown in Table 4-5.

Page | 95
 Table 4-5 - 2415 RCBC reinforcement schedule

Total mass
Bar Mark Grade & Size Qty Shape
(kg)
N1201 N12 10 15.37 straight
N1202 N12 16 14.21 straight
N1203 N12 41 84.84 straight
N1204 N12 16 17.05 straight
N1205 N12 80 130.08 Ligs
N1206 N12 50 76.95 Ligs
N2001 N20 88 327.13 L
N2002 N20 19 59.61 straight

Total reinforcement mass: 725.24 kg

4.8 2418 RCBC Design

To design the RCBC with 2.4 m span and 1.8 m leg (2418 RCBC), the cross
section utilised comprised of a 350 mm leg and 400 mm crown. The overall
culvert width was therefore 3.1 m and the overall height was 2.2 m.

The results from the load combination Matlab script finalscript.m (see
appendix D for code) were as shown in Figure 4-30 and Figure 4-31. The loads
on the top of the culvert are uniformly distributed over the entire top of culvert,
as expected since the truncated prism model (see Section 3.2 for details)
distributes the vehicle and construction loads uniformly over the top of the
culvert. The other loads that act on top of the culvert are also uniformly
distributed, namely the fill and the self-weight. There are 20 straight lines in
Figure 4-30, each for a different fill height. The top line is the load when there
is only 0.1 m of fill over the culvert, giving a load of 653.7 kPa for all values of
x (width of culvert). The bottom lines are very close together, which makes it
difficult to distinguish one from the other. However, the last line represents the
load for the case when there is 2.0 m of fill over the culvert, giving 109.3 kPa.

Page | 96
 Symmetric load combinations - top of culvert
700

Fill height: 0.1 m

600 Load: 653.7 kPa

500
Load (kPa)

Fill height: 0.2 m

400 Load: 454.9 kPa

300 Fill height: 0.3 m

Load: 339.0 kPa

200

100
0 500 1000 1500 2000 2500 3000 3500
Width (mm) Fill height: 2.0 m
Load: 109.3 kPa

Figure 4-30 – Loads on top of 2418 RCBC

The worst load case was that for the least amount of fill, conservatively chosen
to be 0.1 m in this study (refer Section 3.2.3). The worst load combination was
the vertical load number 1 (see Figure 4-1) and horizontal load number 10 (see
Figure 4-2). This is expected since these combinations have the highest load
factors. The horizontal loads combinations were the same as for the 1818
RCBC, since the loads on the culvert are the same (SM1600) and the culvert
height is the same. They are shown in Figure 4-31.

Page | 97
 Symmetric load combinations - sides of culvert
350
X: 501 X: 2001 X: 2201
Y: 344.4 Y: 344.4 Y: 347.2

300 X: 1
Y: 316.4

X: 501 X: 2001 X: 2201

Y: 245 Y: 245 Y: 247.8
250
X: 1
Y: 217

X: 501 X: 2001 X: 2201

200 Y: 187 Y: 187 Y: 189.8
Load (kPa)

X: 1
Y: 159

150

100 X: 501 X: 2001 X: 2201

Y: 70.4 Y: 70.4 Y: 73.2
X: 1
Y: 42.4
50

0
0 500 1000 1500 2000 2500
Height from top of culvert (mm)

Figure 4-31 - Loads on the side of the 2418 RCBC

The 2418 RCBC with crown thickness of 400 mm and the leg thickness of 350
mm is then modelled in Strand7 and the worst load case (for 0.1 m fill) is
applied. The bending moment diagram and shear force diagram are then found
as per Figure 4-32 and Figure 4-33.

Page | 98
Figure 4-32 - Bending moment diagram for 2418 RCBC

Figure 4-33 - Shear force diagram for 2418 RCBC

Page | 99
The next step is to perform a flexibility and shear analysis to determine the
suitable reinforcement for the unit, using the Matlab scripts
flexanalysis.m and checkshear.m, as described in detail in Section
3.3.5 - Non-optimised Culverts and Section 3.3.6 - Reinforcement.

Flexure and shear analysis for the middle of the crown

For the middle of the crown, M*=741.4 kNm (Figure 4-32) and by running
flexanalysis.m with 19-N20 bars as tensile reinforcement, the results are:

>> flexanalysis(400,19,20,0,12);
dn = 41.2800
ku = 0.1163
Cc = 2.9474e+03
Cs = 0
Mu = 1.0037e+03
phiMu = 802.9922
Number of N12 bars required for compression: 0
Number of N20 bars required for tension: 19

This means that:

• the section is under-reinforced since ku<0.36;

• φ M u > M * is suitable and so is the chosen reinforcement

The development length of the tension bars is checked using the Matlab
function devlength.m (see Matlab code in Appendix E) as per described in
Section 3.3.6. The result for 19-N20 in tension, with cover=35 mm and k1=1
is Lsy.tb = 580 mm. This means this bar has to be at least 1060 mm long.

The shear for the middle of the crown is 0 so a shear check does not need to
be carried out and no shear reinforcement is required.

Page | 100
The same procedure is utilised for the end of the crown, the top of the leg and
the bottom of the leg.

Flexure and shear analysis for end of the crown

The design moment at the end of the crown is M*=741.8 kNm (see Figure
4-32). By running flexanalysis.m for 19-N20 bars in tension, the results
are:

>> flexanalysis(400,19,20,0,12)
dn = 41.2800
ku = 0.1163
Cc = 2.9474e+03
Cs = 0
Mu = 1.0037e+03
phiMu = 802.9922
Number of N12 bars required for compression: 0
Number of N20 bars required for tension: 19

This means that:

• the section is under-reinforced since ku<0.36;

• φ M u > M * is suitable and so is the chosen reinforcement

The design shear force at the end of the crown is V*=2157.4 kN (see Figure
4-33). By running checkshear.m for the 19-N20 bars in tension, the results
are:

>> checkshear(2157.4E3,20,19,400);
Vumax = 8520000
Vuc = 8.1887e+05
Vumin = 1.4213e+06
Vusmin = 2.2631e+06
s_vusmin = 69.0196
s = 57.1429

Page | 101
Shear reinforcement is required. Provide 8-N12 bars at
57 mm spacings

This means that (2157400 = V * ) > (φVu .min = 994910) and 8-N12 ligatures
will be provided at 57 mm spacing for an extent of D=400 mm.

The development length of the tension bars is checked using the Matlab
function devlength.m (see Matlab code in Appendix E) as per described in
Section 3.3.6. The result for 19-N20 in tension, with cover=35 mm and k1=1
is Lsy.tb = 580 mm. This means that each leg of this L bar this bar has to be at
least 580 mm long.

Flexure and shear analysis for top of the leg

The design moment at the top of the leg is the same as the one at the end of
the crown, as expected, and is M*=741.8 kNm (seeFigure 4-32). However, this
section is thinner since the leg is 350 mm and the crown is 400 mm. By running
flexanalysis.m for 23-N20 bars in tension, the results are:

>>flexanalysis(350,23,20,0,12) ;
dn = 49.9700
ku = 0.1638
Cc = 3.5679e+03
Cs = 0
Mu = 1.0258e+03
phiMu = 820.6373
Number of N12 bars required for compression: 0
Number of N20 bars required for tension: 23

This means that:

• the section is under-reinforced since ku<0.36;

• φ M u > M * is suitable and so is the chosen reinforcement

Page | 102
A shear check is required to check if shear reinforcement is necessary. The
design shear force at the top of the leg is V*=1183 kN (see Figure 4-33). By
running checkshear.m for the 23-N20 bars in tension, the results are:

>> checkshear(1183E3,20,23,350)
Vumax = 7320000
Vuc = 8.2040e+05
Vumin = 1.3380e+06
Vusmin = 8.6960e+05
s_vusmin = 96.4518
s = 87.5000
Shear reinforcement is required. Provide 5-N12 bars at
88 mm spacings

This means that (1183000 = V * ) > (φVu.min = 936600) and 5-N12 ligatures
will be provided at 88 mm spacing for an extent of D=350 mm.

The development length of the tension bars is checked using the Matlab
function devlength.m (see Matlab code in Appendix E) as per described in
Section 3.3.6. The result for 23-N20 in tension, with cover=35 mm and k1=1.3
is Lsy.tb = 754 mm. This means that each leg of this L bar this bar has to be at
least 754 mm long.

Flexure and shear analysis for bottom of the leg

The design moment at the bottom third of the leg is M*=125 kNm (seeFigure
4-32). By running flexanalysis.m for 10-N12 bars in tension, the results
are:

>> flexanalysis(350,10,12,0,12) ;
dn = 7.7100
ku = 0.0250
Cc = 550.4940
Cs = 0
Mu = 168.6171

Page | 103
phiMu = 134.8937
Number of N12 bars required for compression: 0
Number of N12 bars required for tension: 10

This means that:

• the section is under-reinforced since ku<0.36;

• φ M u > M * is suitable and so is the chosen reinforcement

A shear check is required to check if shear reinforcement is necessary. The

design shear force at the bottom of the leg is V*=455.1 kN (seeFigure 4-33).
By running checkshear.m for the 10-N12 bars in tension, the results are:

>> checkshear(455.1E3,12,10,350)
Vumax = 7416000
Vuc = 4.4247e+05
Vumin = 9.6686e+05
Asvmin = 356.3818
ans =Minimum shear reinforcement is required.

This means that 0.5φVuc < V * ≤ φVu.min and Asv.min will be provided for an

extent of D=350 mm. In this case, V * > φVuc so the minimum shear
reinforcement requirements cannot be waived. Using the maximum spacing of
s=0.5D=175 mm from AS3600-2009 clause 8.2.12.2 into Equation 3-4, the
result is Asv.min = 356.3818 mm, which means there will be 4-N12 ligatures
required at 116 mm spacings.

The development length of the tension bars is checked using the Matlab
function devlength.m (see Matlab code in Appendix E) as per described in
Section 3.3.6. The result for 10-N12 bars in tension, with cover=35 mm and
k1=1 is Lsy.tb = 348 mm. This means this bar has to be at least 696 mm long.

The final 2418 RCBC reinforcement is shown diagrammatically in Figure 4-34.

Page | 104
 Figure 4-34 - 2418 RCBC reinforcement

In order to simplify the assembly of the reinforcement cages, especially with

relation to the ligatures, the L bars at the end of the crown / top of the leg
(N2001) were made the same length with the same spacing. The bars N1201
and N1204 were included to enable the connection of the distribution bars for
cracking and the shear ligatures.

The reinforcement schedule for the 2418 RCBC is shown in Table 4-6.

Page | 105
 Table 4-6 - 2418 RCBC reinforcement schedule

Total mass
Bar Mark Grade & Size Qty Shape
(kg)
N1201 N12 10 15.37 straight
N1202 N12 20 23.18 straight
N1203 N12 45 93.11 straight
N1204 N12 20 28.42 straight
N1205 N12 80 130.08 Ligs
N1206 N12 90 138.51 Ligs
N2001 N20 92 342.00 L
N2002 N20 19 69.46 straight

Total reinforcement mass: 840.13 kg

4.9 Conclusion
The iterative process of designing the non-optimised RCBC to find its
reinforcement layout can be time-consuming if done by hand or with primitive
software, like in the case of this study. Manufacturers in the industry have
access to a great variety of more advanced software that can make this
procedure easy and quick, since all parameters are dictated by Australian
Standards and Main Roads Standards.

An interesting result from this section is the fact that the fill height makes a
great difference in how much load the box culvert is subjected to. There were
great differences in the load supported by the RCBCs depending on the fill
height, especially between 0.1m and 0.4m of fill. This means that the greater
the fill height in real life installations, the more conservative the culvert design
will have been, which can dramatically increase the factor of safety when
utilising these structures.

In summary, due to the fact that different designers can come up with different
compliant structures, the optimised culverts presented in the next section could
end up very different, even though they would all comply with specifications

Page | 106
and be fit for purpose. It would be valid to investigate manners to optimise the
reinforcement, but it is not within the scope of this project.

Page | 107
 CHAPTER 5 - TOPOLOGY OPTIMISATION

5.1 Introduction
The topology optimisation procedure was implemented by using a Matlab
program called top_rcbc4.m, (see Appendix F for code) which was adapted
from Sigmund (2001). The modifications made to the program as well as the
explanation of what each part of the program does can be found in Section
3.4.1.

5.2 1815 RCBC

The topology optimisation Matlab program is applied to the 1815 RCBC with
the following input parameters:

span in mm = 1800
leg height in mm = 1500
volume constraint = 0.25
penalisation factor = 3
filter size divided by the element size (rmin) = 1.2

The command top_rcbc4(1800,1500,0.25,3,1.2) yields the result

shown in Figure 5-1. The simulation comprised of 135 iterations and took 138
seconds.

Page | 108
 Figure 5-1 - Topology optimisation result for 1815 RCBC

The voids in the optimised culvert are to be created only where the colour map
shows white, that is, where the relative density is 0.001. Grey areas mean
composite materials and since this study is not going to utilise composite
materials, only the white parts can become voids. The culvert would look like
Figure 5-2.

Figure 5-2 – Optimised 1815 RCBC section elevation

Page | 109
However, it can be seen the voids are at times too close to the edge of the
culvert. We know the leg reinforcement for the 1815 RCBC leg is comprised
of N16 bars with N12 distribution bars. That means the distance between the
edge of the void to the edge of the leg needs to be a minimum of
35+35+16+12=98 mm, which is shown by the dashed line in Figure 5-3.

Figure 5-3 - Optimised 1815 with trimmed voids

The voids then need to be trimmed to be within the dashed line. Also, the
reinforcement has to fit in and there needs to be 35 mm cover at all times. The
1815 RCBC with trimmed voids is shown in Figure 5-4 with superimposed
reinforcement. It can be seen that some voids will not be possible because they
coincide with the shear ligatures.

Page | 110
 Figure 5-4 - Optimised 1815 with trimmed voids and reinforcement

If the impossible voids are removed, the final optimised culvert would look like
shown in Figure 5-5.

Figure 5-5 - Final 1815 optimised RCBC

Page | 111
5.3 1818 RCBC
The topology optimisation Matlab program is applied to the 1818 RCBC with
the following input parameters:

span in mm = 1800
leg height in mm = 1800
volume constraint = 0.25
penalisation factor = 3
filter size divided by the element size (rmin) = 1.2

The command top_rcbc4(1800,1800,0.25,3,1.2) yields the result

shown in Figure 5-11. The simulation comprised of 135 iterations and took 406
seconds.

Figure 5-6 - Topology optimisation result for 1818 RCBC

The voids in the optimised culvert are to be created only where the colour map
shows white, that is, where the relative density is 0.001. Grey areas mean
composite materials and since this study is not going to utilise composite

Page | 112
materials, only the white parts can become voids. The culvert would look like
Figure 5-12.

Figure 5-7 – Optimised 1815 RCBC section elevation

However, it can be seen the voids are at times too close to the edge of the
culvert. We know the leg reinforcement for the 1815 RCBC leg is comprised
of N16 bars with N12 distribution bars. That means the distance between the
edge of the void to the edge of the leg needs to be a minimum of
35+35+16+12=98 mm, which is shown by the dashed line in Figure 5-3.

Page | 113
 Figure 5-8 - Optimised 1815 RCBC with trimmed voids

The voids then need to be trimmed to be within the dashed line. Also, the
reinforcement has to fit in and there needs to be 35 mm cover at all times. The
1818 RCBC with trimmed voids is shown in Figure 5-14 with superimposed
reinforcement. It can be seen that some voids will not be possible because they
coincide with the shear ligatures.

Page | 114
 Figure 5-9 - Optimised 1815 RCBC with trimmed voids and reinforcement

If the impossible voids are removed, the final optimised culvert would look like
shown in Figure 5-15.

Page | 115
 Figure 5-10 - Final 1818 optimised RCBC

5.4 2412 RCBC

The topology optimisation Matlab program is applied to the 2412 RCBC with
the following input parameters:

span in mm = 2400
leg height in mm = 1200
volume constraint = 0.25
penalisation factor = 3
filter size divided by the element size (rmin) = 1.2

Page | 116
The command top_rcbc4(2400,1200,0.25,3,1.2) yields the result
shown in Figure 5-11. The simulation comprised of 116 iterations and took 119
seconds.

Figure 5-11 - Topology optimisation result for 2412 RCBC

The voids in the optimised culvert are to be created only where the colour map
shows white, that is, where the relative density is 0.001. Grey areas mean
composite materials and since this study is not going to utilise composite
materials, only the white parts can become voids. The culvert would look like
Figure 5-12.

Page | 117
 Figure 5-12 – Optimised 2412 RCBC section elevation

However, it can be seen the voids are at times too close to the edge of the
culvert. We know the leg reinforcement for the 1815 RCBC leg is comprised
of N16 bars with N12 distribution bars. That means the distance between the
edge of the void to the edge of the leg needs to be a minimum of
35+35+20+12=102 mm, which is shown by the dashed line in Figure 5-13.

Page | 118
 Figure 5-13 - Optimised 2412 RCBC with trimmed voids

The voids then need to be trimmed to be within the dashed line. Also, the
reinforcement has to fit in and there needs to be 35 mm cover at all times. The
2412 RCBC with trimmed voids is shown in Figure 5-14 with superimposed
reinforcement. It can be seen that some voids will not be possible because they
coincide with the shear ligatures.

Page | 119
 Figure 5-14 - Optimised 2412 RCBC with trimmed voids and reinforcement

If the impossible voids are removed, the final optimised culvert would look like
shown in Figure 5-15.

Figure 5-15 - Final 2412 optimised RCBC

Page | 120
5.5 2415 RCBC
The topology optimisation Matlab program is applied to the 2415 RCBC with
the following input parameters:

span in mm = 2400
leg height in mm = 1500
volume constraint = 0.25
penalisation factor = 3
filter size divided by the element size (rmin) = 1.2

The command top_rcbc4(2400,1500,0.25,3,1.2) yields the result

shown in Figure 5-16. The simulation comprised of 125 iterations and took 236
seconds.

Figure 5-16 - Topology optimisation result for 2415 RCBC

The voids in the optimised culvert are to be created only where the colour map
shows white, that is, where the relative density is 0.001. Grey areas mean
composite materials and since this study is not going to utilise composite

Page | 121
materials, only the white parts can become voids. The culvert would look like
Figure 5-17.

Figure 5-17 – Optimised 2415 RCBC section elevation

However, it can be seen the voids are at times too close to the edge of the
culvert. We know the leg reinforcement for the 2415 RCBC leg is comprised
of N20 bars with N12 distribution bars. That means the distance between the
edge of the void to the edge of the leg needs to be a minimum of
35+35+20+12=102 mm, which is shown by the dashed line in Figure 5-18.

Page | 122
 Figure 5-18 - Optimised 2415 RCBC with trimmed voids

The voids then need to be trimmed to be within the dashed line. Also, the
reinforcement has to fit in and there needs to be 35 mm cover at all times. The
2415 RCBC with trimmed voids is shown in Figure 5-19 with superimposed
reinforcement. It can be seen that some voids will not be possible because they
coincide with the shear ligatures.

Page | 123
 Figure 5-19 - Optimised 2415 RCBC with trimmed voids and reinforcement

If the impossible voids are removed, the final optimised culvert would look like
shown in Figure 5-20.

Figure 5-20 - Final 2418 optimised RCBC

Page | 124
5.6 2418 RCBC
The topology optimisation Matlab program is applied to the 2418 RCBC with
the following input parameters:

span in mm = 2400
leg height in mm = 1800
volume constraint = 0.25
penalisation factor = 3
filter size divided by the element size (rmin) = 1.2

The command top_rcbc4(2400,1800,0.25,3,1.2) yields the result

shown in Figure 5-21. The simulation comprised of 153 iterations and took 278
seconds.

Figure 5-21 - Topology optimisation result for 2418 RCBC

The voids in the optimised culvert are to be created only where the colour map
shows white, that is, where the relative density is 0.001. Grey areas mean
composite materials and since this study is not going to utilise composite

Page | 125
materials, only the white parts can become voids. The culvert would look like
Figure 5-22.

Figure 5-22 – Optimised 2418 RCBC section elevation

However, it can be seen the voids are at times too close to the edge of the
culvert. We know the leg reinforcement for the 2418 RCBC leg is comprised
of N20 bars with N12 distribution bars. That means the distance between the
edge of the void to the edge of the leg needs to be a minimum of
35+35+20+12=102 mm, which is shown by the dashed line in Figure 5-23.

Page | 126
 Figure 5-23 - Optimised 2418 with trimmed voids

The voids then need to be trimmed to be within the dashed line. Also, the
reinforcement has to fit in and there needs to be 35 mm cover at all times. The
2418 RCBC with trimmed voids is shown in Figure 5-24 with superimposed
reinforcement. It can be seen that some voids will not be possible because they
coincide with the shear ligatures.

Page | 127
 Figure 5-24 - Optimised 2418 RCBC with trimmed voids and reinforcement

If the impossible voids are removed, the final optimised culvert would look like
shown in Figure 5-25.

Page | 128
 Figure 5-25 - Final 2418 optimised RCBC

5.7 Conclusion
The topology optimisation procedure is greatly simplified by the usage of the
script adapted from Sigmund (2001). Great results have been obtained that did
not present checkerboards or too much composite material and voids were able
to be used in the culverts without issues. Also, the computation time is not
significant and is not likely to be prohibitive to the procedure’s adoption. The
presence of shear ligatures greatly influences the final topology of the structure,
in this case. Any decrease in the need for shear ligatures will likely result in great
usage of voids in the culverts.

The successful application of this procedure as was done in this study depends
on some dealing with Matlab and some drafting in AutoCad, which are two
things most engineers should be familiar with. Therefore, it is safe to say most
engineers would not have difficulty with the procedure and would get familiar
with it quickly and be able to efficiently apply the procedure to a variety of
RCBCs.

Page | 129
The adoption of this procedure by manufacturers would greatly depend on
feasibility of the whole exercise, which is analysed and discussed in the next
section.

Page | 130
 CHAPTER 6 - FEASIBILITY ANALYSIS

6.1 Introduction
Because one of the focus points of this project is to draw conclusions regarding
the optimum topology of RCBC commonly used in the industry, it is
paramount that costs are analysed to indicate the feasibility of the method.
During the design phase, the costs and time required to perform the SIMP
analysis will be compared with estimated costs and time required to run a
similar design for a standard RCBC sold by manufacturers. In relation to
manufacture, the complexity and cost of the reinforcement cage required will
be analysed and compared with estimated costs used by manufacturers to
produce non-optimised culverts. Regarding transportation, the cost and time
required to load, unload and transport the optimum RCBC will be compared
between the optimised and non-optimised culverts.

These results will be contrasted with the benefits that using this optimisation
procedure may bring and a conclusion will be reached regarding feasibility of
designing, manufacturing and installing the optimum RCBC.

6.2 Labour and design components of total cost

Because the culverts were not very dissimilar in size, cost changes to allow for
labour and design costs for the optimised culvert were equally applied through
all culverts. Extra labour hours would be required to design the optimised
culverts and to organise the voids. It is important to note that the labour and
overhead costs are calculated based on the mass of the entire unit. The
reinforcement is exactly the same, since the voids are trimmed to fit the
reinforcement configuration. The loading procedures would also be the same
since the centre of gravity of the optimised units was only raised by a few
millimetres to a maximum of 24 mm and that would not pose any problem to
transporting the culverts legs down on the truck. It was assumed in this study
that the cost of labour would increase in 20% to allow for these extra required

Page | 131
activities. As per discussed in Section 3.5.2, one man-hour is assumed to cost
$40.

In addition, the optimised culvert was assumed to take twice as long to design
than the non-optimised culvert because once the non-optimised culvert has
been found, the topology optimisation procedure has to be carried out and will
include some designing and some drafting. However, this assumption leans on
the conservative side because once the designer is used to the software utilised
to carry out the optimisation, design time will become shorter and shorter.
Also, it is possible to improve the optimisation programmes that were utilised
for this study by condensing all functions into one or two programmes to
increase ease of use. That would also mean design time could be reduced.

6.3 1815 RCBC

6.3.1 Materials
The non-optimised 1815 RCBC would utilise 4.92 m3 of concrete minus what
would be displaced by the reinforcement. The reinforcement was found to
weigh 508.6 kg and the density of the steel is assumed 7850 kg/m3 (Standards
Australia 2001, p. 14). Therefore, 508.6 kg of steel would displace 0.0648 m3of
concrete. The total concrete volume would then be 4.86 m3, which at a density
of 2400 kg/m3 equates to 11.657 tonnes. The total mass of the culvert is
therefore 12.166 tonnes.

The optimised 1815 RCBC would utilise 4.55 m3 of concrete minus the 0.0628
m3 displaced by the reinforcement, which gives 4.48 m3. At a density of 2400
kg/m3, this equates to 10.758 tonnes. The total mass of the culvert is therefore
11.251 tonnes. The difference is 899 kg of concrete, representing 7.4% of the
total initial amount of concrete, before the optimisation.

6.3.2 Conclusion
The pertinent parameters namely materials, design, labour and overheads for
the non-optimised and optimised RCBCs were compiled into Table 6-1 and
Table 6-2.

Page | 132
 Table 6-1 - Cost summary of non-optimised 1815 RCBC

COSTS OF NON-OPTIMISED 1815 RCBC

PARAMATER QTY UNIT COST PER UNIT TOTAL COST ($)
Design 1.5 hours 80 $/hour of design $ 120.00
Materials: Steel 508.6 kg 1300 $/tonne of steel $ 661.18
Materials: Concrete 4.86 m3 140 $/m3 of concrete $ 680.40
Labour 2.5 man-hours/tonne $ 1,217.26
Overheads 70 $/tonne $ 852.08
TOTAL COST: $ 3,530.92

Table 6-2 - Cost summary of optimised 1815 RCBC

COSTS OF OPTIMISED 1815 RCBC

PARAMATER QTY UNIT COST PER UNIT TOTAL COST ($)
Design 3 hours 80 $/hour of design $ 240.00
Materials: Steel 508.6 kg 1300 $/tonne of steel $ 661.18
Materials: Concrete 4.48 m3 140 $/m3 of concrete $ 627.20
Labour 3 man-hours/tonne $ 1,351.27
Overheads 70 $/tonne $ 788.24
TOTAL COST: $ 3,667.89

The total manufacture cost of the optimised culvert is approximately 3.9%

higher than the non-optimised culvert.

6.4 1818 RCBC

6.4.1 Materials
Regarding materials, the non-optimised 1818 RCBC would utilise 5.43 m3 of
concrete minus what would be displaced by the reinforcement. The
reinforcement was found to weigh 663.33 kg and the density of the steel is
assumed 7850 kg/m3 (Standards Australia 2001, p. 14). Therefore, 663.33 kg of
steel would displace 0.0845 m3of concrete. The total concrete volume would
then be 5.35 m3, which at a density of 2400 kg/m3 equates to 12.829 tonnes.
The total mass of the culvert is therefore 13.493 tonnes.

The optimised 1818 RCBC would utilise 5.05 m3 of concrete minus what would
be displaced by the reinforcement. The reinforcement was found to weigh
663.33 kg and it would displace 0.0845 m3of concrete. The total concrete

Page | 133
volume would then be 4.97 m3, which at a density of 2400 kg/m3 equates to
11.917 tonnes. The total mass of the culvert is therefore 12.581 tonnes. The
difference is 912 kg of concrete, representing 6.8% of the total initial amount
of concrete, before the optimisation.

Extra labour hours would be required to design the optimised culvert and to
organise the voids. The reinforcement is be exactly the same, since the voids
were trimmed to fit the reinforcement configuration. The loading procedures
would also be the same since the centre of gravity of the optimised unit was
only raised by 10 mm towards the crown at 1.386 m from the leg, and that
would not pose any problem to transporting it legs down on the truck.

6.4.2 Conclusion
The pertinent parameters namely materials, design, labour and overheads for
the non-optimised and optimised RCBCs were compiled into Table 6-3 and
Table 6-4.

Table 6-3 - Cost summary of non-optimised 1818 RCBC

COSTS OF NON-OPTIMISED 1818 RCBC

PARAMATER QTY UNIT COST PER UNIT TOTAL COST ($)
Design 1.5 hours 80 $/hour of design $ 120.00
Materials: Steel 663.3 kg 1300 $/tonne of steel $ 862.33
Materials: Concrete 5.35 m3 140 $/m3 of concrete $ 749.00
Labour 2.5 man-hours/tonne $ 1,350.33
Overheads 70 $/tonne $ 945.23
TOTAL COST: $ 4,026.90

Table 6-4 - Cost summary of optimised 1818 RCBC

COSTS OF OPTIMISED 1818 RCBC

PARAMATER QTY UNIT COST PER UNIT TOTAL COST ($)
Design 3 hours 80 $/hour of design $ 240.00
Materials: Steel 663.3 kg 1300 $/tonne of steel $ 862.33
Materials: Concrete 4.97 m3 140 $/m3 of concrete $ 695.80
Labour 3 man-hours/tonne $ 1,510.96
Overheads 70 $/tonne $ 881.39
TOTAL COST: $ 4,190.48

Page | 134
The total manufacture cost of the optimised culvert is approximately 4.06%
higher than the non-optimised culvert.

6.5 2412 RCBC

6.5.1 Materials
The non-optimised 2412 RCBC would utilise 4.99 m3 of concrete minus what
would be displaced by the reinforcement. The reinforcement was found to
weigh 725.49 kg and the density of the steel is assumed 7850 kg/m3 (Standards
Australia 2001, p. 14). Therefore, 725.49 kg of steel would displace 0.0924 m3of
concrete. The total concrete volume would then be 4.90 m3, which at a density
of 2400 kg/m3 equates to 11.754 tonnes. The total mass of the culvert is
therefore 12.480 tonnes.

The optimised 2412 RCBC would utilise 4.56 m3 of concrete minus what would
be displaced by the reinforcement. The reinforcement was found to weigh
725.49 kg and it would displace 0.0924 m3of concrete. The total concrete
volume would then be 4.47 m3, which at a density of 2400 kg/m3 equates to
10.722 tonnes. The total mass of the culvert is therefore 11.448 tonnes. The
difference is 1032 kg of concrete, representing 8.8% of the total initial amount
of concrete, before the optimisation.

6.5.2 Conclusion
The pertinent parameters namely materials, design, labour and overheads for
the non-optimised and optimised RCBCs were compiled into Table 6-5 and
Table 6-6.

Page | 135
 Table 6-5 - Cost summary of non-optimised 2412 RCBC

COSTS OF NON-OPTIMISED 2412 RCBC

PARAMATER QTY UNIT COST PER UNIT TOTAL COST ($)
Design 1.5 hours 80 $/hour of design $ 120.00
Materials: Steel 725.5 kg 1300 $/tonne of steel $ 943.14
Materials: Concrete 4.9 m3 140 $/m3 of concrete $ 686.00
Labour 2.5 man-hours/tonne $ 1,248.55
Overheads 70 $/tonne $ 873.98
TOTAL COST: $ 3,871.67

Table 6-6 - Cost summary of optimised 2412 RCBC

COSTS OF OPTIMISED 2412 RCBC

PARAMATER QTY UNIT COST PER UNIT TOTAL COST ($)
Design 3 hours 80 $/hour of design $ 240.00
Materials: Steel 725.5 kg 1300 $/tonne of steel $ 943.14
Materials: Concrete 4.47 m3 140 $/m3 of concrete $ 625.80
Labour 3 man-hours/tonne $ 1,374.42
Overheads 70 $/tonne $ 801.74
TOTAL COST: $ 3,985.10

The total manufacture cost of the optimised culvert is approximately 2.95%

higher than the non-optimised culvert.

6.6 2415 RCBC

6.6.1 Materials
The non-optimised 2415 RCBC would utilise 5.50 m3 of concrete minus what
would be displaced by the reinforcement. The reinforcement was found to
weigh 725.24 kg and the density of the steel is assumed 7850 kg/m3 (Standards
Australia 2001, p. 14). Therefore, 725.24 kg of steel would displace 0.0924 m3of
concrete. The total concrete volume would then be 5.41 m3, which at a density
of 2400 kg/m3 equates to 12.978 tonnes. The total mass of the culvert is
therefore 13.704 tonnes.

The optimised 2415 RCBC would utilise 5.09 m3 of concrete minus what would
be displaced by the reinforcement. The reinforcement was found to weigh
725.24 kg and it would displace 0.0924 m3of concrete. The total concrete

Page | 136
volume would then be 5.00 m3, which at a density of 2400 kg/m3 equates to
11.994 tonnes. The total mass of the culvert is therefore 12.719 tonnes. The
difference is 984 kg of concrete, representing 7.6% of the total initial amount
of concrete, before the optimisation.

6.6.2 Conclusion
The pertinent parameters namely materials, design, labour and overheads for
the non-optimised and optimised RCBCs were compiled into Table 6-7 and
Table 6-8.

Table 6-7 - Cost summary of non-optimised 2415 RCBC

COSTS OF NON-OPTIMISED 2415 RCBC

PARAMATER QTY UNIT COST PER UNIT TOTAL COST ($)
Design 1.5 hours 80 $/hour of design $ 120.00
Materials: Steel 725.2 kg 1300 $/tonne of steel $ 942.81
Materials: Concrete 5.41 m3 140 $/m3 of concrete $ 757.40
Labour 2.5 man-hours/tonne $ 1,370.92
Overheads 70 $/tonne $ 959.65
TOTAL COST: $ 4,150.78

Table 6-8 - Cost summary of optimised 2415 RCBC

COSTS OF OPTIMISED 2415 RCBC

PARAMATER QTY UNIT COST PER UNIT TOTAL COST ($)
Design 3 hours 80 $/hour of design $ 240.00
Materials: Steel 725.2 kg 1300 $/tonne of steel $ 942.81
Materials: Concrete 5 m3 140 $/m3 of concrete $ 700.00
Labour 3 man-hours/tonne $ 1,527.03
Overheads 70 $/tonne $ 890.77
TOTAL COST: $ 4,300.61

The total manufacture cost of the optimised culvert is approximately 3.60%

higher than the non-optimised culvert.

6.7 2418 RCBC

6.7.1 Materials
The non-optimised 2418 RCBC would utilise 6.0 m3 of concrete minus what
would be displaced by the reinforcement. The reinforcement was found to

Page | 137
weigh 840.13 kg and the density of the steel is assumed 7850 kg/m3 (Standards
Australia 2001, p. 14). Therefore, 840.13 kg of steel would displace 0.107 m3of
concrete. The total concrete volume would then be 5.89 m3, which at a density
of 2400 kg/m3 equates to 14.143 tonnes. The total mass of the culvert is
therefore 14.983 tonnes.

The optimised 2418 RCBC would utilise 5.74 m3 of concrete minus what would
be displaced by the reinforcement. The reinforcement was found to weigh
840.13 kg and it would displace 0.107 m3of concrete. The total concrete volume
would then be 5.634 m3, which at a density of 2400 kg/m3 equates to 13.522
tonnes. The total mass of the culvert is therefore 14.362 tonnes. The difference
is 621 kg of concrete, representing 4.4% of the total initial amount of concrete,
before the optimisation.

6.7.2 Conclusion
The pertinent parameters namely materials, design, labour and overheads for
the non-optimised and optimised RCBCs were compiled into Table 6-9 and
Table 6-10.

Table 6-9 - Cost summary of non-optimised 2418 RCBC

COSTS OF NON-OPTIMISED 2418 RCBC

PARAMATER QTY UNIT COST PER UNIT TOTAL COST ($)
Design 1.5 hours 80 $/hour of design $ 120.00
Materials: Steel 840.1 kg 1300 $/tonne of steel $ 1,092.17
Materials: Concrete 5.89 m3 140 $/m3 of concrete $ 824.60
Labour 2.5 man-hours/tonne $ 1,497.61
Overheads 70 $/tonne $ 1,048.33
TOTAL COST: $ 4,582.71

Page | 138
 Table 6-10 - Cost summary of optimised 2418 RCBC

COSTS OF OPTIMISED 2418 RCBC

PARAMATER QTY UNIT COST PER UNIT TOTAL COST ($)
Design 3 hours 80 $/hour of design $ 240.00
Materials: Steel 840.1 kg 1300 $/tonne of steel $ 1,092.17
Materials: Concrete 5.634 m3 140 $/m3 of concrete $ 788.76
Labour 3 man-hours/tonne $ 1,723.41
Overheads 70 $/tonne $ 1,005.32
TOTAL COST: $ 4,849.66

The total manufacture cost of the optimised culvert is approximately 5.85%

higher than the non-optimised culvert.

6.8 Conclusions
The difference between the cost of the non-optimised culvert and the
optimised culvert seems closely related to the leg height. The higher the leg
height the higher the difference. Similar leg heights presented similar
differences: for the 1815 the difference was 3.9% and for the 2415 the
difference was 3.6% while for the 1818 the difference was 4.06% and for the
2418 it was 8.85%. This would need to be confirmed by performing an analysis
of all the 24 sizes of culverts, but it would make sense that the price is related
to the leg height since it is in the leg that the majority of the voids is located.

Labour is the biggest factor in the cost of an RCBC, representing from 32% to
37% of the costs of the culvert over the 5 units studied. As pointed out by
Roome (2014), labour would be a key parameter to reduce the costs of the
RCBC.

Design represents between 2.9% and 3.4% of the costs of the non-optimised
unit and between 5.0% and 6.6% of the costs of the optimised unit. This means
that efforts put towards diminishing design time of the optimised culvert will
yield much smaller savings than those put towards standardization of the
optimised units to save on labour costs.

Page | 139
Even though the optimised culvert ended up slightly more expensive than the
non-optimised culvert, it is still very possible that it may be feasible for the
manufacturers to offer that option to the customers. One reason is to
accommodate customers’ requirements. If they need slots to pass cables or for
some other application these requirements can be met by the optimised culvert.
Another reason would be that it can be sold as a more environmentally
sustainable option because it saves around 5% in concrete utilisation. Also, if
methods to standardise the procedure or to reduce labour costs somehow can
be found, it will not take much for the costs of the two culverts to equalise.

Page | 140
 CHAPTER 7 - CONCLUSIONS

7.1 Achievements
This project involved the analysis of SM1600 loads over large reinforced
concrete box culverts (RCBC); the design of non-optimised RCBC units based
on current AS3600-2009 specifications; the determination of the optimum
topology for the five most economically significant RCBCs; and a feasibility
analysis to determine in which situations the application of the process outlined
by this project would be worthwhile.

To aid the accomplishment of these steps, various Matlab programs were

written to calculate load combinations, perform flexibility and shear analysis
and find critical horizontal and vertical loads acting on the culverts. In addition,
a Matlab program was adapted from Sigmund (2001) to perform the topology
optimisation utilising the Solid Isotropic Material with Penalization (SIMP)
method with finite element analysis.

In order to ensure the results achieved by this study were as applicable as

possible to the industry, an interview was carried out with an experienced
manager with over 20 years’ experience in the precast concrete industry. The
information obtained through this interview was invaluable to guaranteeing the
models in this study were as close as possible to the real situations in the
industry.

7.2 Conclusions
The topology optimisation procedure presented in this project yields good
results that can be applied in the industry. It is not a complicated procedure
when the designer makes use of programming in the form of Matlab scripts
and the like to perform the calculations. Even though this project analysed
Queensland Main Roads culverts, the same procedure can be used in other

Page | 141
applications such as subdivision culverts, which are subjected to smaller loads
and may enable greater inclusion of voids in their topology.

The Solid Isotropic Material with Penalization (SIMP) method yielded clean
results, with not many grey areas (composite materials) and without
checkerboard issues due to the utilisation of a mesh-independency filter
developed by Sigmund (2001). That made it possible to establish where the
voids would need to be placed and what size and shape they should be.

The optimised culverts offered a reduction of between 4.4% and 8.8% in the
amount of concrete utilised. The steel reinforcement remained the same since
reinforcement optimisation is out of the scope of this project. However, due
to labour costs, the optimised culverts were estimated to be between 2.95% to
5.85% more expensive to manufacture than the non-optimised culverts. If
procedures are put in place to standardise the inclusion of voids in the culverts
and thus reduce labour costs, it is very likely that the optimised culverts could
cost the same or less to manufacture.

7.3 Possible further work

When examining possible further work, one of the first things that has to be
mentioned is the possibility to analyse all culvert sizes to ascertain the level of
savings and feasibility outside the studied interval of RCBCs. Despite the fact
that they do not represent the majority of sales, it is possible that the exercise
is worthwhile depending on the size and scale of the construction job they will
be used in.

Another significant project would be to improve and condense the Matlab

scripts to increase user-friendliness. A savvy Matlab user would not have much
difficulty understanding how each script interacts with the others and how to
apply them to the topology optimisation process, but for the designers who
only have basic knowledge of Matlab the way the scripts are organised can
prove to be challenging to utilise. Any improvement in that area will decrease

Page | 142
design time and increase the chance of adoption of the procedure by the
industry.

In addition, materials other than concrete and steel can be investigated and
different topologies may be found. The usage of fibre reinforced concrete
(FRC), for instance, can prove to minimise issues with cover to reinforcement.

In a more practical way, the optimised culverts found by this project could be
manufactured and load tested, to confirm assumptions made in this study and
provide the industry with proof of its validity.

Page | 143
 REFERENCES

Ahmed, AU & Amanat, KM 2008, 'Need of 3d finite element analysis for

buried RC box culvert in fill soil', in Annual Conference of the Canadian Society
for Civil Engineering 2008 - "Partnership for Innovation", June 10, 2008 - June
13, 2008: proceedings of theAnnual Conference of the Canadian Society for Civil
Engineering 2008 - "Partnership for Innovation", June 10, 2008 - June 13, 2008
Canadian Society for Civil Engineering, Quebec City, QC, Canada, pp. 2078-
87.

Aschheim, M, Hernandez-Montes, E & Gil-Martin, LM 2008, 'Design of

optimally reinforced RC beam, column, and wall sections', Journal of Structural
Engineering, vol. 134, no. 2, pp. 231-9.

Awwad, E, Mabsout, M, Sadek, S & Tarhini, K 2000, 'Finite element analysis

of concrete box culverts', in 8th International Conference on Computing in
Civil and Building Engineering, August 14, 2000 - August 16, 2000: proceedings
of the8th International Conference on Computing in Civil and Building Engineering, August
14, 2000 - August 16, 2000 American Society of Civil Engineers, Stanford, CA,
United states, pp. 1051-3, <http://dx.doi.org/10.1061/40513(279)136>.

Awwad, E, Mabsout, M, Sadek, S & Tarhini, K 2008, 'Parametric study of load

distribution in four-sided concrete box culverts', Bridge Structures, vol. 4, no. 2,
pp. 99-107.

Bendsøe, MP & Sigmund, O 1999, 'Material interpolation schemes in topology

optimization', Archive of Applied Mechanics, vol. 69, no. 9-10, pp. 635-54.

Bruns, TE 2005, 'A reevaluation of the SIMP method with filtering and an
alternative formulation for solid-void topology optimization', Structural and
Multidisciplinary Optimization, vol. 30, no. 6, pp. 428-36.

Chen, B-G, Zheng, J-J & Han, J 2010, 'Experimental study and numerical
simulation on concrete box culverts in trenches', Journal of Performance of
Constructed Facilities, vol. 24, no. 3, pp. 223-34.

China Ministry of Transport 2004, JTG D60-2004 - General Code for Design of
Highway Bridges and Culverts (English), China Communications Press, Beijing.

Das, S 2013, 'Application of diagnostic load testing and 3D-FEA in load rating
of RC box culverts', Bridge Structures, vol. 9, no. 4, pp. 155-67.

Foley Products 2014, Box Culverts,

<http://www.theconcretecompany.com/test/images/box_culvert-l.jpg>.

Page | 144
Foster, S 2010, Reinforced concrete basics 2E : analysis and design of reinforced concrete
structures, 2nd edition. edn, Pearson Education Australia, Frenchs Forest NSW.

Frederick, GR & Tarhini, KM 2000, 'Structural evaluation of three-sided

concrete culverts', in Concrete Pipe for the New Millennium, May 19, 1999 -
May 20, 1999: proceedings of theConcrete Pipe for the New Millennium, May 19, 1999 -
May 20, 1999 ASTM, Seattle, WA, USA, pp. 36-44.

Garg, AK 2006, 'Experimental and finite element based investigations of shear

behavior in reinforced concrete box culverts', ProQuest, UMI Dissertations
Publishing.

Garg, AK & Abolmaali, A 2009, 'Finite-element modeling and analysis of

reinforced concrete box culverts', Journal of Transportation Engineering, vol. 135,
no. 3, pp. 121-8.

Gil-Martin, LM, Aschheim, M, Hernandez-Montes, E & Pasadas-Fernandez,

M 2011, 'Recent developments in optimal reinforcement of RC beam and
column sections', Engineering Structures, vol. 33, no. 4, pp. 1170-80.

Guest, J & Moen, C 2010, 'Reinforced Concrete Design with Topology

Optimization', in Structures Congress 2010, pp. 445-54.

Guest, J, Lotfi, R, Gaynor, A & Jalalpour, M 2012, 'Structural Topology

Optimization: Moving Beyond Linear Elastic Design Objectives', in 20th
Analysis and Computation Specialty Conference, pp. 245-56.

Guest, JK 2009, 'Topology optimization with multiple phase projection',

Computer Methods in Applied Mechanics and Engineering, vol. 199, no. 1–4, pp. 123-
35.

Guest, JK, Prévost, JH & Belytschko, T 2004, 'Achieving minimum length scale
in topology optimization using nodal design variables and projection functions',
International Journal for Numerical Methods in Engineering, vol. 61, no. 2, pp. 238-54.

Hassani, B & Hinton, E 1998, 'A review of homogenization and topology

optimization III—topology optimization using optimality criteria', Computers &
Structures, vol. 69, no. 6, pp. 739-56.

Huang, X 2010, Evolutionary topology optimization of continuum structures methods and

applications, Wiley, Chichester West Sussex UK,
<http://ezproxy.usq.edu.au/login?url=http://site.ebrary.com/lib/unisouther
nqld/Doc?id=10375595>.

Page | 145
Huang, X & Xie, YM 2007, 'Convergent and mesh-independent solutions for
the bi-directional evolutionary structural optimization method', Finite Elements
in Analysis and Design, vol. 43, no. 14, pp. 1039-49.

Kang, J, Parker, F, Kang, YJ & Yoo, CH 2008, 'Effects of frictional forces

acting on sidewalls of buried box culverts', International Journal for Numerical and
Analytical Methods in Geomechanics, vol. 32, no. 3, pp. 289-306.

Kim, D, Bhowmik, SK, Willmer, JL & Leo, E 2011, 'A case history and finite
element modeling of a culvert failure', in Geo-Frontiers 2011: Advances in
Geotechnical Engineering, March 13, 2011 - March 16, 2011: proceedings of
theGeo-Frontiers 2011: Advances in Geotechnical Engineering, March 13, 2011 - March
16, 2011 American Society of Civil Engineers (ASCE), Dallas, TX, United
states, pp. 1641-50, <http://dx.doi.org/10.1061/41165(397)168>.

Kitane, Y & McGrath, TJ 2006, 'Three-dimensional modeling of live loads on

culverts', in Pipelines 2006, July 30, 2006 - August 2, 2006: proceedings of
thePipelines 2006, July 30, 2006 - August 2, 2006 American Society of Civil
Engineers, Chicago, IL, United states, p. 82,
<http://dx.doi.org/10.1061/40854(211)82>.

Loo, Y-C 2010, Reinforced and prestressed concrete analysis and design with emphasis on
application of AS3600-2009, Cambridge University Press, Leiden,
<http://ezproxy.usq.edu.au/login?url=http://www.USQ.eblib.com.au/EBL
Web/patron?target=patron&extendedid=P_581067_0>.

Martín, I, Adeli, H & Sarma, K 1999, 'Cost Optimization of Concrete

Structures', Journal of Structural Engineering, vol. 125, no. 5, pp. 574-5.

McGrath, TJ, Liepins, AA & Beaver, JL 2005, 'Live load distribution widths for
reinforced concrete box sections', in Transportation Research Board - 6th
International Bridge Engineering Conference: Reliability, Security, and
Sustainability in Bridge Engineering, July 17, 2005 - July 20, 2005: proceedings of
theTransportation Research Board - 6th International Bridge Engineering Conference:
Reliability, Security, and Sustainability in Bridge Engineering, July 17, 2005 - July 20,
2005 Transportation Research Board, Boston, MA, United states, pp. 99-108.

Rath, DP, Ahlawat, AS & Ramaswamy, A 1999, 'Shape optimization of RC

flexural members', Journal of structural engineering New York, N.Y., vol. 125, no. 12,
pp. 1439-46.

Rocla 2014, 'Rocla Box Culvert', viewed 09/09/2014.

Roome, C 2014, RCBC Production Costs 14/10/14.

Page | 146
Rozvany, GIN 2001, 'Aims, scope, methods, history and unified terminology
of computer-aided topology optimization in structural mechanics', Structural and
Multidisciplinary Optimization, vol. 21, no. 2, pp. 90-108.

Sarma, KC & Adeli, H 1998, 'Cost optimization of concrete structures', Journal

of structural engineering New York, N.Y., vol. 124, no. 5, pp. 570-8.

Shojaee, S & Mohammadian, M 2012, 'Structural topology optimization using

an enhanced level set method', Scientia Iranica, vol. 19, no. 5, pp. 1157-67.

Sigmund, O 2001, 'A 99 line topology optimization code written in Matlab',

Structural and Multidisciplinary Optimization, vol. 21, no. 2, pp. 120-7.

Sigmund, O & Petersson, J 1998, 'Numerical instabilities in topology

optimization: A survey on procedures dealing with checkerboards, mesh-
dependencies and local minima', Structural optimization, vol. 16, no. 1, pp. 68-75.

Sigmund, O & Maute, K 2013, 'Topology optimization approaches', Structural

and Multidisciplinary Optimization, vol. 48, no. 6, pp. 1031-55.

Standards Australia 2001, AS4671-2001 - Steel Reinforcing Materials, SAI Global

Limites, Sydney, NSW.

Standards Australia 2004a, AS5100.2-2004 - Bridge Design, Part 2: Design Loads,

SAI Global Limited, Sydney, NSW.

Standards Australia 2004b, AS5100.5 - Beidge Design Part 5: Concrete, SAI Global
Limited, Sydney, NSW.

Standards Australia 2007, AS 5100.2 Supp 1-2007 : Bridge design - Design loads -
Commentary (Supplement to AS 5100.2 - 2004).

Standards Australia 2009, AS3600-2009 - Concrete structures, SAI Global Limites,

Sydney, NSW.

Standards Australia 2013, AS1597.2-2013 - Precast reinforced concrete box culverts,

Part 2: Large culverts (exceeding 1200 mm span or 1200 mm height and up to
and including 4200 mm span and 4200 mm height), SAI Global Limited,
Sydney, NSW.

Stanton, A & Javadi, AA 2014, 'An automated approach for an optimised least
cost solution of reinforced concrete reservoirs using site parameters', Engineering
Structures, vol. 60, pp. 32-40.

Strand7 Pty. Ltd. 2010, Using Strand7 - Introduction to the Strand7 Finite Element
Analysis System, Sydney.

Page | 147
Stromberg, L, Beghini, A, Baker, W & Paulino, G 2011, 'Application of layout
and topology optimization using pattern gradation for the conceptual design of
buildings', Structural and Multidisciplinary Optimization, vol. 43, no. 2, pp. 165-80.

Transport and Main Roads 2010a, Technical Note 20a - Design Criteria for Large
Box Culverts to MRTS24.

Transport and Main Roads, Design Criteria for Large Box Culverts to MRTS24,
2010b.

Van Dijk, NP, Maute, K, Langelaar, M & Van Keulen, F 2013, 'Level-set
methods for structural topology optimization: A review', Structural and
Multidisciplinary Optimization, vol. 48, no. 3, pp. 437-72.

Wang, X, Wang, MY & Guo, D 2004, 'Structural shape and topology

optimization in a level-set-based framework of region representation', Structural
and Multidisciplinary Optimization, vol. 27, no. 1-2, pp. 1-19.

Yamada, T, Izui, K, Nishiwaki, S & Takezawa, A 2010, 'A topology

optimization method based on the level set method incorporating a fictitious
interface energy', Computer Methods in Applied Mechanics and Engineering, vol. 199,
no. 45–48, pp. 2876-91.

Zhao, H, Long, K & Ma, ZD 2010, 'Homogenization Topology Optimization

Method Based on Continuous Field', Computers and Structures, vol. 2010, no. 6,
pp. 1-7.

Zhu, M, Yang, Y, Gaynor, A & Guest, J 2014, 'Considering Constructability in

Structural Topology Optimization', in Structures Congress 2014, pp. 2754-64.

Zhu, P, Liu, R, Liu, W & Wu, X 2012, 'Study on optimal design of a box-culvert
under road', in 2012 International Conference on Civil Engineering and Urban
Planning, CEUP, August 18, 2012 - August 20, 2012: proceedings of the2012
International Conference on Civil Engineering and Urban Planning, CEUP, August 18,
2012 - August 20, 2012 American Society of Civil Engineers (ASCE), Yantai,
China, pp. 614-8, <http://dx.doi.org/10.1061/9780784412435.110>.

Page | 148
 APPENDICES

APPENDIX A – PROJECT SPECIFICATION

The latest version of the project specification is as below.

A typed version is included on the next page, to facilitate reading.

Page | 149
 University of Southern Queensland
FACULTY OF ENGINEERING AND SURVEYING
ENG4111/4112 Research Project
PROJECT SPECIFICATION

FOR: JULIANA SWEENEY

TOPIC: Topology optimisation of large reinforced concrete box
culverts under SM1600 loads
SUPERVISOR: Dr Sourish Banerjee (USQ)
PROJECT AIM: This project aims to find the optimum topology of a
large reinforced concrete box culvert (RCBC) under
SM1600 loads using finite element analysis and Solid
Isotropic Material with Penalization (SIMP) method. A
feasibility analysis will then be conducted to ascertain
how much can be saved using the optimum design and
how practical it would be to design, manufacture and
install it in practice.
PROGRAMME: Issue F, 16/10/14

1. Read and analyse literature relating to optimisation of reinforced concrete

products to produce a literature review.
2. Generate 3D finite-element models of standard reinforced concrete large
box culvert (RCBC) sizes and design them as per AS1597.2-2013 under
SM1600 loads.
3. Use the Solid Isotropic Material with Penalization (SIMP) method to
optimise the RCBC topology
4. Perform a feasibility analysis of the optimum box culvert to ascertain the
level of efficiency and cost savings in designing, manufacturing and
installing the optimised structure.

As time permits:

5. Re-analyse culverts including shear reinforcement and draw conclusions

regarding its influence in the optimum design

AGREED:

Student: Juliana Sweeney _______________________ Date: ___/___/2014

Supervisor: Dr. Sourish Banerjee _________________ Date: ___/___/2014

Page | 150
APPENDIX B – FLEXANALYSIS.M

%ANALYSIS AND DESIGN FOR FLEXURAL STRENGTH

function flexanalysis(D,ntension,astbar,ncomp,ascbar)
%INPUTS:
% D=crown thickness in mm
% ascbar= diameter of compression bars in mm
% astbar= diameter of tension bars in mm
% ncomp= quantity of compression bars
% ntension= quantity of tension bars

clc;
fc=50; %MPa
b=2400; %mm
cover=35; %mm
Es=200E3; %Mpa
bars=[10,80;12,110;16,200;20,310;24,450;28,620;32,800
;36,1020;40,1260];

%calcs
[a a1]=find(bars==ascbar);
Asc=ncomp*bars(a,2); %area of steel in
compression
[a a1]=find(bars==astbar);
Ast=ntension*bars(a,2); %area of steel in tension

T=500*Ast*1E-3; %tension force

gamma=1.05-0.007*fc; %gamma
if gamma>0.85
gamma=0.85;
end
if gamma<0.67
gamma=0.67;
end
alpha2=1-.003*fc; %alpha2
if alpha2>0.85
alpha2=0.85;
end
if alpha2<0.67
alpha2=0.67;
end

rep=1; %rep=1 while dn has not

been found

while rep==1
for dn=0.1:0.01:D
Cc=gamma*dn*b*alpha2*fc*1E-3;
esc=0.003*(dn-(cover+ascbar/2))/dn;

Page | 151
 Cs=Es*esc*Asc*1E-3;
a=Cc+Cs;
if (Cc+Cs>=0.999*T) && (Cc+Cs<=1.001*T)
dnfinal=dn;
rep=0;
end
end
end

dn=dnfinal %dn receives dnfinal value

ku=dn/(D-cover-astbar/2)
if ku>=0.36
fprintf('Ku is outside range. Ku must be <
0.36');
end
esc=0.003*(dn-(cover+ascbar/2))/dn
if esc >=0.0025
fprintf('esc is outside range. esc must be <
0.0025');
end
d=D-cover-astbar/2;
dc=0.5*gamma*dn;
dsc=cover+ascbar/2;
Cc=gamma*dn*b*alpha2*fc*1E-3
Cs=Es*esc*Asc*1E-3
Mu=(Cc*(d-dc)+Cs*(d-dsc))/1E3
phiMu=0.8*Mu
formatSpec = 'Number of N%2.0f bars required for
compression: %2.0f\nNumber of N%2.0f bars required
for tension: %2.0f\n';
fprintf(formatSpec,ascbar,ncomp,astbar,ntension);
end

Page | 152
APPENDIX C – CHECKSHEAR.M

function strresult =
checkshear(Vstar,astbar,ntension,D)
% Checks shear requirements according to AS3600
%inputs: Vstar = V* in N
% astbar = diameter of tension bars in mm
% ntension = quantity of tension bars in mm
% D = beam depth in mm
clc;
%Constants:
fc=50; %MPa
bv=2400; %mm
cover=35; %mm
Es=200E3; %Mpa
bars=[10,80;12,110;16,200;20,310;24,450;28,620;32,800
;36,1020;40,1260];
phi=0.7;
Vus=0;
fsyf=500; %using N12 as ligatures

[a a1]=find(bars==astbar);
Ast=ntension*bars(a,2);

%calcs
d0=D-cover-astbar/2; %the distance from the extreme
compressive fibre to the ...
%...centroid of the most
tensile reinforcement
Vumax=0.2*fc*bv*d0

fcv=fc^(1/3);
if fcv>4
fcv=4;
end

b1=1.1*(1.6-d0/1000);
if b1<1.1
b1=1.1;
end
b3=1;
b2=1;
beta=b1*b2*b3;
Vuc=beta*bv*d0*fcv*(Ast/(bv*d0))^(1/3)

half_phi_vuc=0.5*phi*Vuc;
if Vstar<=half_phi_vuc
strresult='No shear reinforcement is required';
end

Page | 153
Vumin=Vuc+0.1*sqrt(fc)*bv*d0;
if Vumin<(Vuc+0.6*bv*d0)
Vumin=(Vuc+0.6*bv*d0);
end
Vumin
if Vstar>half_phi_vuc
if Vstar<=phi*Vumin
if Vstar<=phi*Vuc
strresult='Minimum shear reinforcement is
required but may be waived.';
else
strresult='Minimum shear reinforcement is
required.';
Asvmin=0.06*sqrt(fc)*bv*(0.5*D)/fsyf
end
end
end

if Vstar>phi*Vumin
nshear=5; %Quantity of N12 bars for shear
reinforcement
Asv=110*nshear; %N12 design area=110mm2

% Minimum Vus = Vusmin

Vusmin=(Vstar-phi*Vuc)/phi
if Vuc+Vusmin>=(Vumax)
strresult='Design fails in shear. Increase
thickness';
end
s_vusmin=Asv*fsyf*d0/(Vusmin*tand(45))
s=D/(nshear-1)

%Maximum Vus
Vusmax=Vumax-Vuc
s_vusmax=Asv*fsyf*d0/(Vusmax*tand(45))

formatSpec = 'Shear reinforcement is required.

Provide %2.0f-N12 bars at %2.0f mm spacings\n';
strresult=fprintf(formatSpec,nshear,s);
end
end

Page | 154
APPENDIX D – FINALSCRIPT.M

function [finalresults,finalresulta,nodes] =
finalscript(lowestfill,
highestfill,leg,crown,nomspan,nomheight)
%inputs: all inputs in metres
clc;
ht=lowestfill;
finalresults=[];
finalresulta=[];

while ht<=(highestfill+1E-10) %for

ht=lowestfill:0.1:highestfill
[Wdc Wfv Wcv Wlv Wfh Wah Wch Wlh Bc oheight crown
leg] = RCBC(ht,leg,crown,nomspan,nomheight);
load_comb;
[a b]=size(scomb);

%generate strand7 inputs for node creation

nodes=[0,0;0,oheight-crown/2;Bc-leg,oheight-
crown/2;Bc-leg,0];

%FIND WORST SYMMETRIC LOAD CASE

worstsymloadcomb=max(scomb,[],1); %This
would be the worst load comb

%confirm it exists:
exists=0;
for counter=1:a
result=scomb(counter,:)-worstsymloadcomb;
ind=find(result);
if isempty(ind);
exists=1;
worst=counter;
end
end

finalresults=[finalresults;ht,worst,(scomb(worst,:))]
;
ht=ht+0.1;

end %end while

%%%%%%%%PLOT SYMMETRIC LOAD COMBINATIONS

[a1 b1]=size(finalresults);
maxsym=0;
for counter7=1:a1

plot(finalresults(counter7,3:(int32(Bc*1000+3))));

Page | 155
 %str7=strcat(num2str((counter7-1)*0.1), ' m of
fill');
%gtext(str7)
if finalresults(counter7,3)>maxsym
maxsym=finalresults(counter7,3);
end
hold on
end
title('Symmetric load combinations - top of culvert')
xlabel('Width (mm)')
ylabel('Load (kPa)')
figure

for counter7=1:a1
plot(finalresults(counter7,int32(Bc*1000+4:b1)));
hold on
end
title('Symmetric load combinations - sides of
culvert');
xlabel('Height from top of culvert (mm)');
ylabel('Load (kPa)');

max2=max(finalresults(:,int32(Bc*1000+4):b1));
end

Page | 156
APPENDIX E – DEVLENGTH.M

%%%calculates development length of bar in tension

according to AS3600
%%%clause 13.1.2 for 50MPa concrete, fsy=500MPa
function
Lsyt=devlength(barqty,db,cover,widthmember,k1)
%INPUTS:
%barqty - quantity of bars
%db - diameter of bar
%cover - cover to reinforcement in mm
%widthmember - width over which quantity of bars will
be srpread
%k1=1.3 for a horizontal bar with more than 300mm of
concrete cast below it
%or k1=1 otherwise
fsy=500;
fc=50;
a=(widthmember-2*cover)/(barqty-1);
cd=min(a/2,cover);
k3=1-0.15*(cd-db)/db;
k2=(132-db)/100;
Lsyt=max(0.5*k1*k3*fsy*db/(k2*sqrt(fc)),29*k1*db);
end

Page | 157
APPENDIX F – TOP_RCBC4.M

%%%% A 99 LINE TOPOLOGY OPTIMIZATION CODE BY OLE

SIGMUND, JANUARY 2000 %%%
%%%% CODE MODIFIED FOR INCREASED SPEED, September
2002, BY OLE SIGMUND %%%

%Adapted by Juliana Sweeney - October 2014

function
[x,U]=top_rcbc4(span,legheight,volfrac,penal,rmin)
% INITIALIZE variables
%calculate nelx, nely
scale=25;
legt=350;
crown=400;
nelx=ceil((span/2+legt)/scale)
nely=ceil((legheight+crown)/scale)
x(1:nely,1:nelx) = volfrac;

%creating box culvert hole

passive(crown/scale+1:nely,1:nelx-legt/scale)=1;
x(crown/scale+1:nely,1:nelx-legt/scale)=0.001;

loop = 0;
change = 1.;
% START ITERATION
while change > 0.01
loop = loop + 1;
xold = x;
% FE-ANALYSIS
[U]=FE(nelx,nely,x,penal);
% OBJECTIVE FUNCTION AND SENSITIVITY ANALYSIS
[KE] = lk;
c = 0.;
for ely = 1:nely
for elx = 1:nelx
n1 = (nely+1)*(elx-1)+ely;
n2 = (nely+1)* elx +ely;
Ue = U([2*n1-1;2*n1; 2*n2-1;2*n2;
2*n2+1;2*n2+2; 2*n1+1;2*n1+2],1);
c = c + x(ely,elx)^penal*Ue'*KE*Ue;
dc(ely,elx) = -penal*x(ely,elx)^(penal-
1)*Ue'*KE*Ue;
end
end
% FILTERING OF SENSITIVITIES
[dc] = check(nelx,nely,rmin,x,dc);
% DESIGN UPDATE BY THE OPTIMALITY CRITERIA METHOD

Page | 158
 [x] = OC(nelx,nely,x,volfrac,dc,passive);
% PRINT RESULTS
change = max(max(abs(x-xold)));
aux1=clock;
disp([' It.: ' sprintf('%4i',loop) ' Obj.: '
sprintf('%10.4f',c) ...
' Vol.: '
sprintf('%6.3f',sum(sum(x))/(nelx*nely)) ...
' ch.: ' sprintf('%6.3f',change ) ...
' time: ' sprintf('%4i',aux1(4)) ' '
sprintf('%4i',aux1(5)) ' ' sprintf('%3.0f',aux1(6))])
% PLOT DENSITIES
colormap(gray); imagesc(-x); axis equal; axis tight;
axis off;pause(1e-6);
end

%%%%%%%%%% OPTIMALITY CRITERIA UPDATE

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
function [xnew]=OC(nelx,nely,x,volfrac,dc,passive)
l1 = 0; l2 = 100000; move = 0.2;
while (l2-l1 > 1e-4)
lmid = 0.5*(l2+l1);
xnew = max(0.001,max(x-
move,min(1.,min(x+move,x.*sqrt(-dc./lmid)))));
%added line for passive
xnew(find(passive))=0.001;

if sum(sum(xnew)) - volfrac*nelx*nely > 0;

l1 = lmid;
else
l2 = lmid;
end
end

%%%%%%%%%% MESH-INDEPENDENCY FILTER

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
function [dcn]=check(nelx,nely,rmin,x,dc)
dcn=zeros(nely,nelx);
for i = 1:nelx
for j = 1:nely
sum=0.0;
for k = max(i-
floor(rmin),1):min(i+floor(rmin),nelx)
for l = max(j-
floor(rmin),1):min(j+floor(rmin),nely)
fac = rmin-sqrt((i-k)^2+(j-l)^2);
sum = sum+max(0,fac);
dcn(j,i) = dcn(j,i) +
max(0,fac)*x(l,k)*dc(l,k);
end
end

Page | 159
 dcn(j,i) = dcn(j,i)/(x(j,i)*sum);
end
end

%%%%%%%%%% FE-ANALYSIS
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
function [U]=FE(nelx,nely,x,penal)
[KE] = lk;
K = sparse(2*(nelx+1)*(nely+1), 2*(nelx+1)*(nely+1));
F = sparse(2*(nely+1)*(nelx+1),1); U =
zeros(2*(nely+1)*(nelx+1),1);
for elx = 1:nelx
for ely = 1:nely
n1 = (nely+1)*(elx-1)+ely;
n2 = (nely+1)* elx +ely;
edof = [2*n1-1; 2*n1; 2*n2-1; 2*n2; 2*n2+1;
2*n2+2; 2*n1+1; 2*n1+2];
K(edof,edof) = K(edof,edof) +
x(ely,elx)^penal*KE;
end
end

%%%%%%%% DEFINE LOADS AND SUPPORTS

%%TOP LINE LOAD: this will be manually changed with

each simulation
lineload(1)=1;
for aux=1:nelx
lineload(aux+1)=lineload(aux)+nelx;
end
F(2*lineload,1) = -1575E3; %load in N/m

%%SIDE LINE LOAD: this will be manually changed with

each simulation
lineload2(1)=2*(nely+1)*(nelx)+1;
for aux=1:nely+1
lineload2(aux+1)=lineload2(aux)+2;
end
F(1*lineload2,1) = -833E3; %load in N/m

%%%%%%%% END DEFINE LOADS AND SUPPORTS

fixeddofs =
union([1:2:2*(nely+1)],[2*(nelx+1)*(nely+1)]);
alldofs = [1:2*(nely+1)*(nelx+1)];
freedofs = setdiff(alldofs,fixeddofs);
% SOLVING
U(freedofs,:) = K(freedofs,freedofs) \ F(freedofs,:);
U(fixeddofs,:)= 0;

Page | 160
%%%%%%%%%% ELEMENT STIFFNESS MATRIX
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
function [KE]=lk
E = 34800E6; %Young's Modulus in Pa
nu = 0.2;
k=[ 1/2-nu/6 1/8+nu/8 -1/4-nu/12 -1/8+3*nu/8 ...
-1/4+nu/12 -1/8-nu/8 nu/6 1/8-3*nu/8];
KE = E/(1-nu^2)*[ k(1) k(2) k(3) k(4) k(5) k(6) k(7)
k(8)
k(2) k(1) k(8) k(7) k(6) k(5) k(4)
k(3)
k(3) k(8) k(1) k(6) k(7) k(4) k(5)
k(2)
k(4) k(7) k(6) k(1) k(8) k(3) k(2)
k(5)
k(5) k(6) k(7) k(8) k(1) k(2) k(3)
k(4)
k(6) k(5) k(4) k(3) k(2) k(1) k(8)
k(7)
k(7) k(4) k(5) k(2) k(3) k(8) k(1)
k(6)
k(8) k(3) k(2) k(5) k(4) k(7) k(6)
k(1)];
%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%%%%%%%%%%%%%%%%%%%%
% This Matlab code was written by Ole Sigmund,
Department of Solid %
% Mechanics, Technical University of Denmark, DK-2800
Lyngby, Denmark. %
% Please sent your comments to the author:
[email protected] %
%
%
% The code is intended for educational purposes and
theoretical details %
% are discussed in the paper
%
% "A 99 line topology optimization code written in
Matlab" %
% by Ole Sigmund (2001), Structural and
Multidisciplinary Optimization, %
% Vol 21, pp. 120--127.
%
%
%
% The code as well as a postscript version of the
paper can be %
% downloaded from the web-site:
http://www.topopt.dtu.dk %

Page | 161
%
%
% Disclaimer:
%
% The author reserves all rights but does not
guaranty that the code is %
% free from errors. Furthermore, he shall not be
liable in any event %
% caused by the use of the program.
%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%%%%%%%%%%%%%%%%%%%%%%

Page | 162
APPENDIX G – LOAD_COMB.M

% GENERATING LOAD COMBINATIONS FOR RCBCs

%Bc= overall outside width of culvert in m - this

input variable is
%passed from main program

clc;
%Load factors:
vfactors=[ 1 1.4 0 1.8
1 1.4 1.5 0
1 0.9 0 1.8
1 0.9 1.5 0
1 0.9 0 0
1 1.4 0 0];

hfactors=[ 0.7 0.7 0 1.8

0.7 1.4 0 1.8
1.4 0.7 0 1.8
1.4 1.4 0 1.8
0.7 0.7 1.5 0
0.7 1.4 1.5 0
1.4 0.7 1.5 0
1.4 1.4 1.5 0
0.7 0.7 0 0
0.7 1.4 0 0
1.4 0.7 0 0
1.4 1.4 0 0];

%Horizontal Loads:
for counter5=1:length(hfactors)

h(counter5,:)=hfactors(counter5,1)*Wfh+hfactors(c
ounter5,2)*Wah+hfactors(counter5,3)*Wch+hfactors(
counter5,4)*Wlh;
end

%Vertical Loads:
for counter4=1:length(vfactors)

v(counter4,:)=vfactors(counter4,1)*Wdc+vfactors(c
ounter4,2)*Wfv+vfactors(counter4,3)*Wcv+vfactors(
counter4,4)*Wlv;
end

Page | 163
%FROM COL=1 TO COL=Bc*1000+1 IT'S VERTICAL, FROM
Bc*1000+2 TO length(scomb) IT'S HORIZONTAL
%Symmetric Combinations:

scomb=[ v(1,:),h(1,:); v(1,:),h(2,:);

v(1,:),h(3,:); v(1,:),h(4,:);
v(2,:),h(5,:); v(2,:),h(6,:);
v(2,:),h(7,:); v(2,:),h(8,:);
v(3,:),h(1,:); v(3,:),h(2,:);
v(3,:),h(3,:); v(3,:),h(4,:);
v(4,:),h(5,:); v(4,:),h(6,:);
v(4,:),h(7,:); v(4,:),h(8,:);
v(5,:),h(9,:); v(5,:),h(10,:);
v(5,:),h(11,:); v(5,:),h(12,:);
v(6,:),h(9,:); v(6,:),h(10,:);
v(6,:),h(11,:); v(6,:),h(12,:)];

%Asymmetric Combinations:
%one side:
acomb1=[v(1,:),h(9,:); v(1,:),h(10,:);
v(1,:),h(9,:); v(1,:),h(10,:);
v(2,:),h(9,:); v(2,:),h(10,:);
v(2,:),h(9,:); v(2,:),h(10,:);
v(3,:),h(9,:); v(3,:),h(10,:);
v(3,:),h(9,:); v(3,:),h(10,:);
v(4,:),h(9,:); v(4,:),h(10,:);
v(4,:),h(9,:); v(4,:),h(10,:);
v(5,:),h(9,:); v(5,:),h(10,:);
v(5,:),h(9,:); v(5,:),h(10,:);
v(6,:),h(9,:); v(6,:),h(10,:);
v(6,:),h(9,:); v(6,:),h(10,:);];
%opposite side:
acomb2=[v(1,:),h(11,:); v(1,:),h(11,:);
v(1,:),h(12,:); v(1,:),h(12,:)
v(2,:),h(11,:); v(2,:),h(11,:);
v(2,:),h(12,:); v(2,:),h(12,:);
v(3,:),h(11,:); v(3,:),h(11,:);
v(3,:),h(12,:); v(3,:),h(12,:);
v(4,:),h(11,:); v(4,:),h(11,:);
v(4,:),h(12,:); v(4,:),h(12,:);
v(5,:),h(11,:); v(5,:),h(11,:);
v(5,:),h(12,:); v(5,:),h(12,:);
v(6,:),h(11,:); v(6,:),h(11,:);
v(6,:),h(12,:); v(6,:),h(12,

Page | 164
APPENDIX H – 1815 RCBC DESIGN

The design process of an RCBC is iterative. The first structure size trialled had
a leg thickness of 200 mm and crown thickness of 250 mm. The results from
the load combination Matlab script finalscript.m (see appendix D for
code) were as shown in Figure 4-5 and Figure 4-10. The loads on the top of
the culvert are uniformly distributed over the entire top of culvert, as expected
since the truncated prism model (see Section 3.2 for details) distributes the
vehicle and construction loads uniformly over the top of the culvert. The other
loads that act on top of the culvert are also uniformly distributed, namely the
fill and the self-weight. There are 20 straight lines in Figure 4-5, each for a
different fill height. The top line is the load when there is only 0.1 m of fill over
the culvert, giving a load of 646.9 kPa for all values of x (width of culvert). The
bottom lines are very close together, which makes it difficult to distinguish one
from the other. However, the last line represents the load for the case when
there is 2.0 m of fill over the culvert, giving 107.3 kPa.

Page | 165
 Symmetric load combinations - top of culvert
700

600 Depth of fill: 0.1 m

Load: 646.9 kPa

500
Load (kPa)

400

300

200

100
0 500 1000 1500 2000 2500
Width (mm) Depth of fill: 2.0 m
Load: 103.5 kPa

Figure H-1 – Loads on top of 1815 RCBC

The worst load case was that for the least amount of fill, conservatively chosen
to be 0.1 m in this study (refer Section 3.2.3). The worst load combination was
the vertical load number 1 (see Figure 4-1) and horizontal load number 10 (see
Figure 4-2). This is expected since these combinations have the highest load
factors.

The horizontal loads have a different form to the vertical loads. They increase
from the top of culvert up to 0.5 m below the top of culvert, from which they
continue uniformly. That is because the only horizontal load which is uniformly
distributed is the horizontal live load WLH. The compaction load WAH, as
dictated in AS1597.2-2013, increases up to 0.5 m below the top of the culvert,
then remains constant up to 1.5 m below the top of the culvert, then decreases
linearly up to 2.0 m below the top of the culvert (see Section 3.2.9 for details).

Page | 166
The horizontal fill load WFH is also non-uniform since it varies with the depth
below the culvert. Because of these non-uniformities, the shape of the
horizontal load graphs differs from the vertical load graphs.

Symmetric load combinations - sides of culvert

350
X: 501
Y: 344.4

300 X: 1
Y: 316.4

250

200
Load (kPa)

150

100 X: 501
Y: 70.4
X: 1
Y: 42.4
50

0
0 200 400 600 800 1000 1200 1400 1600 1800
Height from top of culvert (mm)

Figure H-2 – Loads on the side of 1815 RCBC

The 1815 RCBC is then modelled in Strand7 and the worst load case (for 0.1
m fill) is applied. The bending moment diagram and shear force diagram are
then found as per Figure 4-11 and Figure 4-12.

Page | 167
Figure H-3 – Bending moment diagram for 1815 RCBC

Page | 168
 Figure H-4 – Shear force diagram for 1815 RCBC

The next step is to perform a flexibility and shear analysis to determine the
suitable reinforcement for the unit, using the Matlab scripts
flexanalysis.m and checkshear.m, as described in Section 3.3.5 - Non-
optimised Culverts.

For the middle of the crown, M*=359.2 kNm (Figure 4-11) and by running
flexanalysis.m with 16-N20 bars as tensile reinforcement and 5-N12 bars
as compression reinforcement, the results are:

Page | 169
dn = 35.4800
ku = 0.1731
esc = -4.6674e-04
Cc = 2.5333e+03
Cs = -51.3416
Mu = 479.4426
phiMu = 383.5541
Number of N12 bars required for compression: 5
Number of N20 bars required for tension: 16

This means that:

• the section is under-reinforced since ku<0.36;

• the compressive steel has not yielded since esc < 0.0025;
• φ M u > M * is suitable and so is the chosen reinforcement

The same procedure is utilised for the end of the crown, the top of the leg and
the bottom of the leg. The shear is checked as described in Section 3.3.5 using
the Matlab program checkshear.m. The development length of all bars is
checked using the Matlab function devlength.m (see Matlab code in
Appendix E) as per described in Section 3.3.6.

Flexure and shear analysis for end of the crown

The design moment at the end of the crown is M*=395.1 kNm (see Figure
4-11). By running flexanalysis.m for 24-N20 bars in tension and 11-N20
bars in compression, the results are:

dn = 49.5300
ku = 0.3195
esc = 2.7438e-04
Cc = 3.5364e+03
Cs = 187.1266
Mu = 507.4264
phiMu = 405.9412

Page | 170
Number of N20 bars required for compression: 11
Number of N20 bars required for tension: 24

This means that:

• the section is under-reinforced since ku<0.36;

• the compressive steel has not yielded since esc < 0.0025;
• φ M u > M * is suitable and so is the chosen reinforcement

According to AS3600-2009 clause 8.2.5 (a), if V * ≤ 0.5φVuc no shear

reinforcement is required except where the overall depth of the beam exceeds
750 mm, in which case minimum shear reinforcement shall be provided. Also
according to the same standard, clause 8.2.5 (ii), if V * ≤ φVuc the minimum
shear reinforcement requirements may be waived. Therefore, a shear check is
required to check if shear reinforcement is necessary. The design shear force at
the end of the crown is V*=1527.7 kN (see Figure H-4). By running
checkshear.m for the 18-N20 bars in tension, the results are:

>> checkshear(1527.7E3,20,18,250);
Vumax = 4920000
Vuc = 6.2490e+05
Vumin = 9.7280e+05
Vusmin = 1.5575e+06
s_vusmin = 32.5757
s = 31.2500
Shear reinforcement is required. Provide 9-N12 bars at
31 mm spacings

This means that V * > φVu.min and shear reinforcement is to be provided as per
AS3600-2009 clause 8.2.10. In this case, 9-N12 bars at 31 mm spacings would
have to be provided for an extent of D=250 mm. However, this is not possible
because the actual space between two N12 bars at 31 mm spacings is 19 mm
measured from the outside of the bars. That is smaller than most maximum

Page | 171
aggregate sizes in 50MPa concrete, which is 20 mm. By having bars too close
together, it may impede the passage of the aggregate causing flow problems
while casting. The minimum space between bars should not interfere with the
casting procedure, and therefore this culvert would need to have its sections
increased to better deal with the shear forces imposed on it.

Flexure and shear analysis for top of the leg

The design moment at the top of the leg is the same as the one at the end of
the crown, as expected, and is M*=395.1 kNm (see Figure 4-11). However, this
section is thinner since the leg is 220 mm and the crown is 250 mm. By running
flexanalysis.m for 24-N20 bars in tension and 22-N20 bars in
compression, the results are:

dn = 48.2700
ku = 0.3114
esc = 2.0323e-04
Cc = 3.4465e+03
Cs = 277.2082
Mu = 506.4705
phiMu = 405.1764
Number of N20 bars required for compression: 22
Number of N20 bars required for tension: 24

This means that:

• the section is under-reinforced since ku<0.36;

• the compressive steel has not yielded since esc < 0.0025;
• φ M u > M * is suitable and so is the chosen reinforcement

A shear check is required to check if shear reinforcement is necessary. The

design shear force at the top of the leg is V*=894.2 kN (see Figure H-4). By
running checkshear.m for the 22-N20 bars in tension, the results are:

Page | 172
>> checkshear(894.2E3,20,22,220);
Vumax = 4200000
Vuc = 6.1417e+05
Vumin = 9.1116e+05
Vusmin = 6.6325e+05
s_vusmin = 43.5354
s = 44
Shear reinforcement is required. Provide 6-N12 bars at
44 mm spacings

This means that V * > φVu.min and shear reinforcement is to be provided as per
AS3600-2009 clause 8.2.10. In this case, 6-N12 bars at 44 mm spacings would
have to be provided for an extent of D=220 mm.

Flexure and shear analysis for bottom of the leg

The design moment at the bottom third of the leg is M*=103.2 kNm (see
Figure 4-11). By running flexanalysis.m for 16-N12 bars in tension and 0
bars in compression, the results are:

dn = 12.3300
ku = 0.0775
esc = -0.0079
Cc = 880.3620
Cs = 0
Mu = 136.1784
phiMu = 108.9427
Number of N20 bars required for compression: 0
Number of N12 bars required for tension: 16

This means that:

• the section is under-reinforced since ku<0.36;

• φ M u > M * is suitable and so is the chosen reinforcement

Page | 173
A shear check is required to check if shear reinforcement is necessary. The
design shear force at the bottom of the leg is V*=413.2 kN (see Figure H-4).
By running checkshear.m for the 14-N20 bars in tension, the results are:

Vumax = 4200000
Vuc = 5.2828e+05
Vumin = 8.2526e+05
ans =Minimum shear reinforcement is required.

This means that 0.5φVuc < V * ≤ φVu.min and minimum shear reinforcement

Asv.min is to be provided as per AS3600-2009 clause 8.2.8.

However, when the topology optimisation Matlab program is applied to the

1815 RCBC with the following input parameters:

span in mm = 1800
leg height in mm = 1500
volume constraint = 0.25
penalisation factor = 3
filter size divided by the element size (rmin) = 1.5

it yields the result shown in Figure 5-1. It can be seen that the only areas with
white spaces (voids) are located in the bottom of the leg. That is the area where
there is less bending moment in the RCBC and that also requires less
reinforcement. The load on the crown is much larger than the load on the leg,
and that means that the crown will not have voids, since the material in the
crown is working hard to support the loads.

Page | 174
 Figure H-5 – Topology optimisation result for of 1815 RCBC

This simulation took 99 seconds and required 135 iterations. The optimised
culvert would look like shown in Figure H-6. It can be seen in detail 1 that there
is no space to insert bars in the region of the voids, since the cover to
reinforcement on either side of the bar needs to be 35 mm. Therefore, for
topology optimisation to be performed on the 1815 RCBC, the design domain
(crown thickness and leg thickness) will have to be increased to generate results
that can be achieved in practice.

Page | 175
Figure H-6 – Optimised 1815 RCBC

Page | 176