Académique Documents
Professionnel Documents
Culture Documents
by
A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy
University of Toronto
Doctor of Philosophy
University of Toronto
2010
Abstract
In this thesis, new design concepts for arch bridges using ultra high-performance fibre-reinforced concrete
are developed for spans of 50 to 400 m. These concepts are light-weight and efficient, and thus have the
potential to significantly reduce the cost of construction. Lightness is achieved by the thinning of structur-
al components and the efficient use of precompression in the arch, rather than by the decrease of bending
stiffness. Using the advanced properties of the material, the design concepts were shown to reduce the
consumption of concrete in arch bridges by more than 50% relative to arches built using conventional con-
crete technology. In addition to span length, other design parameters including span-to-rise ratio and
Other important contributions made in this thesis include: (1) the development of a simple analytical
model that describes the transition of shallow arches between pure arch behaviour and pure beam beha-
viour, (2) a comprehensive comparative study of 58 existing concrete arch bridges that characterizes the
current state-of-the-art and serves as a valuable reference design tool, and (3) the development and experi-
mental validation of general and simplified methods for calculating the capacity of slender ultra high-per-
formance fibre-reinforced concrete members under compression and bending. The research presented in
this thesis provides a means for designers to take full advantage of the high compressive and tensile
strengths of the concrete and hence to exploit the economic potential offered by the material.
ii
Acknowledgments
I would like to thank, first and foremost, Professor D. P. Gauvreau, for all the guidance and encouragement
he has given me as a supervisor, teacher, and mentor over the past five years. His passion for bridges, with
respect to their design, construction, and aesthetics has certainly been contagious and will forever be an
inspiration to me.
I would also like to thank Professor F. J. Vecchio, Professor P. C. Birkemoe, Professor S. A. Sheikh, and Pro-
fessor V. Sigrist for serving on my Ph.D. Examination Committee and for reviewing this thesis.
Financial assistance during the majority of my graduate studies was provided in large part by the Natural
Sciences and Engineering Research Council of Canada, and the University of Toronto. For this, I am
sincerely grateful.
Many others have given their time selflessly to help me along the way. Experimental work would not have
been possible without the help of my colleagues Ivan Wu, Jimmy Susetyo, Kris Mermigas, Brent Visscher,
Davis Doan, Lulu Shen, Nabil Mansour, Boyan Mihailov, James Liu, and Serguei Bagrianski, and laborat-
ory staff Joel Babbin, Renzo Bassett, Giovanni Buzzeo, and John MacDonald. I am also indebted to others
who have accompanied and helped me along the way, including: Billy Cheung, Negar Elhami Khorasani,
Jeff Erochko, Eileen Li, Jamie McIntyre, Michael Montgomery, Talayeh Noshiravani, Sandy Poon, Carlene
Ramsay, Nick Zwerling, and especially Lydell Wiebe. A special thanks is dedicated to Catherine Chen who
assisted me over two summers in compiling the database of concrete arch bridges.
Last, I thank my wife Sarah, my parents, Danilo and Elizabeth, and my siblings, Michael, Elaine, and
iii
Table of Contents
Abstract, ii
Acknowledgments, iii
Table of Contents, iv
List of Figures, x
Chapter 1. Introduction, 1
1.1 Motivation, 1
1.2 State-of-the-art, 5
3.2.1 Methodology, 37
v
Chapter 4. Statical Analysis of Arch Bridges, 82
vi
Chapter 5. Comparative Study of 58 Concrete Arch Bridges, 153
vii
Chapter 6. New Concepts for Arch Bridges, 197
References, 288
Appendices, 294
ix
List of Figures
Figure 1-1. Flow of forces and sectional forces in arches and beams, 2
Figure 1-2. Construction of Sandö Bridge (top) and Colorado River Bridge (bottom), 3
Figure 1-3. Cross-section of ultra high-performance fibre-reinforced concrete girder used in Virginia Bridge, 6
Figure 1-6. Elevation, cross-section, and joint detail views of WILD Bridge concept, 9
Figure 2-2. Load versus displacement data of cylinders from Set 1, tested 27 days after casting, 19
Figure 2-3. Load versus displacement data of cylinder from Set 2, tested 60 days after casting, 19
Figure 2-7. Tangent modulus of elasticity versus compressive strength of tested cylinders, 25
Figure 2-11. Creep strain of an untreated ultra high-performance fibre-reinforced concrete specimen over time, 31
Figure 3-2. Load deflection and sectional response of a slender column given different analytical assumptions, 35
x
Figure 3-4. Numerical integration procedure, 43
Figure 3-5. Column deflection curves and sectional responses of mid-length sections for various column lengths, 47
Figure 3-6. Member capacity interaction diagrams for various slenderness ratios, 52
Figure 3-10. Clamped steel end plate (left), spherical bearing at base (right), 58
Figure 3-12. Position of cohesive laser radar scanner targets at zero load, 59
Figure 3-15. Load-deflection response of specimens and photographs at each load stage, 63
Figure 3-17. Comparison of load test results and general method predictions, 67
Figure 3-19. Conventional strain limits and Menn’s reduced strain limits, 71
Figure 3-20. Interaction and moment-curvature diagrams for a 30 MPa concrete box, 72
xi
Figure 4-3. Arch and analogous column buckling models, 87
Figure 4-4. Comparison of effective arc length factors with stability analysis results obtained by Rambøll (1944), 88
Figure 4-5. Pressure lines for uniformly distributed and concentrated loads, 90
Figure 4-6. Bending moments caused by concentrated loads in fixed, two-hinged, and three-hinged arches, 93
Figure 4-7. Bending moments caused by distributed loads in fixed, two-hinged, and three-hinged arches, 94
Figure 4-8. Bending moment influence lines for fixed, two-hinged, and three-hinged arches, 95
Figure 4-9. Comparison of bending moment envelopes and variations in depth along the arch, 96
Figure 4-10. Bending moments caused by restrained deformations in fixed, two-hinged, and three-hinged arches, 97
Figure 4-11. Moment distribution factors for the Colorado River Bridge, 100
Figure 4-13. Classification of fixed arch systems, and their respective simplified and analogous structural models, 104
Figure 4-14. Stiffened arch model and moments and axial forces caused by unit redundant forces, 105
Figure 4-15. Effect of arch, deck, and column flexural rigidities on the distribution of bending moment, 106
Figure 4-17. Transition between fixed arch and fixed beam, 116
Figure 4-18. Square of slope of parabolic arches at their springing lines, 118
Figure 4-22. Flat hydraulic jacks at crown of Rio Parana Bridge, 132
Figure 4-23. Free body diagrams of cantilevered arch during crown jacking, 132
Figure 4-24. Influence of deck-stiffening on flexural stresses in the arch and deck, 145
xii
Figure 4-26. Billington’s graph of arch stress versus stiffness ratio between arch and deck, 147
Figure 5-5. Year of completion and method of construction versus span length, 167
Figure 5-9. Colorado River Bridge construction using tower and stay method, 169
Figure 5-10. Krk Bridge construction using trussed cantilever method, 170
Figure 5-11. Argentobel Bridge construction using the cantilever rotation method, 171
Figure 5-12. Section type and structural depths of arches and decks versus span length, 172
Figure 5-13. Structural depth of arch and deck versus span length, 173
Figure 5-15. Moment of inertia of arch, deck, and system versus span length, 175
Figure 5-17. Ratio of arch inertia to system inertia versus span length, 178
Figure 5-18. Ratio of average column inertia to system inertia versus span length, 179
Figure 5-20. Ground and arch profiles of selected bridges in the database, 181
xiii
Figure 5-21. Variations of span-to-rise ratio based on the Colorado River Bridge, 182
Figure 5-24. Efficiency threshold curve based on average system radius of gyration trend line, 185
Figure 5-25. Efficiency threshold curve based on subjective high estimate of system radius of gyration, 187
Figure 5-26. Equivalent slab thickness by volume versus span length, 188
Figure 5-27. Equivalent slab thickness, span length, and year of construction, 189
Figure 5-29. Eccentricity of resultant compressive force caused by live load bending moments, 191
Figure 5-30. Normalized live load eccentricity versus span length, 193
Figure 6-12. Interaction diagrams of all partially deck-stiffened arch concepts, 225
Figure 6-13. Sections of selected design concepts and existing arch bridges, 226
xiv
Figure 6-14. General arrangement drawing, 230
Figure 6-16. Proposed match-casting method for abutting end ribs, 233
Figure 6-19. Comparison of simplified and general method interaction diagrams, 242
Figure 6-20. Comparison of equivalent slab thicknesses of new arch concepts and existing arches in database, 245
Figure 6-22. Comparison of normalized system moment of inertia and system radius of gyration, 249
Figure 6-27. Comparison of first-order live load portions of governing moments, 256
Figure A-2. Arch database: geometry, structural types, and ratios, 297
xv
Figure C-7. Instrumentation, 313
xvi
List of Tables
Table 2-1. Mix design for ultra high-performance fibre-reinforced concrete, 17
Table 2-3. Compressive strengths of cylinders organized by batch, age, and curing regime, 22
Table 4-1. Relative comparison between fixed, two-hinged, and three-hinged arches, 86
Table 4-2. Expressions for axial strains and flexural rigidities in arches, 97
Table 4-3. Bending moment diagrams for fixed arch systems, 109
Table 5-1. Concrete arch bridge data compiled from reference databases, 155
Table 5-3. Location and global geometry of arch bridges in database, 161
Table 6-5. Sequence of trial dimensions used in the design of hollow box sections, 207
Table 6-6. Summary of quantities to be designed for parametric design study, 212
Table 6-7. Maximum plate widths for thin ultra high performance fibre-reinforced concrete plates, 215
xvii
Table 6-8. Proportions of design concepts, 219
Table 6-9. Geometric ratios and structural demands of design concepts, 220
Table 6-15. Equivalent slab thicknesses based on volume of longitudinal concrete among all concepts, 244
Table 6-16. Quantity of longitudinal prestressing strands in deck girder among all concepts, 245
Table 6-17. Relative volume of arch and deck among all concepts, 246
Table 6-20. Normalized system moment of inertia among all concepts, 248
Table 6-24. Maximum deflections at ultimate limit states among all concepts, 252
Table 6-26. Governing live load cases among all concepts, 254
Table 6-27. Maximum flexible system moments at ultimate limit states among all concepts, 254
Table 6-28. Portions of governing moments caused by first-order live loads among all concepts, 256
Table 6-29. Most common geometrical quantities used among design concepts, 257
xviii
List of Symbols
§ reference to clause
!V extra weight factor, equal to total weight of system divided by weight of continuous longitudinal concrete
$0 first-order deflection
$ij flexibility coefficients, or deformations of primary system caused by unit redundant forces
$M bending deformation
$N axial deformation
$V shear deformation
xix
ΔL change in span length
ΔT change in temperature
Δf crown deflection
%c compressive strain
%t tensile strain
κ system stiffness factor, or ratio between system secant bending stiffness and system flexural rigidity
( slenderness ratio
xx
(f modified slenderness ratio
) buckling coefficient
* Poisson’s ratio
+ aging coefficient
E modulus of elasticity
EA axial rigidity
EI flexural rigidity
EI! secant bending stiffness (secant to the moment-curvature diagram calculated for a given point)
xxii
EIarch flexural rigidity of arch
EIcrown flexural rigidity of arch crown; prime symbol indicates secant stiffness
EIdeck flexural rigidity of deck girder; prime symbol indicates secant stiffness
EIsys flexural rigidity of system (sum of EIdeck and EIcrown); prime symbol indicates secant stiffness
f rise of arch
fc compressive stress
ft tensile stress
xxiii
hV equivalent slab thickness, or normalized concrete volume of system
kf shallowness factor
xxiv
(kL)arch effective length of arch leg between spandrel columns
l length of element
mN=0 normalized bending moment resistance in pure flexure (axial force is zero)
M* ultimate moment, or bending moment carried by the critical section of a column at ultimate load Q*
Mfalse bending moment if structural system was built on falsework and then released
xxv
Mlane bending moment caused by multilane lane load
n integer value; number of design traffic lanes; normalized axial force; virtual axial force
nE; nE linearity modular ratio between Ec and Ec!; normalized axial force at Euler buckling
N axial force
xxvi
N-M sectional capacity interaction diagram
q distributed load
Q concentrated load
Q* ultimate load, or maximum axial load of column of given length, cross-section, material, and eccentricity
r radius of gyration
R2 coefficient of determination
t elapsed time (i.e. after casting, after initial stress); wall thickness
V total longitudinal concrete volume of system, including the deck girder, spandrel columns, and arch
x longitudinal axis
xxvii
x1, x2 redundant end moments of fixed arch system
z lever arm between flexural tensile force and flexural compressive force
xxviii
List of Uncommon Terms
arch-beam parameter—a non-dimensional parameter that situates a given fixed arch system within a transition
balanced point—in the context of ultra high-performance fibre-reinforced concrete, the sectional failure mode
associated with the simultaneous failure of concrete in compression and tension at the extreme fibres
deck-stiffened arch system—a structural system comprised of an arch, spandrel columns, and a deck girder, in which
the flexural rigidity of the arch is much less than the flexural rigidity of the deck
eccentricity of load—the distance between the centroidal axis of a column and line of action of applied load
eccentricity constant—the ratio of the maximum deflection of a member to the maximum eccentricity of load
effective secant stiffness—conservative values of secant bending stiffness selected according to the highest bending
effective arc length—similar to the concept of effective length in slender columns, this term is equal to the arc length
of the arch times a factor that accounts for the boundary conditions and buckled shape of the arch
equivalent slab thickness—a normalized volume quantity, equal to the total volume of longitudinal concrete divided
extra weight factor—the ratio of the total weight of the system to the weight of longitudinal concrete
fixed system moment—the moments that arise from local deformations of the deck and arch between spandrel
columns, assuming that the joints at each end of the columns do not translate
flexible system moment—the moments that arise from the global deformations of the deck and arch between
springing lines; these moments are typically much greater than fixed system moments
member capacity interaction diagram—an envelope of maximum sectional forces carried by the critical section of a
slender member of given length, cross-section, and material model; this envelope is a collection of
ultimate limit states of the member, corresponding to all possible values of initial eccentricity of load
xxix
modified slenderness ratio—a non-dimensional ratio related to the arch-beam parameter that situates a given fixed
arch system within a transition between pure arch behaviour and pure beam behaviour; this ratio also
partially deck-stiffened arch system—a structural system comprised of an arch, spandrel columns, and a deck girder, in
which the flexural rigidity of the arch is roughly the same as the flexural rigidity of the deck
secant bending stiffness—a secant to the moment-curvature diagram calculated for a given point; this quantity is often
sectional capacity interaction diagram—an envelope of maximum sectional forces carried by a given section and
material model; this envelope corresponds to the sectional failure (i.e. concrete crushing or fibre pull-out)
self-stiffened arch system—a structural system comprised of an arch, spandrel columns, and a deck girder, in which
the flexural rigidity of the arch is much greater than the flexural rigidity of the deck
shallow arch—an arch that is sensitive to axial deformations; in indeterminate arches, these axial deformations induce
system flexural rigidity—the sum of flexural rigidities of the arch and deck
system moment of inertia—the sum of moments of inertia of the arch and deck
system radius of gyration—the square root of (system moment of inertia divided by cross-sectional area of arch
crown)
system stiffness factor—a reduction factor less than one that accounts for material nonlinearity, equal to effective
ultimate deflection—in the context of slender members, the deflection at ultimate load
ultimate load—in the context of slender members, the maximum compressive axial load that can be sustained by a
ultimate moment—in the context of slender members, the bending moment at the critical section of a column at
ultimate load
xxx
“The form controls the forces; and the more clearly
the designer can visualize those forces the surer he is of his form”
—D. P. Billington
xxxi
Chapter 1. Introduction
This thesis concerns the development and validation of new structural systems for arch bridges using ultra
high-performance fibre-reinforced concrete. To design structures effectively using new materials, it is im-
portant that designers understand how the material affects structural behaviour at a variety of scales, in-
cluding that of the material, cross-section, member, and overall arch system. Correspondingly, this re-
search study consists of five parts: (1) an experimental investigation of the stress-strain response of ultra
members under compression and bending, (3) a review and synthesis of the statical analysis of arch
bridges, (4) an empirical study of fifty-eight concrete arch bridges built over the last century, and (5) a
parametric design study of arch bridge concepts using ultra high-performance fibre-reinforced concrete.
1.1 Motivation
Arches are intrinsically efficient because they carry a large share of their applied loads in compression. The
geometric form of an arch is typically chosen such that the centroidal axis of the arch coincides with the
line of action of force, or pressure line, of a given set of permanent loads (see Figure 1-1). Shaping arches in
this way allows them to carry permanent loads in a state of pure compression. Because concrete is strong
in compression and weak in tension, carrying permanent loads in compression is considered to be a more
economical use of material than carrying permanent loads in bending. To carry tensile stresses that arise
due to bending, concrete members must rely on longitudinal steel reinforcement to maintain equilibrium.
1
1. Introduction
Thus the required amount of steel and concrete is generally lower in arches than in beams for permanent
loads.
Figure 1-1. Flow of forces and sectional forces in arches and beams
Despite their material efficiency, arches have seldom been built in the past fifty years due to the high cost
of labour and temporary structures. In general, construction costs increase with increases in complexity
and duration of construction and with increases in weight of temporary and permanent structures. Up un-
til the mid-20th century, concrete arches were usually built using elaborate timber scaffolding. For ex-
ample, the Sandö Bridge in Sweden was built this way in 1943, as shown in Figure 1-2. This method of con-
struction is time-consuming, generates a large amount of throwaway material, and requires the work of
highly skilled carpenters. In the second half of the twentieth century, modern cantilevered methods of
construction led to significant cost savings and greater span lengths for arches. The Colorado River Bridge
on the border of Arizona and Nevada was built this way in 2009, as shown in Figure 1-2. These cantilever
methods significantly reduced construction costs by reducing the amount of falsework and throwaway ma-
terial and by reducing the amount of manual labour through increased mechanization of work. The use of
precast concrete segments has allowed for further improvements in the speed and quality of construction.
Parallel advances in the construction technology of competing structural systems, such as variable-depth
girder and cable-stayed systems, were also made. The construction of cable-stayed bridges requires essen-
tially zero throwaway materials, which remains a lot less than arch bridges built using cantilever methods.
2
1. Introduction
Thus, despite innovations in arch construction techniques, concrete arches have remained, in most cases,
Figure 1-2. Construction of Sandö Bridge (top) and Colorado River Bridge (bottom).
Adapted from Mondorf (2006) and usdot-fhwa (2003). Illustrations are not shown at the same scale.
Advances in material technology have produced concretes that have compressive strengths in excess of 150
MPa. This represents a significant increase in strength relative to the conventional 35 MPa concretes that
were common twenty years ago. In addition to exceptional strength, this new class of concrete called ultra
high-performance fibre-reinforced concrete also tends to exhibit superior durability, long-term stability
(i.e. reduced creep and shrinkage strains), stiffness, and tensile strength when compared to conventional
concrete (AFGC 2002). These enhanced properties are achieved through the use of steel fibres, water redu-
cers, a high binder content, and a special selection of fine aggregates. Steel fibres are used to improve the
ductility of concrete, to provide crack control, and to carry tensile forces, potentially reducing the need for
conventional steel reinforcing. In mixes with high fibre content, tensile strengths up to 20 MPa have been
Using ultra high-performance fibre-reinforced concrete in arch bridges may lead to significant reductions
in concrete consumption and construction cost. The benefits of higher compressive strength are compoun-
3
1. Introduction
ded in terms of material efficiency: increases in member capacity allow for thinner members, thinner
members are lighter and so decrease dead load demand. Because arches carry loads primarily in compres-
sion, these benefits are applicable and should reduce concrete consumption.
In conventionally reinforced concrete members, minimum slab and flange thicknesses are often governed
by minimum concrete cover and spacing requirements for mild and prestressing steel. Minimum cover
depths to reinforcing steel ensure a minimum level of durability of the structure, including protection
against the corrosion of reinforcing steel. Where mild or internal prestressing steel is used in conjunction
with ultra high-performance fibre-reinforced concrete, minimum cover requirements can be reduced due
to the low permeability of the concrete. Where steel fibres are designed to carry tensile forces in lieu of
mild or internal prestressing steel, conventional cover requirements are less applicable because steel fibres
are not spatially continuous. The spatial discontinuity of the fibres prevents the flow of chloride ions in a
closed circuit, effectively limiting the the undesirable effects of corrosion. As cover requirements are re-
duced or eliminated through the use of ultra high-performance fibre-reinforced concrete, greater thinness
of components can be achieved. Thinness of components are instead governed by local stability problems,
by space requirements related to obtaining random fibre-orientation during concrete placement, and by
member strength requirements. Based on the amount of additional compressive and tensile strength
afforded by ultra high-performance fibre-reinforced concrete over conventional concrete, this research
If the weight of concrete in arches can be reduced by half, the cost of labour, equipment, and temporary
structures will also be reduced. Because lighter structures are more easily lifted and manoeuvred, load ca-
pacities of cranes and temporary structures can be reduced. Further, construction should proceed more
quickly and total hours of labour reduced. The benefits of tensile strength provided by steel fibres can be
used to reduce the quantity of passive reinforcing, reducing the labour associated with assembling rebar
cages. All these factors should offset the higher unit cost of ultra high-performance fibre-reinforced con-
crete relative to conventional concrete. If these cost savings equal or exceed the economic differences
4
1. Introduction
between arch bridges and other structural systems, then concrete arches will be renewed as viable and eco-
arches may create unique opportunities in aesthetic expression, leading to greater achievements in visual
slenderness, shallowness, and transparency than were previously possible. Reductions in concrete con-
sumption also have benefits in terms of environmental impact, such as reduced use of natural resources
and reduced carbon emissions in the manufacturing and transporting of concrete. Improvements in dur-
ability would likely increase the service life of the bridge and decrease the cost and frequency of
maintenance.
1.2 State-of-the-art
In this section, the state-of-the-art of bridges using ultra high-performance fibre-reinforced concrete is re-
viewed. The currently under-researched aspects of this body of knowledge, which are addressed by this
Real-world progress has been made over the past decade in the development and implementation of new
structural systems for vehicular bridges using ultra high-performance fibre-reinforced concrete. In her
doctoral thesis, Spasović (2008) reviews and compares five short-span highway bridges built between 2001
and 2006 which utilize girders made from ultra high-performance fibre-reinforced concrete. These bridges
were built in Virginia, USA, in Shepherds Creek, Australia, in Bourg-lès-Valence, France, in Saint-Pierre-
la-Cour, France, and over Highway A51 in France. The cross-section of the precast, π-shaped girders used
in the Virginia road bridge is shown in Figure 1-3 (Graybeal and Hartmann 2005). The girders span 21.3
metres and are simply supported. At their thinnest, the girders have slab thicknesses of 76 mm and web
thicknesses of 58 mm.
5
1. Introduction
Figure 1-3. Cross-section of ultra high-performance fibre-reinforced concrete girder used in Virginia Bridge.
Adapted from Graybeal and Hartmann (2005). Dimensions are in mm.
Spasović (2008) concluded that the ultra high-performance fibre-reinforced concrete girders among the
five bridges reviewed were all designed to function in a conceptually similar way: prestressing strands were
used to carry flexural tensile forces and ultra high performance fibre-reinforced concrete was used to carry
flexural compressive forces. Spasović observed, however, that the high compressive capacities of the con-
cretes were not being used to their fullest potential. In general, these girders were precast and assembled
in-situ using dry or wet joints and prestressing. In some of the five bridges, transverse moments were car-
ried by ultra high-performance fibre-reinforced concrete alone, without conventional reinforcing. Conven-
tional stirrups for shear reinforcement were also omitted in the designs due to the capacity of the concrete
and fibres to carry shear stresses. According to the designers, these bridges were economically competitive
in comparison to conventional concrete girder systems. These project studies demonstrate that structural
members with thin slabs, webs, and walls are capable of transmitting localized forces, such as wheel loads,
to primary longitudinal structural elements of bridges. They also set precedence for the successful applica-
6
1. Introduction
Figure 1-4. Sun-Yu Pedestrian Bridge. Adapted from Huh and Byun (2005). Dimensions are in mm.
The Sun-Yu Bridge in Korea, built in 2002, is the first and longest pedestrian bridge to use ultra high-per-
formance fibre-reinforced concrete in an arch form (Huh and Byun 2005). The arch, which spans 120 m
and rises 15 m, consists of a single π-shaped arch rib, as shown in Figure 1-4. The arch rib is prestressed
transversely with 13 mm monostrands at each transverse stiffening rib spaced at 1225 mm along the length
of the arch. Longitudinally, the rib is prestressed with two tendons of 9-15 mm strands in the lower webs,
and one tendon of 12-15 mm strands in the upper web. The specified compressive strength of the reactive
Instead of conventional spiral reinforcement details, special reinforcement-free micro-anchorages (see Fig-
ure 1-5) were used to confine the concrete and to control the high tensile splitting forces near the an-
chorage zone. Using these micro-anchorages remove the slenderness limits normally imposed by the cover
requirements of spiral reinforcement, and allow for unprecedented thinness in the slabs, webs, and ribs of
structural members. These micro-anchorages were first developed for the Sherbrooke Bridge in Quebec,
7
1. Introduction
which is a trussed pedestrian bridge that has struts made from ultra high-performance fibre-reinforced
concrete (Aitcin et al. 1998). The Sun-Yu and Sherbrooke Bridges demonstrate that thin, ultra high-per-
formance fibre-reinforced concrete slabs, webs, and ribs can be prestressed and detailed without mild
reinforcing.
Figure 1-5. Reinforcement-free micro-anchorage detail. Adapted from Aitcin et al. (1998).
To date, there has been no real-world application of ultra high-performance fibre-reinforced concrete in
highway arch bridges. Conceptual design studies, however, have been done by Freytag et al. (2009), Can-
drlic et al. (2001), and Radić et al. (2005) for arch spans of 69 m, 432 m, and 1000 m, respectively.
Freytag et al. (2009) investigate a conceptual design for the wild Bridge, which is designed for a 69 m
crossing in Austria using 165/185 MPa ultra high-performance fibre-reinforced concrete. The design,
shown in Figure 1-6, proposes two, hollow concrete ribs for the arch. These ribs are slender and thin, with
structural depths of 1200 mm (corresponding to a span-to-arch-depth ratio of 58:1) and wall thicknesses of
60 mm. Each leg in the polygonal arch is built as individual precast segments, which are later joined by
special knee-joints that form the angle breaks of the polygon. The knee-joints share the same exterior di-
mensions as the arch segments, but are made inwardly thicker to make space for embedded deviators for
prestressing tendons. The prestressing in the arch ribs are external and unbonded within the arch seg-
8
1. Introduction
ments, and are distributed along the inside perimeter of the section to obtain concentric prestress. The
proposed bridge deck consists of a conventional 600 mm thick slab made from 40/50 MPa concrete.
Figure 1-6. Elevation, cross-section, and joint detail views of WILD Bridge concept.
Scale is shown at 80% of original. Adapted from Freytag et al. (2009).
Freytag et al. (2009) conducted compression and bending tests on a prototype prestressed arch segment
and knee-joint assembly, which was built at full scale. The specimens were subjected to the forces expected
at all critical design states, and were found to be adequate. Results were also compared to finite element
Candrlic et al. (2001) propose a conceptual design for a 432 m crossing over the Straits of Bakar in Croatia,
between Rijeka and Senj. The study is intended as a probe into the possibilities offered by using ultra high-
performance fibre-reinforced concrete with compressive strength of 200 MPa in the design an arch bridge.
The elevation and sections of the design are shown in Figure 1-7. Both the deck and arch are proposed as
three-cell box girders, with top and bottom slabs of 120 mm and web thicknesses of 200 mm. These mem-
bers are given oblong, aerodynamic shapes to reduce drag forces caused by transverse wind loading. The
arch rib is 6500 mm deep, corresponding to a span-to-arch-depth ratio of 66:1. Precast, segmental canti-
9
1. Introduction
lever construction is proposed, whereby match-cast segments, each 3800 mm in length, are joined in-situ
using epoxy resin and temporary external prestressing. Transverse ribs along the top and bottom slabs of
the box girders are provided at the ends of each segment. These ribs are used as a form of anchorage block
for temporary and permanent longitudinal prestressing. Using linear-elastic analysis, Candrlic et al. (2001)
calculated stresses caused by dead load, live load, wind, earthquake, and temperature, and found that con-
Figure 1-7. Bakar Bridge concept. Adapted from Candrlic et al. (2001). Dimensions are in mm.
10
1. Introduction
Radić et al. (2005) propose a conceptual design for a fictitious 1000 m crossing. Elevation and cross-sec-
tion views of their design are shown in Figure 1-8. The arch is a three-cell box girder and is 16.5 m deep,
corresponding to a span-to-arch-depth ratio of 61:1. The arch rib is composed of 5000 mm long segments,
each weighing 5.6 MN. A truss cantilevered method of construction is proposed, whereby the arch forms
the bottom chord, spandrel columns form the vertical struts, and steel ties form the diagonal ties and top
chord. A statical analysis was performed for all critical stages of construction. Maximum compressive and
tensile stresses were found to be 61 MPa and 14 MPa, respectively. Beyond this, very few details of the ana-
Figure 1-8. Bridge concept for a 1000 m crossing. Adapted from Radić et al. (2005). Dimensions are in mm.
The three arch bridge concepts described above begin to reveal the possibilities offered by using ultra high-
performance fibre-reinforced concrete in arch bridges. For all concepts, most bridge components are pro-
posed as modular precast segments with thin slabs and webs. Embedded steel fibres are used to displace
11
1. Introduction
most, if not all, of the conventional mild steel reinforcement. Prestressing steel is used temporarily to con-
nect segments during construction and also used as permanent reinforcement to provide flexural tensile
strength. All proposed segments incorporate some form of transverse stiffening through the local thicken-
ing of the section or the use of transverse ribs. Only Freytag et al. (2009) provide any experimental valida-
Although these concept studies put forth are new and innovative ideas, they are narrow in scope and are
only directly applicable to the spans and rises specified in their designs. The calculated stresses and design
checks given in the studies are not helpful to designers who wish to design their own arch systems. Instead,
general guidelines and simplified methods of analysis are needed to aid designers. This would include a
thorough exposition on member behaviour and system behaviour, and how they differ from convention
reinforced concrete structural systems. This would also include simplified design methods for calculating
maximum sectional forces in the system, and recommendations on the proportion and size of members.
Structural efficiency needs to be quantified and compared with existing concrete arch technology to furth-
er substantiate the material and cost savings offered by these new systems. This thesis undertakes these
tasks in hope of educating designers and bringing these innovative structural systems one step closer to
real-world application.
1. to design and validate new structural systems for arch bridges that reduce concrete consumption by
at least 50% relative to arch bridges built using conventional reinforced concrete technology, and
2. to provide designers with guidance on the structural analysis and design of these structural systems.
The secondary objectives of this thesis, which support the primary objectives are:
3. to characterize a local ultra high-performance fibre-reinforced concrete mix and to propose material
4. to identify how differences between ultra high-performance fibre-reinforced concrete and conven-
6. to review, synthesize, and illustrate the statical analysis of concrete arch bridges,
7. to identify thresholds for efficient arch behaviour related to the shallowness of concrete arch bridges,
8. to characterize the state-of-the-art of arch bridges made from conventional concrete, and
9. to determine which load combinations cause maximum sectional forces in these arch systems.
The content of this thesis reflects the primary and secondary objectives stated above. Topics are discussed
in sequence of scale of structural behaviour, starting with behaviour of the material, then of slender mem-
bers, and last of arch systems. In Chapter 2, the material behaviour of a local mix design of ultra high-per-
formance fibre-reinforced concrete is investigated experimentally with respect to uniaxial compression, in-
direct tension, creep, and shrinkage. From these tests, simple materials models are proposed for use in
analysis and design. In Chapter 3, the behaviour of slender members subjected to compression and bend-
ing is investigated. A general analysis method, which accounts for material and geometric nonlinearities, is
described and is implemented as a computer program that calculates load-deflection response of eccent-
rically-loaded columns in compression. This method is validated experimentally with load tests of slender
columns made from ultra high-performance fibre-reinforced concrete. Simplified design methods for cal-
culating member capacity are also proposed. In Chapter 4, the behaviour of arch structural systems is re-
viewed, synthesized, and illustrated. Using the force method, simplified methods for calculating sectional
forces caused by dead loads, live loads, and volumetric changes are presented and solved. The effects of
arch slenderness (sensitivity to flexural buckling) and arch shallowness (sensitivity to axial deformations)
are discussed in detail. In Chapter 5, a comparative study of fifty-eight concrete arch bridges built over the
past century is presented. Trends of various geometrical ratios and quantities among bridges in the data-
base are identified and discussed. These trends represent the current state-of-the-art of concrete arch
13
1. Introduction
bridges. In Chapter 6, a parametric design study of arch bridges using ultra high-performance fibre-rein-
forced concrete is described. Span length, arch rise, and degree of deck-stiffening are the primary design
parameters considered. Dimensions and sections of satisfactory design solutions are presented. Insights re-
lated to the efficiency, consumption of materials, slenderness, and shallowness are discussed. Design con-
cepts are also compared with the database of existing concrete arch bridges. In the final chapter, the most
important results and conclusions from each of the preceding chapters are summarized. Potential research
• the validation of new light-weight and efficient arch bridge design concepts using ultra high-per-
• the development of a simple analytical model that describes the transition of shallow arches between
• a comprehensive comparative study of 58 existing concrete arch bridges that characterizes the current
• the development and experimental validation of a general and simplified method for calculating the
and bending.
The research presented in this thesis provides a means for designers to take full advantage of the high com-
pressive and tensile strengths of the concrete and hence to exploit the economic potential offered by the
material.
14
Chapter 2. Ultra High-Performance Fibre-Reinforced
Concrete
The purpose of this chapter is to describe the production of ultra high-performance fibre-reinforced con-
crete and to develop workable stress-strain models based on experiments conducted at the University of
Toronto. These stress-strain models will be used in Chapter 3 to investigate the behaviour of slender mem-
bers subjected to compression and bending and in Chapter 6 to design concepts for new arch bridge
systems.
French Interim Recommendations on the use of ultra high-performance fibre-reinforced concrete have
been previously developed (AFGC 2002). The first set of international design rules for the material is under
development by the International Federation for Structural Concrete (fib) Task Group 8.6 (Walraven
2008). However, because ultra high-performance fibre-reinforced concrete mixes are not standardized and
are not yet widely produced, a universal material model for ultra high-performance fibre-reinforced con-
crete does not exist (Graybeal 2006). Thus, mechanical properties of these concrete mixes are sensitive to
differences in local raw materials, mixing procedures, and testing procedures. It was therefore necessary to
develop a stress-strain model appropriate for the concrete mix used at the University of Toronto.
An ultra high-performance fibre-reinforced concrete mix was developed by Habel et al. (2008) at the Uni-
versity of Toronto and École Polytechnique de Montréal using locally available materials. The mix exhib-
15
2. Ultra High-Performance Fibre-Reinforced Concrete
ited a compressive strength of 128 MPa and a tensile strength of 11.0 MPa (Habel and Gauvreau 2008).
High compressive strength is achieved by using water reducers, a low water-to-cement ratio, and no coarse
aggregate. High tensile strength is achieved by incorporating steel fibre reinforcement into the concrete.
These fibres are intended to provide crack control and member strength, and offer the potential to reduce
The French Interim Recommendations (AFGC 2002) gives three criteria for differentiating ultra high-per-
formance fibre-reinforced concrete from other high-performance cementitious materials: (1) material has
compressive strength that is systematically greater than 150 MPa, (2) material systematically uses fibres to
improve ductility and to modify conventional requirements for mild and prestressing steel, and (3) materi-
The University of Toronto mix used in this study satisfies the last two criteria but does not satisfy the com-
pressive strength criterion. Habel et al. (2008), who developed this mix based on an existing European
mix, attributes the lower compressive strength of the Toronto mix to the composition of the selected local
cement. The cement composition was found to be less appropriate for the production of concretes with low
water-to-binder ratio than those used in the European mix. They also suspect that the higher specific sur-
face area of the silica fume (15 to 20 g/m2 as opposed to 12 g/m2) might have led to higher water demand in
the Toronto mix as compared to the European mix. The compressive strength could be increased to satisfy
the strength criteria by further optimizing the selection and mix proportions of materials, by improving
the equipment and procedure used for mixing and casting, or by modifying the curing method through
the use of pressure or heat treatment. While these changes could be made, the concrete mix still exhibits
excellent material properties that can be used in its present formulation to design efficient, light-weight
structures. Increasing the strength of the concrete above 128 MPa and beyond 150 MPa does not necessar-
ily reduce the weight of the structure, since member geometry becomes increasingly governed by local sta-
16
2. Ultra High-Performance Fibre-Reinforced Concrete
The ultra high-performance fibre-reinforced concrete mix developed by Habel et al. (2008) is presented in
Table 2-1. Fibre content is 5.5% of total volume and water-to-cement ratio is 0.25 by mass. This mix exhib-
ited good workability, low permeability, and high strength. This mix design was used as the basis for all ex-
Six batches of ultra high-performance fibre-reinforced concrete were prepared according to the mix design
shown in Table 2-1. The first three batches were 48 litres in volume, while the remaining batches were 52
litres in volume. Materials were mixed in a 0.17 m3 mortar mixer (see Figure 2-1) in the following order: (1)
the cement, silica fume, and sand were dry mixed until they were homogeneous, (2) the superplasticizer
and water, which were pre-mixed, were added and mixed until no dry materials remained, and (3) the steel
fibres were added and mixed in. The mixing process took 5 to 10 minutes per batch.
The finished batches were poured into a concrete bucket (see Figure 2-1) and then placed into a wheelbar-
row. Cylinder moulds were filled using hand shovels. Since the concrete was self-consolidating, no vibra-
17
2. Ultra High-Performance Fibre-Reinforced Concrete
Two sets of cylinder tests were conducted to investigate the stress-strain behaviour of ultra high-perform-
ance fibre-reinforced concrete in compression. All cylinders were nominally 100 mm in diameter and 200
mm in length. Cylinders from Set 1 were moist-cured at room temperature until they were tested 27 days
after casting. Cylinders from Set 2 were moist-cured at room temperature for 7 days and tested 60 days
after casting. The ends of all cylinders were saw-cut and left uncapped. All cylinders were tested using an
mts testing machine, with initial displacement rates of 4.44 μm/sec. To produce this displacement rate,
machine head displacements were continuously used to control the application of load during each test.
For cylinders from Set 1, vertical displacements of the cylinder were measured by two linear variable di-
fferential transformers, with gage lengths of 153.5 mm. For cylinders from Set 2, vertical displacements
were measured from machine head displacements. Load versus vertical displacement data of the tested
18
2. Ultra High-Performance Fibre-Reinforced Concrete
batch U2 batch U5
specimen 6 specimen 5
Load
in kN
1000
780 kN 865 kN 866 kN
500
0
0 0.5 1 1.5 batch U5 batch U5
displacement data is displacement in mm specimen 4 specimen 6
averaged from two batch U1
transformers specimen 1
Figure 2-2. Load versus displacement data of cylinders from Set 1, tested 27 days after casting
load applied by
MTS testing machine
batch U5 batch U6
specimen 7 specimen 7
Load
in kN
1000 1024 kN
951 kN 950 kN
500
0
0 0.5 1 1.5 batch U5 batch U6
displacement in mm specimen 8 specimen 8
batch U4
specimen 4
Figure 2-3. Load versus displacement data of cylinder from Set 2, tested 60 days after casting
19
2. Ultra High-Performance Fibre-Reinforced Concrete
The cylinders exhibited nearly linear behaviour until their peak load, except for Cylinders U2-6 and U5-7
(cylinder id numbers correspond to batch number and specimen number, respectively). The linearity of a
material can be quantified by calculating the ratio of the tangent modulus of elasticity in compression to
the secant modulus at peak compressive stress. The tangent modulus of elasticity is given by the stress at
40% of peak compressive stress divided by the strain of the cylinder at this stress. The secant modulus at
peak compressive stress is given by the peak compressive stress divided by the strain of the cylinder at this
stress. Stress in the concrete is given by the applied load divided by the cross sectional area of the cylinder
(7850 mm2). The strain in the concrete is given by the average displacement of the linear variable differen-
tial transformers divided by their gage lengths of 153.5 mm. The average ratios between the tangent and
secant moduli for cylinders from Set 1 was 1.28, as shown in Table 2-2. This average ratio value is consistent
with results obtained from cylinder tests conducted by Graybeal (2006). Strain and modulus values of cyl-
inders from Set 2 cannot be calculated accurately based on machine head displacements, and so are not
tabulated in Table 2-2. Machine head displacements were not linearly related to transformer displacement
Calculated averages exclude values from U5-5 because of its inordinately high tangent modulus of elasticity.
20
2. Ultra High-Performance Fibre-Reinforced Concrete
Because cylinder tests were displacement controlled, post peak ductility could be investigated. Most cylin-
ders exhibited ductility beyond their peak compressive load. As shown in Figures 2-2 and 2-3, the post-
peak response of each cylinder differed. Cylinders U5-4 and U5-6 from Set 1 lost about half their load ca-
pacity soon after sustaining their peak load, while the other three cylinders from this set exhibited long
yield plateaus. Cylinders from Set 2 all exhibited some decrease in load beyond the peak load. In all but
Based on the test results, the load-carrying capacities of the cylinders beyond their peak loads are high, but
inconsistent. The sudden loss of strength in Cylinders U5-4, U5-6, and U4-4 may be a result of: (1) subop-
timal mixing procedures and equipment, (2) the small size of production, (3) misalignments of the saw-cut
ends of the cylinders causing non-uniform stresses, or (4) the use of machine head displacements as the
feedback for displacement control. Calibrating and standardizing the mixing procedure to the local mater-
ials and equipment can help to improve the consistency of results. This optimization process is most suit-
able in an industrial setting, where the size of production can be increased. Using machine displacements
to control the application of load on cylinders is not ideal because the displacement of the machine head is
not perfectly correlated with the displacement of the cylinder. This discrepancy may cause the machine to
apply loads that are not in sync with the desired cylinder displacements, causing Cylinders U5-4, U5-6,
and U4-4 to be unloaded as they would in a load-controlled test. If this is indeed the case, then the sudden
loss of strength observed in these cylinders may be an artifact of the test equipment. Improving the test
Given that the inherent and observed consistency and reliability of the concrete can be improved through
the means discussed above, it is assumed that the University of Toronto ultra high-performance fibre-rein-
forced concrete mix is inherently more ductile in compression than conventional concrete. This limited
ductility is incorporated into the material model proposed later in this section.
21
2. Ultra High-Performance Fibre-Reinforced Concrete
All cylinders tested did not fail by brittle crushing, which is the mode of failure of conventional concrete
cylinders. Instead, the cylinders continued to carry load at strains higher than conventional concrete due
to the ability of the dispersed fibres to keep the concrete matrix together. A photograph of a failed ultra
Table 2-3. Compressive strengths of cylinders organized by batch, age, and curing regime
batch- age curing peak stress batch- age curing peak stress
specimen days MPa specimen days MPa
U2-1 7 moist 90.1 U6-4 28 moist 90.3
U2-2 7 moist 84.2 U6-5 28 moist 96.1
U5-1 7 moist 89.2 U6-6 28 moist 111.2
U5-2 7 moist 83.7 U1-1 28 moist 99.3
U5-3 7 moist 90.6 U1-2 48 air 124.3
U2-3 14 moist 94.7 U1-3 48 air 138.3
U2-4 14 moist 99.3 U2-7 49 air 111.4
U6-1 14 moist 89.6 U2-8 49 air 118.4
U6-2 14 moist 80.7 U2-9 49 air 118.6
U6-3 14 moist 75.3 U3-4 54 air 127.3
U2-5 27 moist 106.5 U3-5 54 air 134.4
U2-6 27 moist 105.2 U3-6 54 air 134.6
U5-4 27 moist 110.1 U5-7 60 air 137.7
U5-5 27 moist 106.3 U5-8 60 air 130.4
U5-6 27 moist 110.3 U4-4 60 air 121.1
U3-1 28 moist 113.1 U6-7 60 air 129.2
U3-2 28 moist 113.4 U6-8 60 air 121.0
U3-3 28 moist 101.1
22
2. Ultra High-Performance Fibre-Reinforced Concrete
A total of thirty-five ultra high-performance fibre-reinforced concrete cylinders were tested for compress-
ive strength. Cylinders differed by batch, age of concrete after casting, and curing regime. Test results are
Age after casting and compressive strength of the tested cylinders are plotted in Figure 2-5. The results
show an increase in strength with age between 7 and 60 days. The average strength of the cylinders tested
27 or 28 days after casting was 107 MPa. The average strength of the cylinders tested at 7 days after casting
was 87.6 MPa, which is 82% of the 27 or 28 day strength. This shows that this material gains strength
quickly. The average strength of the air-cured cylinders tested 60 days after casting was 128 MPa, which is
100
500
50
load applied by
MTS testing machine
0 0
0 28 56
age after casting in days
The proposed compressive stress-strain models for the concrete mix under investigation is shown in Fig-
ure 2-6 and is labelled as “material model.” These model curves are based on average values of modulus of
elasticity in compression Ec, peak compressive stress fc!, and strain at peak stress %c!, observed from the cyl-
inder tests described earlier in this section. The model curve is defined by four points: (0, 0), (0.7fc!/Ec,
0.7fc!), (%c!, fc!), and (1.4%c!, 0.7fc!). The strain at peak stress %c! can also be expressed as fc!/Ec!, where Ec! is
the secant modulus at peak compressive stress. Stress-strain results from the cylinder tests from Set 1 are
shown in the figure to show their variance, and the degree of simplification that the model curve repres-
ents. The proposed strain limit of 1.4%c! represents a practical limit for calculation rather than the complete
loss of load-carrying capacity by concrete crushing. Straining the extreme fibres of a section beyond this
23
2. Ultra High-Performance Fibre-Reinforced Concrete
strain limit tends to result in bending moments with curvatures greater than the curvature at maximum
moment. Since designers are typically interested in determining the moment capacity of a section, then
50
0
0 !c′ Compressive strain cylinder test data
material model
0
0 4·10-3 8·10-3
Compressive strain
According to this compressive stress-strain model, three input variables fc!, Ec!, and Ec are required. Cylin-
der tests are often conducted without measuring displacement. In these cases, only ultimate load is meas-
ured, which gives the concrete strength fc! of the cylinder. Compressive stress-strain models for these tests
Empirical relationships between Ec and fc! have been proposed by Ma and Orgass (2004) and Graybeal
(2007) using other ultra high-performance fibre-reinforced concrete mixes. Many cylinders were tested in
these studies at different ages after casting, resulting in a wide spectrum of compressive strengths and
elastic moduli. From these tests, Ma and Orgass fitted their data to a one-third power equation form
(Equation 2-1) and Graybeal fitted his test data to a square root equation form (Equation 2-2):
24
2. Ultra High-Performance Fibre-Reinforced Concrete
Modulus of elasticity 80
data points are results from 5 cylinder tests
in compression
in GPa
curve proposed by Graybeal (2007):
60
Ec = 3840 fc! in MPa
40
curve proposed by Ma and Orgass (2004)
Figure 2-7. Tangent modulus of elasticity versus compressive strength of tested cylinders
Figure 2-7 shows that the empirical equations proposed by Ma and Orgass (2004), and Graybeal (2007)
underestimate the measured moduli of elasticity of the cylinders tested at the University of Toronto. Thus,
in most cases using these empirical equations will result in design calculations that lead to conservative
member capacities.
Using the linearity modular ratio nE between the tangent modulus of elasticity in compression and the sec-
ant modulus at peak compressive stress, Equation 2-2 can be modified to relate fc! and Ec!:
1
Ec! = " 3840 fc! in MPa Equation 2-3
nE
Using Equation 2-3, together with Equations 2-1 or 2-2 results in a material model that requires only com-
pressive strength fc! as an input parameter. It is assumed that this material model is sufficiently accurate for
the conceptual design studies presented in Chapter 6. Because the shape of the material model was de-
termined from the cylinder test results shown in Figure 2-2 and Figure 2-3, it is prudent to restrict the ap-
plicability of this design material model to compressive strengths between 100 MPa and 140 MPa. These
strengths are roughly those which were observed among all cylinders tested at the University of Toronto.
Extending the model beyond this range or to different concrete mixes may result in gross model errors.
25
2. Ultra High-Performance Fibre-Reinforced Concrete
Modulus of rupture tests were conducted at the University of Toronto1 to investigate the behaviour of ultra
high-performance fibre-reinforced concrete in tension. All specimens were square prisms, 150 mm by 150
mm in section, spanning 900 mm between roller supports. Elongations along the tension side of the prism
were recorded with two longitudinally oriented linear variable differential transformers with gage lengths
of 900 mm. Three specimens were moist-cured for 7 days at room temperature and tested 54 days after
casting under four-point loading. Load points are located at the third-points of the span. Load-displace-
50
steel spreader
free rotation about axis of pin free rotation about axis of pin
beam cross-section
150
mount
Load
in kN
80 79.3 kN
74.5 kN 72.9 kN
60
40
20
0
0 4 8 12 0 4 8 12 0 4 8 12
displacement data is displacement in mm displacement in mm displacement in mm
averaged from two trans- batch U4 batch U5 batch U6
formers, one on each side
1. These prism tests were part of a related research project conducted by Ivan Wu.
26
2. Ultra High-Performance Fibre-Reinforced Concrete
Because the transformers shown in Figure 2-8 do not bend along with the prism, there is some discrep-
ancy between the relative horizontal displacements of the ends of the beam and the average change in
length of the longitudinal fibres of the beam at the same level. The error associated with this second-order
effect at the initial loss of stiffness (or first significant change in slope on the diagrams) is about one-hun-
dredth of a percent, which is small and therefore can be neglected. This initial loss of stiffness corresponds
to the initiation of distributed microcracking. Data recorded after initial microcracking is not considered
crete and will not be used to characterize the stress-strain behaviour of the material.
Based on elasticity, the flexural tensile stress fflex at the extreme fibre of the prism between the third-point
loads can be calculated using Equation 2-4 (Neville 1995), where Q is the total applied load, L is the span
length, and b and d are the width and depth of the prism:
QL
fflex = Equation 2-4
bd 2
Because Equation 2-4 assumes linear-elastic behaviour, only the stresses of the prism before cracking can
be calculated accurately. These flexural tensile stresses are further reduced by a factor of 0.727 to estimate
axial tensile stresses ft according to Equation 2-5 (afgc 2002), where d is the prism depth of 150 mm, and
2 ( d / d0 )
0.7
Equation 2-5 accounts for the overestimation of stresses in small-scale flexural tests due to strain gradient
effects and is recommended in the French Interim Recommendations (afgc 2002). After applying the cor-
rection factor of 0.727, the average cracking stress of the three prisms tested were found to be 7 MPa.
Fibre strains at the level of the transformers were obtained by dividing the average displacement of the
transformers by their gage length. By assuming plane sections remain plane, bottom fibre strains were es-
27
2. Ultra High-Performance Fibre-Reinforced Concrete
timated by multiplying the strains at the level of the transformers by two, since the transformers are at one
fourth the depth from the bottom. The average cracking strain of the three prisms tested were 0.15·10-3.
Dividing the average cracking stress by the average cracking strain results in a tangent modulus of elasti-
city in tension of 46.7 GPa. This value falls within the range of moduli obtained from direct tension tests
conducted by Habel and Gauvreau (2008) who used the same ultra high-performance fibre-reinforced
Three split cylinder tests were conducted to investigate the peak tensile stress, or post-cracking strength, of
the material. All three cylinders were cast from batch U4, moist-cured for 7 days at room temperature, and
then air-cured for 21 days until they were tested. The peak tensile stresses of the cylinders were 18.1, 12.0,
and 13.8 MPa. The peak tensile stress ft! was calculated by Equation 2-6 (Neville 1995), where Q! is the peak
applied load, l is the cylinder length of 200 mm, and d is the cylinder diameter of 100 mm.
2Q !
ft! = Equation 2-6
" ld
Average measured values from the prism and split-cylinder tests, and the direct-tension tests by Habel and
Gauvreau (2008) were used to define a bilinear curve representing the tensile stress-strain response of the
ultra high-performance fibre-reinforced concrete mix being investigated at the University of Toronto (see
Figure 2-9). The material model for tension is only directly applicable to this particular mix design.
Obsolete
Et = 46.7 GPa
20 prism test data only !cr = 0.15·10-3 fcr! and !cr! based on
valid before cracking averages from prism data
fcr = 7.0 MPa
fcr
!t! = 1.5·10-3 !t! estimated using results
ft! = 14.6 MPa from direct tension tests by
10
Habel and Gauvreau (2008)
Tensile strain
28
2. Ultra High-Performance Fibre-Reinforced Concrete
The initial linear portion of the material model curve represents the uncracked, linear-elastic behaviour of
the material in tension. As the material is loaded in tension, stress increases linearly with strain until the
cracking stress fcr is reached. Upon cracking, the material exhibits strain-hardening, as represented by the
second linear portion of the material model. In this domain, strain-hardening occurs due to: (1) the forma-
tion of multiple uniformly distributed cracks that are invisible to the naked eye, and (2) the progressive
pull-out of fibres (Habel 2004). Because the damage to the concrete matrix is gradual, there is a gradual
decrease in secant modulus. This strain-hardening behaviour concludes with the peak tensile stress ft! of
the material, after which a localized, visible crack forms. After this point, the material exhibits strain-
softening behaviour, in which the stress-carrying capacity of the material diminishes as a function of the
opening of the localized crack. Because of the continued, progressive pull-out of fibres, an abrupt loss of
stress does not occur. This strain-softening domain of behaviour of ultra high-performance fibre-rein-
forced concrete is neglected in the material model proposed in this thesis. This leads to conservative estim-
For the purposes of this study and the analytical models developed later in this thesis, this tensile material
model will be assumed to remain constant. Tensile properties of conventional concrete are often related to
the compressive strength of concrete fc!. There is, however, insufficient data available to correlate tensile
cracking stresses with compressive strength. The steel reinforcement fibres, whose properties do not
change with the maturity of the concrete, provide a large share of the tensile capacity of ultra high-per-
formance fibre-reinforced concrete. Thus, assuming that the tensile behaviour of the cured material re-
29
2. Ultra High-Performance Fibre-Reinforced Concrete
Concrete shrinkage is caused by autogenous shrinkage and drying shrinkage. Autogenous shrinkage is
caused by self-desiccation, in which water is withdrawn from capillary pores by the continued hydration of
concrete after it has set. Drying shrinkage is caused by the movement of water out of the non-rigid, porous
body of concrete.
total shrinkage
0.4·10-3
autogenous shrinkage
0.2·10-3
0
0 20 40 60
Figure 2-10. Shrinkage strains as a function of days after casting. Adapted from Habel et al. (2008).
Shrinkage tests on six specimens made from the concrete mix design shown in Table 2-1 were done by
Habel et al. (2008) at the University of Toronto. Specimens were 305 mm in length and 76 mm in width
and height and were tested according to astm c 157. Autogenous shrinkage was measured from the three
specimens that were sealed with two layers of self-adhesive aluminum foil tape. Total shrinkage (the sum
of drying and autogenous shrinkage) was measured from the three unsealed specimens. These strains are
The French Interim Recommendations on ultra high-performance fibre-reinforced concrete (AFGC 2002)
suggest using a long-term shrinkage strain of 0.550·10-3. This estimate of long-term shrinkage strain correl-
ates well with the shrinkage strain data shown in Figure 2-10 obtained by Habel et al. (2008). Shrinkage
tests done by Graybeal (2006) on a similar material resulted in a long-term shrinkage strain of 0.555·10-3.
30
2. Ultra High-Performance Fibre-Reinforced Concrete
Creep is the increase of strain over time caused by a constant stress in a concrete member. No tests were
done at the University of Toronto to study the creep behaviour of ultra high-performance fibre-reinforced
0
0 100 200 300 400
Figure 2-11. Creep strain of an untreated ultra high-performance fibre-reinforced concrete specimen over time.
Adapted from Graybeal (2006).
Creep is normally quantified using the creep coefficient, which is defined as the ratio of additional creep
strain with time to the initial elastic strain caused by a constant applied stress. The French Interim Recom-
mendations on ultra high-performance fibre-reinforced concrete (AFGC 2002) suggest using a long-term
creep coefficient of 0.8 if no heat treatment is applied to the concrete. Creep tests conducted by Graybeal
(2006) resulted in a long-term creep coefficient of 0.78 for specimens without heat or steam treatment, as
shown in Figure 2-11. The long-term creep coefficient and shrinkage strains presented above will be used in
31
Chapter 3. Slender Ultra High-Performance Fibre-Reinforced
Concrete Columns
In this chapter, an analytical model for determining the ultimate capacity of slender ultra high-perform-
ance fibre-reinforced concrete members in combined flexure and axial compression is developed. This
model will be used to calculate member capacities for all the primary load-carrying members in arch
bridges, including the arch, spandrel columns, and prestressed deck. The methods and insights described
in this chapter will be used in Chapter 6 to design new concepts for arches using ultra high-performance
fibre-reinforced concrete.
The content of this chapter is divided into four sections. In Section 3.1, the structural model of an eccent-
rically loaded slender column is described and the importance of considering nonlinear behaviour is illus-
trated. In Section 3.2, a general method of analysis is proposed for estimating the capacity and load-deflec-
tion response of given ultra high-performance fibre-reinforced concrete columns, considering material
and geometric nonlinearities. The general method is validated experimentally with a set of column tests, as
described in Section 3.4. In the final section, a simplified design method is proposed for calculating ulti-
mate member capacity. This approximate method is shown to be accurate and conservative relative to the
more rigorous general method of analysis. All test specimens described in this chapter were made using
the concrete mix design given in Table 2-1. All sectional analyses described in this chapter use the material
model proposed in Section 2.1 adjusted according to the assumed or observed compressive strengths.
32
3. Slender Ultra High-Performance Fibre-Reinforced Concrete Columns
The research described in this chapter is the first known study of the behaviour of slender ultra high-per-
formance fibre-reinforced concrete members in combined flexure and axial compression. It offers a means
for designers to take advantage of the high compressive strength of ultra high-performance fibre-rein-
forced concrete and hence to exploit the economic potential offered by this material.
Q Q Q
w +
x M
line of action
of load N
Q Q M = Qe + Qw(x)
A structural model for eccentrically loaded columns is shown in Figure 3-1. In this system, moments M are
caused by axial load Q times the distance between its line of action and the centroidal axis of the member.
As load Q is applied, the centroidal axis of the member deforms away from the line of action, producing
additional moments in the system. These additional moments can be accounted for by second-order ana-
lysis, in which states of equilibrium are calculated on the geometry of the deformed structure. Moments M
calculated using second-order analysis are called second-order moments and are given by Equation 3-1,
M (x) = Qe + Qw(x)
Equation 3-1
M (x) = M 0 + M 1 (x)
The first term Qe in Equation 3-1 is called the first-order moment M0, because it corresponds to first-order
analysis, which is based on a state of equilibrium calculated on the geometry of the undeformed structure.
The second term Qw represents the additional moments M1 caused by member deformation.
33
3. Slender Ultra High-Performance Fibre-Reinforced Concrete Columns
The relative magnitude of first-order moments M0 and additional moments M1 is related to the bending
stiffness of the member. When members are stiff, deflections w are small and therefore additional mo-
ments M1 are small relative to first-order moments M0. When members are slender, deflections w are large
and therefore additional moments M1 can be large relative to first order moments M0. Thus, it is important
to calculate second-order moments M accurately when designing slender members, so that moment de-
Calculating second-order moments M accurately requires an analysis that considers both geometric and
material nonlinearities. Geometric nonlinearity refers to the nonlinear load-deflection behaviour of the
column. One way to account for geometric nonlinearity is to use Vianello’s method of successive approx-
imations (Vianello 1898, Menn 1990). According to this method, deflections w can be calculated approxim-
ately using Equation 3-2 (Menn 1990), where w0 is first-order deflection, Q is applied load, and QE is Euler
buckling load:
1
w(x) = w0 (x) Equation 3-2
1 ! Q / QE
The Euler buckling load QE is given by Equation 3-3, where EI is flexural rigidity, k is effective length factor
EI
QE = ! 2 Equation 3-3
( kL )2
Material nonlinearity refers to nonlinear stress-strain behaviour of the composite material, such as caused
by the cracking of concrete or the yielding of reinforcing steel. When used in sectional analysis, nonlinear
stress-strain curves produce nonlinear moment-curvature diagrams. One way to account for material non-
linearity when calculating deflections w is to replace flexural rigidity EI in Equation 3-3 with secant bend-
ing stiffness EI!. Secant stiffness is given by Equation 3-4 (Menn 1990), where , is curvature of the section:
34
3. Slender Ultra High-Performance Fibre-Reinforced Concrete Columns
M
EI !(N ) = Equation 3-4
"
Secant stiffness EI! varies with axial force N and moment M and is calculated as a secant to the moment-
Load Q A
Axial force N A
sectional capacity
Q Q interaction diagram
A—geometric linearity, material linearity
B—geometric nonlinearity, material linearity
C—geometric nonlinearity, material nonlinearity
e
ultimate limit state ultimate limit state
B
w 1
e B
M ultimate N*
N load Q*
C C
Figure 3-2. Load deflection and sectional response of a slender column given different analytical assumptions
Comparing the linear and nonlinear response of a given eccentrically loaded slender column illustrates the
importance of considering both geometric and material nonlinearities. The column considered is 2300
mm long, is square in section with side lengths of 100 mm, and is made of ultra high-performance fibre-
reinforced concrete without conventional reinforcing steel. The applied load has an initial eccentricity e of
10 mm to the centre of the column. Figure 3-2 shows the load-deflection and axial force-moment response
of the column according to three sets of analytical assumptions that follow, corresponding to Curves A, B,
and C:
Curve A: This response considers geometric linearity and material linearity. Equilibrium is calculated
based on undeformed geometry, thus only first-order moments M0 are considered. The stress-strain beha-
viour of the material is assumed to be linear-elastic. First order deflections w0 are calculated using an elast-
Curve B: This response considers geometric nonlinearity and material linearity. Equilibrium is calcu-
lated based on deformed geometry, thus additional moments M1 are considered and first-order deflections
35
3. Slender Ultra High-Performance Fibre-Reinforced Concrete Columns
w0 are magnified according to Equation 3-2. Because a linear-elastic material is assumed, first-order deflec-
tions w0 and the Euler buckling load QE are calculated using an elastic modulus E of 42000 MPa and un-
cracked moment of inertia of 833 cm4. Deflections w from Curve B are larger than those from Curve A for
any given load Q. The ratio between additional moments M1 and first-order moments M0 increases as the
applied load Q is increased. For the slender column considered, this ratio increases beyond two. This
shows that neglecting second-order effects, as done for Curve A, can lead to gross underestimates of bend-
Curve C: This response considers both geometric and material nonlinearities and is calculated based
on the general method described in Section 3.2. This method does not use the approximations considered
by Equations 3-2, 3-3, and 3-4, which are purely sectional calculations. Instead, the general method con-
siders the overall stability of the member. It uses the nonlinear stress-strain model proposed in Chapter 2
(assuming a compressive strength of 120 MPa) to calculate nonlinear moment-curvature diagrams, which
are then used to determine curvatures and deflections along the length of the column. Because the materi-
al behaves elastically at first, Curve C closely follows Curve B until the ultimate load Q* is reached. Ulti-
mate loads Q* are defined as the maximum applied axial loads that can be sustained by columns of given
length, cross-section, material, and eccentricity. Beyond this ultimate limit state, the applied load Q must
be decreased to maintain a state of equilibrium. Any attempt to increase load beyond the ultimate load will
rapidly increase deflections and moments causing the member to become unstable. As such, the ultimate
load Q*, which cannot be calculated using sectional calculations as in Curve B, marks an ultimate limit
state that occurs before sectional failure. The bending moment that is sustained by the critical section of
the column when the ultimate load Q* is reached will be called the ultimate moment M* (this is not the
As illustrated by the different analyses described in the previous example, it is important to consider geo-
metric and material nonlinearities and the overall stability of the system when designing slender members.
36
3. Slender Ultra High-Performance Fibre-Reinforced Concrete Columns
For slender members loaded in compression and bending, ultimate limit states can be reached before sec-
The primary objective of the general analysis method described in the following section is to determine
the ultimate loads Q* of a given slender ultra high-performance fibre-reinforced concrete column for all
all possible values of initial eccentricity of load e. The structural model considered has been previously dis-
cussed and is shown in Figure 3-1. The model has a constant cross section, pins at each end of the column,
and an axial load, which is applied eccentric to the centre of the column.
From external quantities Q* and w*, the limiting sectional forces N and M* of the critical section of the
column can be calculated (see Figure 3-2). Ultimate moment M* is given by Equation 3-5, where e is the
Expressing slender column capacity in terms of limiting sectional forces of the critical section allows for
the use of N-M* interaction diagrams, which correspond to the various ultimate limit states (Q=Q*) of the
column for all possible values of initial eccentricity. These ultimate N-M* interaction diagrams are reduced
versions of traditional N-M interaction diagram, which are instead calculated based on sectional failure.
Calculating N-M* interaction diagrams for a given column is the secondary objective of the general ana-
lysis method.
3.2.1 Methodology
A rational method for calculating N-M* interaction diagrams for slender prestressed concrete columns
(arranged as in Figure 3-1) is presented by Nathan (1985). Nathan’s method requires the following input:
material stress-strain curves, cross sectional geometry, and pin-to-pin length of the column. The method
37
3. Slender Ultra High-Performance Fibre-Reinforced Concrete Columns
Step 1. A set of sectional analyses is used to calculate contours of equal curvature on an axial force-
moment diagram. These contours are used to calculate a set of nonlinear moment curvature diagrams,
Step 2. Column deflection curves w(x), which are column deflections along their length for a given
axial load Q and mid-length deflection w(0), are calculated by solving Equation 3-1 numerically. A set of
column deflection curves are calculated by changing the mid-length deflection. The ultimate deflection
w*, corresponding to Q=Q*, is taken as the mid-length deflection w(0) of the column deflection curve
with the largest end eccentricity e. From w*, the ultimate moment M* of the critical section is calculated.
Step 3. The member capacity N-M* interaction diagram of the slender member is determined by re-
cording ultimate moments M* of the critical section for all values of axial force (N=Q) up to the maximum
In the following sections, this method will be adapted for use with ultra high-performance fibre-reinforced
concrete by using the material model proposed in Chapter 2. The interactions among axial force, moment,
and curvature of a given section, which are unique to the stress-strain curve and geometry of the section
assumed, will be discussed. In addition to proposing a general method for calculating N-M* interaction
diagram for slender columns, a method for calculating load-deflection response of slender columns will be
presented, which is not included in Nathan’s formulations. This latter calculation method predicts the ulti-
mate load Q* and corresponding ultimate deflection w* of columns of given length, sectional geometry,
and eccentricity of load. Load-deflection response will be used to validate the general method
experimentally.
The first step in the general method is to calculate a set of nonlinear moment-curvature M-ϕ diagrams,
each diagram corresponding to a different axial force N. An effective way to do this is to calculate contours
of equal curvature on an axial force-moment diagram and then interpolate M-ϕ diagrams from those con-
tours. This computational method offers two advantages: (1) it does not require iterative calculations, and
38
3. Slender Ultra High-Performance Fibre-Reinforced Concrete Columns
(2) the contours of equal curvature give insight into the sectional behaviour of ultra high-performance
fibre-reinforced concrete sections. A given contour of equal curvature is found by calculating sectional
forces corresponding to a set of planes of strain, each with the same curvature ϕ but different strains on the
flexural compression face of the column %cf. Contours of equal curvature can be determined by dividing
the ranges of %cf and ϕ into intervals as in Equation 3-6, where %t! is strain at peak stress in tension, %c! is
strain at peak stress in compression, and d is depth of section. The limits of %cf and ϕ correspond to the
Contours of equal curvature for a 100 mm by 100 mm square section using the material model presented
in Chapter 2 and a compressive strength of 130 MPa are shown in Figure 3-3. Figure 3-3b illustrates how
contours of equal curvature are calculated. First, various planes of strain with the same value of curvature
are chosen. Second, stress profiles for each plane of strain are calculated using the constitutive laws of the
material. Third, axial forces and moments are calculated through the integration of stresses. Last, the res-
ulting sectional forces from each plane of strain are plotted in axial force-moment space (see Figure 3-3a).
The envelope of all contours of equal curvature forms the axial force-moment N-M interaction diagram
For this section, setting the strain on the flexural compression face %cf to 1.2%c! and setting the strain on the
flexural tension face %tf to %t! gives a good estimate of the plane of strain that results in the maximum mo-
ment of resistance of the section. This state is analogous to the balanced point in a conventionally rein-
forced concrete section, in which the section fails simultaneously by the concrete crushing at the flexural
compression face and by the yielding of steel near the flexural tension face. This state is labelled as the “bal-
39
3. Slender Ultra High-Performance Fibre-Reinforced Concrete Columns
–600 100 mm
! = 10 20 30 40 50
100 mm
! = 30 rad/km ... ...
–400
4 D
–200
0 0
0 4 8 12 16 kN·m 0 30 60 90 120 rad/km
Moment M Curvature !
Moment-curvature diagram points for a given axial force N can be found by interpolating values of mo-
ment M from each contour of equal curvature. In Figures 3-3c through 3-3f, ten moment-curvature dia-
grams are calculated for ten different values of axial force N using contours of equal curvature. Points A
through F illustrate how moment-curvature diagrams for N=–300 kN and N=–800 kN are calculated. For
40
3. Slender Ultra High-Performance Fibre-Reinforced Concrete Columns
example, Point B lies on the intersection of N=–300 kN and contour of equal curvature with ϕ=30 rad/km.
The moment M associated with Point B is 9.5 kN·m, which is plotted on the moment-curvature diagram
shown in Figure 3-3d. Interpolating values of moment from each contour of equal curvature in this way
creates a complete nonlinear moment-curvature diagram for a given section and axial force N.
The axial force-moment-curvature N-M-ϕ relationship depicted in Figure 3-3 reveals trends in the mo-
spacing of the contours of equal curvature at a given axial force N is related to the linearity of the moment-
curvature diagram. Constant spacing corresponds to linear behaviour, while decreasing spacing corres-
ponds to decreases in secant stiffness EI!. The shape of the moment-curvature diagram depends on wheth-
er the axial force N is less than or greater than the axial force at the balanced point Nb, as described below:
Case 1: For states of strain where N<Nb, the spacing between contours of equal curvature get smaller
and smaller until they converge as curvature is increased (see Figure 3-3c). This corresponds to sectional
failure caused by the pull-out of steel fibre reinforcement. The initial slope of the resulting moment-
curvature diagram, shown in Figure 3-3d is equal to the uncracked, elastic flexural rigidity EI0, where E is
the elastic modulus of the concrete, and I0 is the moment of inertia of the gross uncracked section. The
cracking point, which occurs when the flexural tension face reaches the cracking strain %tf=%cr, is indicated
for the curve labelled N=–300kN in Figure 3-3d. For states of strain where N<Nb, the secant stiffness at ini-
tial cracking EI!cr is equal to EI0. After the section cracks, the secant stiffness EI! begins to decrease and
small increases in moment require large increases in curvature. The difference between cracking moments
and peak moment for the N<Nb curves shown in Figure 3-3d are relatively large.
Case 2: For states of strain where N≥Nb, the contours of equal curvature fold back on one another
(see Figure 3-3e), suggesting that straining the section beyond 1.4%!c leads only to states of equilibrium on
the descending post-peak branch of moment-curvature diagrams. The non-convergence of the contours
corresponds to sectional failure caused by concrete crushing. The ascending branch of moment-curvature
diagrams in Figure 3-3f is bilinear before cracking due to the bilinear form assumed for the compressive
41
3. Slender Ultra High-Performance Fibre-Reinforced Concrete Columns
stress-strain model. For states of strain where N≥Nb, the secant stiffness at cracking EI!cr decreases as axial
force N is increased. The difference between cracking moments and peak moments of the N≥Nb curves
With the axial force-moment-curvature N-M-ϕ relationship of the section established, the equilibrium of a
given member can now be studied. Equating moment demand (Equation 3-1) and moment resistance
(Equation 3-4) gives Equation 3-7, where Q is applied load, e is eccentricity of load, w(x) is deflection of the
column along its length, EI! is secant stiffness, N is axial force of the section, and ϕ(x) is curvature of the
By approximating curvature ϕ(x) as the second derivative of the column deflection w"(x) using a second-
order, small-displacement approach, and by equating axial force N with applied load Q, the following di-
Solving Equation 3-8 for w(x) for a given applied load Q, mid-length moment M(0), and column length
produces one column deflection curve. Because secant stiffness and corresponding nonlinear moment-
curvature diagram cannot be expressed in functional form, Equation 3-8 has to be integrated numerically
over the length of the column. Only half the column needs to be considered, since the column and load
configurations are symmetric. To perform the numerical integration, the half-length of the column is di-
42
3. Slender Ultra High-Performance Fibre-Reinforced Concrete Columns
–EI!(x)·w"(x)+Qw(x)+Qe=0
e+w(x)
e
Q Q
start end
column column
mid-length L end
2
w(0) w M w w(L/2)
w!(0) w! w" w! w!(L/2)
!x !x !x !x !x !x
2 2
x
One column deflection curve w(x) is calculated given the following input parameters: column length L,
axial load Q, moment-curvature diagram M-ϕ, and mid-length moment M(0). The numerical integration
procedure begins at mid-length (x=0) and proceeds toward the end of the column (x=L/2). The value of
initial eccentricity e is solved at the end of the integration procedure, rather than being specified as an in-
put parameter. Listed below are the steps for calculating one column deflection curve w(x), adapted from
1. The total eccentricity w(0)+e between the line of action of load and the centroid of the column at
mid-length is calculated based on the specified mid-length moment M(0) and axial load Q:
M (0)
w(0) + e = Equation 3-9
Q
3. The total eccentricity w+e and slope w! at mid-length are assigned to the total eccentricity and slope
43
3. Slender Ultra High-Performance Fibre-Reinforced Concrete Columns
4. The total eccentricity w+e and moment M at the middle of the current element are approximated:
5. The curvature ϕ at the middle of the current element is found by dividing moment M by secant stiff-
ness EI! using the given moment-curvature diagram. The second derivative of deflection w" is as-
M (x0 + 12 "x)
w!!(x0 + 12 "x) # $ (x0 + 12 "x) = Equation 3-14
EI !
6. The total eccentricity w+e at the end of the current element is approximated using the first three
7. The slope w! at the end of the current element is approximated using the first two terms of its Taylor
series expansion:
8. The total eccentricity w+e and slope w! at the end of the current element are assigned to the total ec-
9. Steps 4 through 8 are repeated for all elements until the end of the column is reached.
44
3. Slender Ultra High-Performance Fibre-Reinforced Concrete Columns
10. Because the deflection of the column is zero at its end, the total eccentricity w+e at the end of the
11. The column deflection curve w(x) can be determined by subtracting the initial eccentricity e from
The purpose of calculating column deflection curves w(x) is to identify which combinations of axial load Q
and initial eccentricity e cause the column to fail. To do this, a family of column deflection curves needs to
be calculated, each with the same axial load Q, but with different specified mid-length moments M(0). The
mid-length moment of the first curve in the family is assigned the peak moment from the given moment-
curvature diagram. Subsequent curves in the family are assigned mid-length moments that are success-
ively less than the peak moment, until a mid-length moment of zero is reached. As the number of column
deflection curves in the family is increased, the resolution of the analysis is increased. The higher the resol-
ution is, the higher the accuracy is in determining the ultimate load. Some of the resulting column deflec-
tion curves will have initial eccentricities e that are negative. These curves represent unstable column
deflection states that occur after the ultimate load has been reached. The process of identifying the ulti-
mate limit state among deflection curves in a family is explained in the next section.
45
3. Slender Ultra High-Performance Fibre-Reinforced Concrete Columns
This section describes how member capacity N-M* interaction diagrams are calculated. Each point on an
N-M* interaction diagram corresponds to a unique ultimate limit state at which axial load Q has reached
the ultimate load Q*. One point on a member capacity N-M* interaction diagram is calculated by identify-
ing which curve in a family of column deflection curves corresponds to an ultimate limit state.
Column deflection curves for an ultra high-performance fibre-reinforced concrete member subjected to
an axial load Q1 of 600 kN are shown in Figure 3-5a,c,e. The dashed line directly below the column deflec-
tion curves is the lines of action of the applied load. Three column lengths are considered: 1200 mm, 2050
mm, and 2800 mm. The nonlinear moment-curvature diagram used to calculate the column deflection
curves is based on a 100 mm by 100 mm square section and the material model presented in Chapter 2 us-
ing a compressive strength of 130 MPa. Curves A, D, and G correspond to mid-length moments causing
sectional failure (peak moment from moment-curvature diagram). All other curves correspond to mid-
46
3. Slender Ultra High-Performance Fibre-Reinforced Concrete Columns
e+w(x)
e
Q Q
(c)
D
(f)
Axial N-M interaction
Unstable column (e)
force N
This 2800 mm column is G
unstable at this axial load H N=Q1
Q1, since all column
I N=QE (Euler buckling)
deflection curves in the Q1
Q1
family extend below the
line of action of load. N-M* interaction
1400 mm
Moment M=Q(e+w(0))
Figure 3-5. Column deflection curves and sectional responses of mid-length sections for various column lengths
In Figure 3-5a,c,e, the left sides of the column deflections curves are at mid-length, while the right sides are
at the pinned ends of the columns. These two boundaries define the length over which the half-column is
allowed to bend. Mathematically, this corresponds to the bounds over which Equation 3-8 is integrated.
Since the deflection of the columns is zero at the pinned supports w(L/2)=0, the distance from the line of
47
3. Slender Ultra High-Performance Fibre-Reinforced Concrete Columns
action to the end of the column is equal to the initial eccentricity of load e. The value of e is different for
each column deflection curve in the family of curves. In general, the column deflection curve with the
greatest eccentricity at the end of the column is the curve that corresponds to ultimate load Q*. This will
be explained in greater detail according to the three domains of slenderness described below:
Slender Column: The curve with the greatest end eccentricity e is the curve with the greatest mid-
length eccentricity w(0)+e. The mid-length moment M(0) of this curve is equal to the peak moment of the
moment-curvature diagram, and thus corresponds to sectional failure. This means that the column reaches
sectional failure without first reaching ultimate load Q*. This type of slender column behaviour is illus-
trated by the column deflection curves shown in Figure 3-5a. All column deflection curves in the diagram
are subjected to an axial load Q1 of 600 kN. Curve A is the critical column deflection curve with end ec-
centricity e1. Curves B and C are noncritical column deflection curves because their end eccentricities are
The diagram in Figure 3-5b shows possible axial force-moment responses of the mid-length section that
lead to the column deflection curves represented by Curves A, B, and C. Each axial force-moment re-
sponse corresponds to a unique eccentricity of load. Only Curve A, which has the highest eccentricity of
load and lies on the section capacity N-M interaction diagram, represents a state of failure when Q=Q1.
Very Slender Column: The curve with the greatest end eccentricity e does not correspond to the curve
with the greatest mid-length eccentricity w(0)+e. In this case the curve with the greatest end eccentricity
has a mid-length moment M(0) that is less than the peak moment of the moment-curvature diagram. This
48
3. Slender Ultra High-Performance Fibre-Reinforced Concrete Columns
means that the column reaches ultimate load Q* before reaching sectional failure. This type of very slender
column behaviour is illustrated by the column deflection curves shown in Figure 3-5c. All column deflec-
tion curves in the diagram are subjected to an axial load Q1 of 600 kN. Curve E is the critical column
deflection curve with end eccentricity e2. Curves D and F are noncritical column deflection curves because
their end eccentricities are less than e2. Curve D, which has the highest mid-length eccentricity w(0)+e,
folds back on to the curves with lower mid-length eccentricities. This means that two column deflection
curves may have the same end eccentricity e and may be part of the same axial force-moment response.
This is the case for Curves D and F, which share the same end eccentricity e3. This pair of column deflec-
(c)
D
The diagram in Figure 3-5d shows possible axial force-moment responses of the mid-length section that
lead to the column deflection curves represented by Curves D, E, and F. The points representing the mid-
length sectional forces of Curves F and D lie on the same axial force-moment response. The former point
lies on the ascending branch of the sectional response, while the latter point lies on the descending branch.
Only the former point (Curve F) represents a stable state of equilibrium for which the ultimate load Q*
has not been reached. The point representing the mid-length sectional forces of Curve E lies at the apex of
the second axial-force moment response, and thus corresponds to an ultimate limit state. The mid-length
moment of this curve is the ultimate moment M* for this value of applied load Q=Q1. The ultimate mo-
49
3. Slender Ultra High-Performance Fibre-Reinforced Concrete Columns
ment M* is less than the moment causing sectional failure (sectional failure is marked by the dashed N-M
interaction diagram).
A second envelope is marked in Figure 3-5d with a dotted line. This envelope, called the member capacity
N-M* interaction diagram, is a locus of all possible ultimate limit states for the specific length of column
considered. At each point along this envelope, different combinations of ultimate load Q* and ultimate
Unstable Column: At the end of the column, all curves are below the line of action of applied load,
corresponding to negative end eccentricities e<0. This is the case for all column deflection curves shown
in Figure 3-5e. This means that the column is unstable under axial load Q1. As shown in Figure 3-5f, the
dotted N-M* interaction diagram for this specific length of column is less than Q1. The upper boundary of
this interaction diagram is equal to the Euler buckling load of the column, as calculated by Equation 3-3.
(f)
Axial N-M interaction
Unstable column (e)
force N
This 2800 mm column is G
unstable at this axial load H N=Q1
Q1, since all column
I N=QE (Euler buckling)
deflection curves in the Q1
Q1
family extend below the
line of action of load. N-M* interaction
1400 mm
Moment M=Q(e+w(0))
With the three domains of slenderness described above, ultimate moments M* for any applied axial load Q
can be determined. The axial load Q equals the ultimate load Q* when the moment of the mid-length sec-
tion M(0) reaches ultimate moment M*. Ultimate moments M*, thus, represent the highest moment M
that the given member can carry safely at the specified axial load Q. Member capacity N-M* interaction
diagrams are the collection of ultimate moments M* for all different values of axial force N. By equating
axial force N with axial load Q, sectional limits of the critical mid-length section can be defined, represent-
ing states at which ultimate loads Q* of the member have been reached. Constructing the member capa-
50
3. Slender Ultra High-Performance Fibre-Reinforced Concrete Columns
Step 1. A family of column deflection curves is calculated for a given axial load Q.
Step 2. The axial load Q and mid-length moment M of the deflection curve with the highest eccentri-
city e at the end of the column are recorded (these are the ultimate load Q* and ultimate moment M*, re-
spectively). The mid-length moment recorded is zero if all curves are below the line of action of the load at
the end of the column (this corresponds to the unstable column condition).
Step 3. The axial load Q is incremented and Steps 1 and 2 are repeated until the failure load in pure
axial compression Q! is reached. Q! is given by Equation 3-20, where fc! is the compressive strength of con-
Step 4. Recorded values are plotted on an axial force-moment diagram, setting recorded values of Q
equal to N. This locus of points represents all possible ultimate limit states of the given column.
The member capacity N-M* interaction diagram for a 100 mm × 100 mm ultra high-performance fibre re-
inforced concrete column with a length of 2310 mm is shown in Figure 3-6a (dotted curve). This length
corresponds to a slenderness ratio ( of 80. Slenderness ratios ( are normalized measures of column length
L as defined by Equation 3-21, where r is radius of gyration of the section and k is effective length factor:
kL
!= Equation 3-21
r
Each point on the N-M* interaction diagram corresponds to a unique combination of ultimate load Q*
(N=Q*) and ultimate moment M*. If the given column were loaded axially with initial eccentricities e of
0.1, 0.5, 1, 2, 4, 8, 16, and 32 mm, the sectional response of the mid-length section of the column would fol-
low the response curves shown in Figure 3-6a. These response curves show that each point on the N-M*
51
3. Slender Ultra High-Performance Fibre-Reinforced Concrete Columns
Figure 3-6. Member capacity interaction diagrams for various slenderness ratios
Member capacity N-M* interaction diagrams define the sectional capacity of the critical section of the
column. If the axial-force and second-order moment for given load cases are within the space bounded by
N-M*, then the member can safely carry the load. Load cases with N-M combinations that lie outside the
The member capacity N-M* interaction diagram changes with the slenderness ratio ( of the column. Fig-
ure 3-6b shows the member capacity interaction diagrams for different slenderness ratios of the 100 mm ×
100 mm column considered previously. Decreasing ( increases the allowable axial forces and moments
that can be sustained by the mid-length section of the column. At some value of (, the N-M* interaction
diagram utilizes the full capacity of the section. In this case, this occurs when ( is decreased below 60. Dis-
playing the trade-off between slenderness and member capacity visually as in Figure 3-6b can help design-
ers select member sizes that are slender, safe, and structurally efficient.
The load-deflection Q-w response of slender concrete columns with given initial eccentricities of load e
can be calculated using the column deflection curves described in the previous section. The calculation
process is similar to the way ultimate moments are calculated, except that the column deflection curve of
52
3. Slender Ultra High-Performance Fibre-Reinforced Concrete Columns
interest within a family of curves is the one that produces the given eccentricity at the end of the column
(previously, the curve with the highest end eccentricity was selected among those in the family).
Figure 3-5c, which shows column deflection curves for a very slender column, can be used to illustrate this
process. The column arrangement considered is a 100 mm × 100 mm ultra high-performance fibre-rein-
forced concrete column with a length of 2050 mm and a constant eccentricity of load e3. Based on the fig-
ure, the mid-length moments of both Curves D and F would be recorded for this value of axial load Q1 and
eccentricity e3. Setting axial force equal to axial load N=Q1, two points, labelled D and F, can be plotted as
a part of the axial force-moment response of the mid-length section (see Figure 3-5d). Other N-M points
can be plotted by repeating this process with different families of column deflection curves corresponding
Once the complete axial force-moment N-M response of the mid-length section has been calculated, the
load-deflection Q-w response of the column can be calculated. In this column arrangement, mid-length
deflection w(0) is given by the following equation, where axial load Q is taken as axial force N:
M (0)
w(0) = !e Equation 3-22
Q
Responses calculated using this method assume that the loads applied are deflection-controlled and that
they can be reduced once ultimate load Q* has been reached. These load-deflection responses will be used
53
3. Slender Ultra High-Performance Fibre-Reinforced Concrete Columns
The computer program qult was written by the author to model the behaviour of ultra high-performance
fibre-reinforced concrete slender members under combined compression and bending, according to the
general method described in Section 3.2. The program considers nonlinear material behaviour in sectional
analysis and calculates equilibrium based on deformed geometry (i.e. second-order analysis). For a given
stress-strain curve, constant sectional geometry, and column length, qult returns the member capacity N-
M* interaction diagram of the column, which bounds all admissible combinations of axial force and mo-
ment that can be carried safely by the critical section of the member (see Section 3.2.4). For a given stress-
strain curve, constant sectional geometry, column length, and eccentricity of load, qult returns the load-
deflection Q-w response of a column subjected to an eccentrically applied axial load (see Section 3.2.5).
From this, the ultimate load, or maximum load sustained by the column, is determined.
The overall structure of the program is based on a similar program developed by Nathan in his paper en-
titled "Rational Analysis and Design of Prestressed Concrete Beam Columns and Wall Panels" (1985). As
shown in Figure 3-7, qult is organized as a hierarchy of functions: matmodel defines the stress-strain re-
sponse of the material, geom defines the geometry of the section, sectforce performs sectional analysis,
contourcurv calculates contours of equal curvature of the section, momcurv calculates the axial force-
moment-curvature response of the section, deflection calculates column deflection curves of the
column, response calculates the load-deflection response of the column for a given eccentricity of load,
and capacity calculates the member capacity interaction diagram of the column. The function ultimate
returns the ultimate load Q*, corresponding deflection w*, and ultimate moment M* of the column for a
given eccentricity of load. Functions at any calculation stage in the program can be executed to obtain in-
termediate results. For example, a set of column deflection curves can be calculated using deflection giv-
en that the following input parameters are defined: stress-strain curve, sectional geometry, member length,
54
3. Slender Ultra High-Performance Fibre-Reinforced Concrete Columns
sectional
member
geometry & sectional member
response
material analysis deflection
& capacity
response
input parameters stress-strain curve stress-strain curve stress-strain curve stress-strain curve
sectional geometry sectional geometry sectional geometry sectional geometry
member length member length
applied load applied load
eccentricity of load
FUNCTIONS GEOM
superscripts refer CENTROID1
to the sample output SECTPROP
diagrams below
MATMODEL2 SECTFORCE3
GETSTRESS NMBOUND
MFACTOR CONTOURCURV4 MOMCURV5 RESPONSE7
ULTIMATE
DEFLECTION6 CAPACITY8
depth
ENVELOPES
plane stress
width of strain
element strip
moment
axial moment
force
1. CENTROID
curvature
axial load
3. SECTFORCE ultimate load
stress
5. MOMCURV
contour of
axial equal curvature
force 7. RESPONSE
strain + eccentricity
column
6. DEFLECTION
moment
8. CAPACITY
qult is written in matlab r2007a and is included with this thesis digitally as a set of matlab m-files.
Table B-1, which is in Appendix B, lists all the major functions used in qult and describes their input
55
3. Slender Ultra High-Performance Fibre-Reinforced Concrete Columns
Four eccentrically-loaded ultra high-performance fibre-reinforced concrete columns were tested at the
University of Toronto for the purpose of validating the general analysis method described in Section 3.2.
The primary objectives of this test program were (1) to study the effect of slenderness on ultimate load Q*,
(2) to assess the accuracy of the load-deflection response predicted by the general method, and (3) to ob-
serve the mode of failure of the columns. Details of the test specimens and connections are shown in Fig-
ure 3-8, along with their as-built measurements. The complete set of column drawings, including measure-
ments can be found in Appendix C. The pin-to-pin lengths of the specimens ranged from 1737 mm to 3466
56
3. Slender Ultra High-Performance Fibre-Reinforced Concrete Columns
Three 48 litre batches of ultra high-performance fibre-reinforced concrete were prepared according to the
mix design in Table 2-1 on page 17. Materials were mixed in a 0.17 m3 mortar mixer in the following order:
(1) the cement, silica fume, and sand were dry mixed until they were homogeneous, (2) the superplasti-
cizer and water (pre-mixed) were added and mixed until no dry materials remained, and (3) the steel fibres
were mixed in. The mixing process took 5-10 minutes per batch.
The finished batches were poured into a concrete bucket and then placed directly into the horizontally ori-
ented forms as shown in Figure 3-9. Specimens A and B each had their own batch, while Specimens C and
D shared the same batch. Because the concrete is self-consolidating, no vibration was done during place-
ment. The specimens and cylinders were left in the laboratory at approximately 20°C and moist-cured for 7
days after concrete was placed. The average compressive strengths fc! of the cylinders on the days of
The desired eccentricity of load was built into steel end plates as shown in Figure 3-8. Welded steel studs
were offset 10 mm from the centre of each plate. After the studs were manually aligned with the centroid of
the form void, the steel end plates were clamped to the plywood form (see Figure 3-10), which ensured a
57
3. Slender Ultra High-Performance Fibre-Reinforced Concrete Columns
Figure 3-10. Clamped steel end plate (left), spherical bearing at base (right)
Spherical bearings were built into receiving steel plates at the two ends of the column, allowing for rotation
about all axes (see Figure 3-10). This eliminated any directional bias of the column deflection, except for
that of the eccentricity of load. Friction between the bearing and pins were not measured and were not ac-
A few days before testing, 60 mm surface strain gauges were installed along the centreline of the flexural
compression face of each specimen at a centre-to-centre spacing of 300 mm. Along the centreline of one
adjacent face, 6 mm tooling balls targets were glued to the column at a spacing of 150 mm. Tooling ball tar-
gets were also glued to the centre of each pin. A metris cohesive laser radar scanner was used to measure
the three spatial coordinates of each tooling ball with respect to a predefined frame of reference. With
these measurements, column deflection curves w(x) could be recorded directly at any given load stage.
In order to prevent premature concrete cracking, forms were designed to be stiff enough to support the
columns during their transportation, erection, and installation into the test machine (see Figure 3-11).
Forms were removed after each column was pinned to the machine head. A linear variable differential
transformer was installed at the middle of the column to measure mid-length deflections of the column in
58
3. Slender Ultra High-Performance Fibre-Reinforced Concrete Columns
3000
top pin
2500
top pin
2000
1500
12.3 mm A Point A marks the
mid-length target
of each column
A 8.8 mm
1000
A 5.3 mm
500
all measurements
taken at zero load
bottom pin at (0,0) 0
0 5 10 15 20 0 5 10 15 20 0 5 10 15 20 0 5 10 15 20
Horizontal distance of centreline targets from a line passing through the top and bottom pin targets in mm
Figure 3-12. Position of cohesive laser radar scanner targets at zero load
For each column, initial readings of all tooling ball targets were taken before load was applied. Figure 3-12
shows the position of each target on the column relative to the line of action of load, which is taken as a
59
3. Slender Ultra High-Performance Fibre-Reinforced Concrete Columns
line passing through the top and bottom pin targets. For each column, the as-built initial eccentricity of
load was measured as the horizontal distance between the middle target (Point A) and the line of action of
load. The error in manually positioning each target using a ruler is assumed to be ±1 mm. The estimated
margin of error in measuring the initial eccentricity of load is thus ±2 mm. Pin-to-pin length was taken as
the vertical distance between the top and bottom pin targets.
The nominal offset of 10 mm (built into the steel end plates) was chosen to produce a load-deflection re-
sponse in which second-order effects could be observed. The resulting as-built eccentricities of load varied
between 5.3 mm and 12.3 mm. The variations are not ideal, but are acceptable because they can be accoun-
ted for in general method calculations. Increasing the overall scale of the specimens to reduce the effects of
construction error was not possible due to restrictions on the height clearance of the testing machine.
A large-scale universal mts mobile testing machine with a load capacity of 2650 kN was used to exert a
downward load on the top pin of each specimen. The load tests were displacement-controlled using a com-
puter-controlled, electro-hydraulic servo testing system. Specimens were loaded in 10 to 15 stages. At each
load stage, the machine displacement was held so that the scanner could measure the position of each tar-
get. Because the tests were vertical displacement-controlled, the columns would slowly lose load and
deflect while scanner measurements were being taken. Figure 3-13 shows the effect of this creep behaviour
on the load-deflection response of Specimen C between load stages. The plots of tangent slopes dQ/dw and
secant slopes Q/w reveal that the stiffness fluctuations caused by the discontinuous application of load did
not adversely affect the overall active load-deflection response of the column.
Due to safety concerns, close examination of the columns during the post-peak load stages of the test was
not possible. At failure, each specimen developed a full depth crack within 300 mm of the middle of the
column. The failure surface of Specimen C is shown in Figure 3-14. Fibres were left protruding from the
failure surface. The concrete matrix that had held these fibres together had disintegrated into powder. Ex-
cept for the failure crack, all other cracks that had formed in the specimen were too small to be seen.
60
3. Slender Ultra High-Performance Fibre-Reinforced Concrete Columns
e 400
w
200
Q 0
w
100 spikes correspond to
Tangent slope creeping of column
dQ/dw in kN/mm between load stages
Q tan shape of curve with
50 continuous loading
–50
w
Secant slope 60
fluctuation of secant
Q/w in kN/mm slope between load
stages are minimal
c
se
Q 40
w
20
0
0 10 20 30
Horizontal deflection of
column at mid-length w in mm
61
3. Slender Ultra High-Performance Fibre-Reinforced Concrete Columns
Load-deflection responses of all specimens are shown in Figure 3-15 along with photographs of the
columns at different load stages. Load stages are marked on the load-deflection plot with small solid
circles. Load stage 0 refers to the unloaded state of the column. At each subsequent load stage, photo-
graphs were taken each time the vertical displacement of the column was held to allow for scanner meas-
urements. This figure presents a visual record of the deflected shape of the column deflection w(x) as a fun-
ction of applied load Q. The lines superimposed on the photographs indicate the location of the mid-
length section of the column and the line of action of load, defined by a line passing through the centre of
the top and bottom pins. These lines create a frame of reference that can help the reader compare subtle
changes in the deflection of the column. The last photograph in each set shows the location of the full
depth crack in each failed column, except for Specimen B for which a photograph at failure was not taken.
The displacement measurements taken by the laser radar scanner at each load stage were fitted to 4th order
polynomial equations w(x). These polynomials were then differentiated twice to obtain approximate ex-
pressions for curvature ϕ(x) along the column. These curvatures ϕ along with the surface strain gauge
measurements on the flexural compression face of the column were used to calculate the planes of strain of
several cross-sections along the column. Strains on the flexural tension face %tf of the column were calcu-
lated according to Equation 3-23, where %cf is strain of the flexural compression face and d is the depth
62
3. Slender Ultra High-Performance Fibre-Reinforced Concrete Columns
axial load
Q in kN
800
600
400
200
w*
0
0 20 40 60
deflection w at
mid-length in mm
600
400 mid-length
200
w*
w
0
0 20 40 60
600
400
w*
200
w
0
0 20 40 60
600
w*
400
200
w
0
0 20 40 60
Figure 3-15. Load-deflection response of specimens and photographs at each load stage
63
3. Slender Ultra High-Performance Fibre-Reinforced Concrete Columns
Figure 3-16 presents all key measurements recorded during each load test. From top to bottom, the follow-
ing data sets are shown for each column load test: (1) photographs at the unloaded, ultimate, and failed
states, (2) as-built measurements, (3) ultimate loads Q*, deflections w*, and moments M*, (4) load-deflec-
tion Q-w response, (5) axial force-moment N-M response of the mid-length section, (6) strain of the flex-
ural tension face %tf and strain of the flexural compression faces %cf of the mid-length section versus mo-
The effect of slenderness on ultimate load can be observed from the load-deflection diagrams in Figure 3-
16. For columns with the same initial eccentricity of load e, the ultimate load of the column decreases as
column slenderness is increased. Eccentricity of load and sectional geometry also affect the value of ulti-
mate load according to the analytical model developed in Section 3.2. These parameters were not held per-
fectly constant in the load tests due to construction tolerances, but are at least close in value.
Observations on the behaviour of the mid-length section can be made by comparing the bottom three sets
of plots in Figure 3-16. The dashed lines that link the bottom three plots mark the value of the observed
ultimate moment M* for each load test. On the diagrams of extreme fibre strains versus moment, the
shaded regions represent the typical elastic stress range based on the material model developed in Chapter
2, using a compressive strength of 131 MPa. It can be seen that the mid-length sections behave inelastically
(data points move outside the shaded area) before the ultimate loads of the columns have been reached.
Thus, it would be unconservative to assume a purely elastic stress-strain response to simplify the calcula-
64
3. Slender Ultra High-Performance Fibre-Reinforced Concrete Columns
e Q=0 Q=Q*
w
M Q=0 Q=Q* Fail
N
mid-length
Q Q=0 Q=Q* Fail
Specimen A B C D
mid-length eccentricity in mm 9.4 12.3 8.8 5.3
mid-length depth in mm 102.2 101.7 102.5 102.6
pin-to-pin length in mm 3466 2892 2317 1737
compressive strength in MPa 132.1 116.1 131.3 131.3
Results
ultimate load Q* in kN 205 258 393 661
ultimate deflection w* in mm 24.3 21.1 21.0 10.0
ultimate moment M* in kN·m 6.89 8.64 11.7 10.1
400
typical elastic stress
range is shaded , 200
inelastic response
occurs before M* M M M M
0
M* M* M* M*
extreme fibre strains 2 ! ! ! !
at mid-length in 10-3 0
M M M M
strain on flexural
tension face of column –2 !!depth
strain on flexural com-
pression face of column –4
M M M M
0
curvature ! at mid-
length in rad/km 10
EI*=0.77·EI 0.82·EI
curvature based on 20 0.75·EI
laser scanner data 1
30 1 0.64·EI 1
EI is taken as the bending
stiffness of the uncracked 40
section " " " 1 "
50
moment M at 0 5 10 15 0 5 10 15 0 5 10 15 0 5 10 15
mid-length in kN·m apparent drop in stiff- secant stiffness at
ness after ultimate load ultimate load EI*
65
3. Slender Ultra High-Performance Fibre-Reinforced Concrete Columns
The moment-curvature diagram in Figure 3-16 for Specimen D (bottom-right plot) shows a bending mo-
ment of 1.8 kN·m and a nearly zero curvature at the first load stage. At this stage, the linear variable differ-
ential transformer measured a displacement of 1.03 mm, while the cohesive laser radar scanner measured a
displacement of 0.31 mm at the middle target of the column. The scanner measurement, which is used in
the calculation of curvature, may be the source of the uncharacteristically small curvature at the first load
stage. Another possible source of error is friction at the pins, which would cause some rotational fixity at
the ends of the column. If friction was present, the resultant line of action of load would be shifted toward
the centroidal axis of the column, resulting in much less curvature, as was observed during the test. Sub-
sequent load stages had much larger mid-length curvatures at higher loads, similar to those observed in
the other load tests. Measurements between the scanner and transformer agreed well after the first load
stage, suggesting that the displacement error did not continue past the first load stage.
The curvature-moment ϕ-M diagrams in Figure 3-16 are similar to conventional moment-curvature dia-
grams except that axial force N is not held constant in each diagram. Secant stiffnesses at ultimate limit
state EI* are shown on the diagram as secant slopes and are expressed in terms of the initial uncracked
flexural rigidity of the column EI0. The lowest recorded secant stiffness EI* was 0.64EI0, which gives furth-
er evidence that the column was well into its nonlinear stress range at ultimate limit state. Specimens A
and B exhibited a distinct drop in secant stiffness after their ultimate loads were reached. This was prob-
ably true for Specimens C and D, but no curvature data was recorded after their ultimate loads were
reached.
The input parameters required by the general analysis method are column length, geometry of section, ec-
centricity of load, and stress-strain curve. Pin-to-pin column length and eccentricity of load at mid-length
were measured by the cohesive laser radar scanner from the zero-displacement scans described in Section
3.4.2. Average depth and average width of section were taken as the average measured value along the
length of column as shown in Figure C-8 on page 314. Compressive strengths of each specimen were based
66
3. Slender Ultra High-Performance Fibre-Reinforced Concrete Columns
on the average strength of cylinders that were cast from the same concrete batch, cured under the same
conditions, and tested at the same age as their respective specimens. The elastic modulus E for all speci-
mens was taken as 46.7 GPa, as determined by the average elastic modulus of the three prism tests de-
scribed in Section 2.1.4. The average elastic modulus of 48.1 GPa obtained from the 28-day cylinder tests
was not used because the prisms tested were more similar to the column tests in terms of age of concrete
and stress gradients. The linearity ratio between the secant modulus at peak stress Ec! and elastic modulus
Ec was taken as 1.3 based on the cylinder test results in Table 2-2. From these material properties, a stress-
strain curve based on the material model proposed in Chapter 2 was used as input for the general method.
Specimen | Model A A B B C C D D
mid-length eccentricity in mm 9.4 9.4 12.3 12.3 8.8 8.8 5.3 5.3
average depth in mm 101.9 101.9 101.4 101.4 102.4 102.4 101.3 101.3
average width in mm 103.2 103.2 103.0 103.0 101.7 101.7 102.9 102.9
pin-to-pin length in mm 3466 3466 2892 2892 2317 2317 1737 1737
compressive strength in MPa 132.1 132.1 116.1 116.1. 131.3 131.3 131.3 131.3
elastic modulus in GPa 46.7 46.7 46.7 46.7 46.7 46.7 46.7 46.7
percent
Results | Prediction difference
ultimate load Q* in kN 205 218 +6.1% 258 260 +0.8% 393 450 +14.5% 661 728 +10.1
ultimate deflection w* in mm 24.3 25.1 +3.3% 21.1 23.5 +11.4% 21.0 18.4 -12.4% 10.0 13.3 +33.0
ultimate moment M* in kN·m 6.89 7.51 +9.0% 8.64 9.31 +7.8% 11.7 12.2 + 4.3% 10.1 13.5 +33.7
w 400
† w*
†
200 w*
w*
Q deflection 0
w w w w
w in mm 0 20 40 60 0 20 40 60 0 20 40 60 0 20 40 60
‡
Q axial force 800
N N N N
N in kN
600 M*
‡
400
M M*
N
200 M*
M*
M M M M
moment 0
M in kN·m 0 5 10 15 0 5 10 15 0 5 10 15 0 5 10 15
Figure 3-17. Comparison of load test results and general method predictions
67
3. Slender Ultra High-Performance Fibre-Reinforced Concrete Columns
Figure 3-17 shows a comparison between load test results and general method predictions. The table at the
top of the figure shows the measured values that were used as input for the general method. Percent differ-
ences between actual and predicted ultimate loads Q* of Specimens A, B, C and D were +6.1%, +0.8%,
+14.5%, and +10.1%, respectively. Hence, all predictions moderately overestimated the observed ultimate
loads of the column. The ratio between assumed error and measured initial eccentricity ranges between
16.2% and 37.7%. Thus, considering the ±2 mm error in measuring initial eccentricities of load, the general
The load-deflection Q-w and axial force-moment N-M diagrams shown in Figure 3-17 include both ob-
served test results and general method predictions. The sensitivity of the ±1 mm error associated with the
manual positioning of the tooling balls is captured by two boundary prediction curves. These curves use
initial eccentricity of load values of 2 mm greater than and 2 mm less than the measured value. Test results
of all specimens, except for Specimen C, stay within the error bounds. Overall, the predictions and experi-
mental results agree reasonably well, attesting to the validity of the analytical model described by the gen-
eral method.
The measured deflections beyond the ultimate load of Specimens A, B, and C extend further than the
those predicted by the general method. This discrepancy may be accounted for by the tensile ductility of
the material beyond peak tensile stress. This post-peak tensile ductility is neglected in the stress-strain
models used for calculation. Neglecting this ductility is thus conservative in terms of estimating deflection
capacities of slender columns. Further, this omission does not appear to adversely affect the accuracy of
Specimen D failed prematurely according to the response predicted by the general method. One possible
explanation is that there was a zone of weakness in the column where it failed, about 200 mm above the
middle of the column. Throughout the load test of Specimen D, the strain gauge at mid-length recorded
compressive strains smaller in magnitude than those above and below it. This means the section at mid-
length was not the critical section. The load-deflection response of Specimen D shows an abrupt change in
68
3. Slender Ultra High-Performance Fibre-Reinforced Concrete Columns
stiffness at 150 kN. This suggests that there may have been some non-uniformity in the column or some
friction at the pins, neither of which were accounted for in general method calculations.
The general method presented in Section 3.2 requires a large amount of computation, and thus cannot be
done by hand in a reasonable amount of time. General method calculations are only practical with the use
of a computer, as done in the program qult written by the author. In order to reduce the analytical effort
required by the general method, approximate simplified methods for designing slender ultra high-per-
The objective of the simplified design method is to determine whether a given slender column is capable of
sustaining the combined axial compression and bending caused by a given load. Given the following input
parameters: length of column, sectional geometry, and stress-strain curve, an approximate member capa-
city N-M* interaction diagram is produced. This interaction diagram serves as an envelope for all com-
binations of axial force and moment that can safely be carried by the critical section of the member.
The proposed simplified method is inspired by a simple and elegant design method proposed by Menn
(1990) for the design of slender reinforced concrete members. Menn uses a reduced N-M interaction dia-
gram based on a sectional analysis that is limited by the yield strain of steel. Designing slender members
using this reduced interaction diagram ensures that curvatures and second-order deflections remain small.
Menn’s simplified design method is particularly useful because it is simple, concise, and can be calculated
by hand. Menn’s method will be reviewed in Section 3.5.1 and then used as a model for the simplified
Menn’s design method (1990) is best described within the context of an example. The hollow, reinforced
concrete box section shown in Figure 3-18 will be used to illustrate Menn’s design method. A concrete
compressive strength of 30 MPa is assumed. The box is lightly reinforced with two layers of fifty-five 10M
69
3. Slender Ultra High-Performance Fibre-Reinforced Concrete Columns
bars in each of the top and bottom slabs, corresponding to top and bottom reinforcement ratios ρs! and ρs
of 0.15%. Various material and section properties are given in Figure 3-18.
30
400
concrete in compression
A = 7.36 m2
steel I = 17.8 m4 fs
reinforcement r = 1.55 m 400
As = 0.011 m2 E = 25100 MPa
!s = 0.15% EIo = 447000 MN·m2
400
Sectional capacities of reinforced concrete sections are typically calculated based on the following modes
of failure: (1) the crushing of concrete in compression at a strain of -0.0035, and (2) the rupturing of rein-
forcing steel at a strain of about +0.010. Straining reinforced concrete members to these limits tend to sig-
nificantly reduce the stiffness of the system and cause large plastic deformations. Slender members subjec-
ted to axial compression are particularly sensitive to these effects and can reach their ultimate load at
strains much lower than these conventional strain limits. Thus, in many cases, it is unconservative to rely
upon the full moment capacity of the section, as calculated by these strain limits. As previously discussed
in this chapter, the member capacities of slender members are less than their sectional capacities.
To reduce the severity of second-order effects in slender members, Menn proposes that sectional capacity
be calculated based on the initial yielding of reinforcing steel. The yield strain of conventional reinforcing
steel is typically ±0.002, which is significantly less than the crushing strain of concrete or the rupture strain
of steel. The ranges of admissible planes of strain based on conventional material failure strain limits and
based on Menn’s yield strain limits are illustrated in Figure 3-19. The figure shows that curvature and axial
strain are significantly reduced when the lower yield strain limits are imposed.
70
3. Slender Ultra High-Performance Fibre-Reinforced Concrete Columns
Figure 3-19. Conventional strain limits and Menn’s reduced strain limits
Menn’s method is intended for solving specific design problems related to the use of slender members in
bridges, particularly in the design of piers and arches. The choice of yield strain limits is foremost a design
choice intended to limit deflections and second order-effects. The reduced strain limits effectively limit the
moment capacity of the section from that of the sectional moment of resistance Mu to the yield moment of
the section My. In most practical slender member designs, the ultimate moment M*, as defined and de-
scribed in Section 3.1, occurs somewhere between the two moments: My<M*<Mu. Because the calculation
of M* is complex, My is used as a conservative estimate of the ultimate moment capacity of the member.
The difference between My and Mu, which affects the degree of conservatism, varies with the shape of the
cross-section and the arrangement of longitudinal steel. The additional moment capacity gained by strain-
ing the section beyond the yield strain of steel is typically small. Once the steel has yielded, the moment
resistance of the section can be increased by increasing the lever arm between the resultant force of rein-
forcing steel in tension and the resultant force of concrete in compression. For hollow box sections, the
resultant compression force is typically confined within the thickness of the top slab. Since the thickness of
the top slab is very small relative to the overall depth of member, only small increases in lever arm and
moment capacity are possible. Thus, the differences between My, M*, and Mu are small, as is the degree of
conservatism. For solid sections and for sections with multiple layers of reinforcing steel spread over a sig-
nificant portion of the depth, the degree of conservatism is higher, since the difference between My and Mu
is larger. This is because there is (1) a greater capacity for the lever arm to increase after the reinforcing
71
3. Slender Ultra High-Performance Fibre-Reinforced Concrete Columns
steel yields, and (2) the inside layers of steel reinforcing have not yet reached their yield strength. This
means that in some cases, Menn’s design method may lead to overly conservative designs. However,
without a simple way to calculate the ultimate moment M*, there is no choice but to use My as an estimate
of the moment capacity of the member to ensure that the design is satisfactory. If designers are unsatisfied
with restricting the capacity of their slender member to My, then they must resort to more complex analyt-
Post-yield increases in moment resistance can also be attributed to increases in stress in the reinforcing
steel due to strain-hardening. Accounting for strain-hardening using manual calculation, however, is com-
plex. Hence, the over-strength provided by strain-hardening is normally neglected in design, and is instead
N M
150
!200
My Mu
Mcr
!150
!100
N-My 0
N-Mu !
!cr !y !u
!50 N=-55.2 MN
EI10 EIcr
0
0 50 100 150 M EIy
EIu
0
N-Mu interaction diagram based 0 0.5 1 1.5 !
on material failure strain limits
Secant stiffness-curvature diagram
N-My interaction diagram based on in EI0 and rad/km N=-55.2 MN
Menn's reduced yield strain limits
(a) (b)
Figure 3-20. Interaction and moment-curvature diagrams for a 30 MPa concrete box
Results from a sectional analysis of the example reinforced concrete box section are shown in Figure 3-20a.
Two interaction diagrams are shown, one based on conventional strain limits (dashed N-Mu curve) and
one based on yield-strain limits (solid N-My curve). For the given section, the difference between the two
72
3. Slender Ultra High-Performance Fibre-Reinforced Concrete Columns
interaction diagrams are minimal. This means that for any value of axial force N there is very little penalty
in using yield moments My as design moments of resistance rather than using ultimate moments Mu.
The behaviour of the reinforced concrete box section is further explored in Figure 3-20b, which shows the
moment-curvature diagram of the section for N=–55.2 MN. According to this diagram, the section has
three distinct domains of behaviour. The transitions between domains are marked by: the initial cracking
of concrete at the flexural tensile face of the section (Mcr), the yielding of the bottom layer of steel rein-
forcement in tension (My), and the rupture of steel or crushing of concrete (Mu). Beyond the yield moment
The softening of the concrete box section is illustrated by the secant stiffness-curvature diagram shown in
Figure 3-20b. Secant stiffness EI! is calculated by dividing each value of moment M on the moment-
curvature diagram by its corresponding value of curvature ϕ. Between zero curvature and the curvature at
cracking ,cr, the section has a secant stiffness higher than the gross flexural rigidity EI0 of the section be-
cause of the stiffness contributed by the reinforcing steel. After cracking, the section loses stiffness, with
secant stiffness dropping down to about 33% of EI0, at which point reinforcing steel begins to yield. This
value is called the secant stiffness at yield EIy. Beyond the yield condition, secant stiffness continues to
drop as the reinforcing steel continues yielding until sectional failure is reached.
The secant stiffness at yield EIy is an important quantity because it represents a lower limit of secant stiff-
ness for members designed according to Menn’s reduced N-My interaction diagram. Provided that moment
demands at ultimate limit states are less than the yield moment, then it is always conservative to assume
EIy as the effective stiffness of the member being designed. Once calculated, this effective stiffness can be
used in the calculation of defections, buckling resistances, and moments caused by restrained deformation.
In lieu of detailed calculations, EIy can roughly be taken as one-quarter of the gross flexural rigidity of the
EI y ! 41 EI 0 Equation 3-24
73
3. Slender Ultra High-Performance Fibre-Reinforced Concrete Columns
The accuracy of this estimate of EIy varies with axial force N, as shown by the axial force-secant stiffness
diagram of the 30 MPa concrete section in Figure 3-21b. Three curves are shown corresponding to secant
stiffnesses at cracking, yielding, and ultimate (calculated assuming no strain-hardening). The EIy curve
shows that detailed calculations of EIy give values that are close to "EI0. For this section, it is unconservat-
N N
!200
!150
!100 EIcr
Mcr
EIy
!50
My
EIu
0
0 50 100 150 M 0 !EI 10
EI EI!
0
(a) (b)
With means of calculating effective secant stiffness EIy and reduced N-My interaction diagrams, Menn’s
simplified design method can now be fully described. The method proceeds as follows:
Step 1. A reduced N-My design interaction diagram is calculated for the trial section based on planes
of strain that are limited to the yield strain of steel %sy at the outer layers of steel reinforcing (Figure 3-19).
Step 2. Axial force demand N is calculated based on the applied design loads. Based on this axial
force, yield moment My and yield curvature ,y is calculated using sectional analysis. From these values,
Step 3. Second-order deflections and moments are calculated using Vianello’s method using Equa-
tions 3-2 and 3-3 on page 34. In the calculation of deflections, moments, and buckling resistance, flexural
74
3. Slender Ultra High-Performance Fibre-Reinforced Concrete Columns
Step 4. The adequacy of the slender member is determined by plotting axial force-moment demands
onto the reduced N-My design interaction diagram. If the points lie outside the interaction diagram, then a
Menn’s simplified design method is validated experimentally by load tests of reinforced concrete beams in
compression and bending. Original work is reported in a university report by Gruber and Menn (1978).
This section describes a simple method for designing slender members made from ultra high-performance
fibre-reinforced concrete. The method is inspired by Menn’s design approach, and involves using a reduced
interaction diagram for member capacity, and also an effective secant stiffness in the calculation of second-
order moments. The key difference that prevents the direct application of Menn’s method to ultra high-
performance fibre-reinforced concrete members without conventional steel reinforcing is that there is no
distinct yield point to use as a strain limit. An alternative means of defining a reduced member capacity in-
As done for Menn’s method, the proposed simplified design method will be described within the context
of an example. The ultra high-performance fibre-reinforced concrete box considered is shown in Figure 3-
22. The outside dimensions of the box are the same as the conventionally reinforced concrete box analyzed
in the previous section. The strength of concrete assumed is 120 MPa, which is four times the strength of
the concrete in the other box. The webs and slabs of the higher strength box have specified thicknesses of
100 mm. The thicknesses are chosen such that the product of cross-sectional area and concrete strength
are roughly equal among the high-strength and conventional-strength boxes. This improves the validity of
75
3. Slender Ultra High-Performance Fibre-Reinforced Concrete Columns
100
84
A = 1.96 m2 ft
I = 5.48 m4
r = 1.67 m
E = 42000 MPa 14.6
EI0 = 230000 MN·m2
100
7
Ultra high-performance
fibre-reinforced concrete section 0.15 1.5 !t
concrete in tension
Moment-Curvature
in MN·m and rad/km N=-55.2 MN
Axial force- Axial force-
Moment interaction Secant stiffness M
in MN and MN·m in MN and EI0
150
Mu
N N
Mcr
post-peak response
!200 depends on the control
of localized cracking
EIu
!150
0
"cr "u "
!100 EIcr
EI10 EIcr
Mcr
Mu EIu
!50
0
0 0 0.5 1 1.5 "
0 50 100 150 M 0 10
EI EI!
!EI0
Secant stiffness-Curvature
(a) (b) in EI0 and rad/km N=-55.2 MN
(c)
The ultra high-performance fibre-reinforced concrete box section considered has three domains of beha-
viour, as shown by the moment-curvature diagram in Figure 3-23c. The transitions between domains are
marked by: initial cracking of concrete at the flexural tensile face of the section (Mcr) and the localization
of cracks after strain-hardening (Mu). The moment at crack localization is assigned to be the ultimate mo-
76
3. Slender Ultra High-Performance Fibre-Reinforced Concrete Columns
ment Mu because (1) the post-peak tensile stress-strain behaviour of the material is not precisely known,
and (2) any increase in moment capacity after crack localization will be at best, minimal. A distinct yield
moment for this section cannot be characterized because the fibre reinforcement embedded in the materi-
The amount of post-peak ductility (often measured by the area under the moment-curvature diagram)
depends on the ability of the embedded steel fibres to bridge the the localized crack while being pulled out
from the concrete matrix. Volume and orientation of steel fibres are thus the most important factors that
affect post-peak ductility. The ductility of the ultra high-performance fibre-reinforced concrete mix used
in this thesis can be indirectly observed from the load-deflection results of the prism tests shown in Figure
2-8 on page 26. The post-peak curves of two out of the three prisms show gradual decreases in load with
increasing deflection. This type of ductility behaviour is only exhibited if the test is deflection-controlled.
The secant stiffness-curvature behaviour of the 120 MPa concrete section is shown below the moment-
curvature diagram in Figure 3-23c. The diagram shows that the section has relatively high stiffness at ulti-
mate limit state. For the given axial force N=–55.2 MN, the section maintains 68% of the gross flexural ri-
gidity of the section, which is much higher than the 30 MPa concrete section based on percentage of EI0.
Comparing the actual values of secant stiffness, the 120 MPa concrete section has EIu=156000 MN·m2 at
ultimate and the 30 MPa concrete section has EIy=148000 MN·m2 at yield. Thus, even though the 120 MPa
concrete section is about four times lighter and thinner than the 30 MPa concrete box section, both sec-
tions have effective secant stiffnesses at the design limit state that are more or less the same.
As shown in Figure 3-23b, secant stiffnesses of the 120 MPa box section vary with axial force N. For most
values of N below the balanced condition, secant stiffness at cracking EIcr is roughly equal to the gross flex-
ural rigidity of the section EI0. Thus, if the demand moment is less than the cracking moment M<Mcr, then
it is reasonable to assume that the design secant stiffness is equal to EI0. This design criterion is most ap-
propriate for serviceability limit state calculations, for which it is favourable to have the structure remain
crack-free. For most values of N, secant stiffness at ultimate EIu ranges between 50% and 75% of EI0. If the
77
3. Slender Ultra High-Performance Fibre-Reinforced Concrete Columns
demand moment is less than the ultimate moment M<Mu, then it is conservative to assume that the design
secant stiffness is EIu. In lieu of more detailed calculations, EIu can be estimated using Equation 3-25,
where N is axial force, and N! is maximum force sustained by the section in pure axial compression:
+ # N &
- 1
EI 0 % + 1( 0 ) N ) 12 N "
-
2
$ N" '
EI u ! , Equation 3-25
- # N&
1
EI 0 % 2 * ( 1
N" ) N ) N"
- 2
$ N"'
2
.
Estimates of EIu using Equation 3-25 were found to be in close agreement with detailed calculations of EIu
for many different ultra high-performance fibre-reinforced concrete sections that are symmetric about the
axis of bending. These estimates are considered to be sufficiently accurate for preliminary design.
The N-Mu interaction diagram shown in Figure 3-23a is not a reduced interaction diagram, and is thus un-
conservative for slender members whose ultimate loads Q* are reached before sectional failure (as exhib-
ited by the columns tested in Section 3.4). A strain-limited, reduced interaction diagram is not possible be-
cause a distinct yield point does not exist. Instead, a reduced interaction diagram is proposed as a
conservative approximation of the member capacity N-M* interaction diagrams produced by the general
method. The proposed reduced interaction diagram is shown in Figure 3-24 in terms of normalized axial
78
3. Slender Ultra High-Performance Fibre-Reinforced Concrete Columns
0.5
n —n
0
n 0.5 1 1.5
n
m
mn=
Normalized axial force n and moment m are given by Equations 3-26 and 3-27, where A is cross sectional
area, fc! is compressive strength of concrete, d is depth of section, and I is moment of inertia of the un-
cracked section:
N 1
n= ! Equation 3-26
A fc"
Md 1
m= ! Equation 3-27
I fc"
The outside boundary of the lightly shaded area in Figure 3-24 represents the proposed interaction dia-
gram for serviceability limit states. The interaction diagram is defined by three coordinates: (m=0, n=0.5),
(m=1, n=0.5), and (m=0, n=–ncr).The normalized axial force at cracking ncr is equal to the cracking
strength of concrete fcr divided by the compressive strength of concrete fc!. The upper horizontal boundary
ensures that stresses in the slender member remain in the linear range of the compressive stress-strain
curve. The lower diagonal boundary approximates the contour of cracking, which is a set of sectional
forces that correspond to the initial cracking of the flexural tensile face of the section. These limits ensure
that the member remains in the linear-elastic range under service loads, and that secant stiffness is equal
79
3. Slender Ultra High-Performance Fibre-Reinforced Concrete Columns
to or slightly less than the gross flexural rigidity of the section. These limits also ensure that some nominal
The outside boundary of the darkly shaded area in Figure 3-24 represents the proposed reduced interac-
tion diagram for ultimate limit states. The interaction diagram is defined by five coordinates: (m=0, n=nE),
(m=1, n=0.5), (m=1, n=0.5–(nt–ncr)), (m=mN=0, n=0), and (m=0, n=–nt). The normalized axial force at
Euler buckling nE is equal to the buckling resistance of the member divided by the maximum axial force of
the section in pure compression: QE/N!, or equal to 0.8, whichever is less. The normalized axial force at
peak tensile stress nt is equal to the tensile strength of concrete ft! divided by the compressive strength of
concrete fc!. The normalized moment resistance in pure flexure mN=0 (zero axial force) is calculated as the
The proposed reduced n-m interaction diagram is validated in Section 6.3.4. This section discusses the sec-
tional force demands and capacities of the concept arches that were designed as part of the parametric
design study. In Figure 6-19 on page 242, the member capacity N-M* interaction diagrams for 72 concept
arch members are displayed and superimposed on to the proposed reduced interaction diagram. These 72
interaction diagrams are calculated based on trial arch sections that were found to be satisfactory under
the imposed bridge loads. Each of the 72 concepts were designed for a unique set of span length, span-to-
rise ratio, and degree of deck-stiffening. For the arch cross-sections considered, the proposed reduced n-m
interaction diagram was found to be conservative and accurate in comparison to the member capacity N-
Only hollow and solid-filled, rectangular ultra high-performance fibre-reinforced concrete sections were
considered in the parametric design study. The reduced interaction diagrams for these two types of section
differ in terms of their degree of conservatism. Reduced interaction diagrams for hollow sections tend to
follow the diagrams calculated using the general method reasonably well, resulting in a nominal degree of
conservatism. As the walls of the hollow section are made inwardly thicker, the degree of conservatism in-
80
3. Slender Ultra High-Performance Fibre-Reinforced Concrete Columns
creases. Thus, solid-filled sections tend to have actual ultimate moment M* capacities that are moderately
With means of calculating effective secant stiffness EIu and reduced n-m interaction diagrams, the pro-
posed simplified design method for slender ultra high-performance fibre-reinforced concrete members
can now be fully described. For the design at ultimate limit states, the method proceeds as follows:
Step 1. A reduced N-M* design interaction diagram for the trial section is calculated based on the
Step 2. Axial force demand N is calculated based on the applied design loads. Based on this axial
Step 3. Second-order deflections and moments are calculated using Vianello’s method using Equa-
tions 3-2 and 3-3 on page 34. In the calculation of deflections, moments, and buckling resistance, flexural
Step 4. The adequacy of the slender member is determined by plotting axial force-moment demands
onto the reduced N-M* design interaction diagram. If the points lie outside the interaction diagram, then a
81
Chapter 4. Statical Analysis of Arch Bridges
In this chapter, the statical analysis of concrete arch bridges is presented. The analytical methods described
in this chapter give estimates of maximum sectional forces in the structural system that are sufficiently ac-
curate for preliminary design. Further, this chapter examines how slenderness, shallowness, and distribu-
tions of flexural rigidity affect the structural behaviour of arches. Insights drawn from these studies along
with the proposed analytical methods are used to validate new design concepts for concrete arch bridges in
Chapter 6.
Early literature on the statical analysis of arch bridges includes work by Ritter (1877) and Culmann and
Ritter (1906). The former is a paper that discusses stiffening trusses in arch and suspension bridges, while
Robert Maillart, a student of Ritter, was one of the early designers of reinforced concrete arch bridges. His
design methods are described by Maillart (1934), Billington (1979), and Laffranchi and Marti (1997). One
important contribution of Maillart’s is the deck-stiffened arch form, for which he assumed the deck girder
Classical methods of arch analysis neglect the deformations of the structure when formulating equilibrium
equations. These methods are suitable for designing and analyzing heavy arches since deflections tend to
82
4. Statical Analysis of Arch Bridges
be small in comparison to the depth of the arch ribs. As arches become more slender, considerable addi-
tional bending moments can be caused by second-order deflections of the arch. Deflection theory analysis
of arches, which accounts for arch displacements in formulating equations of equilibrium, was studied by
Melan and Steinman (1913) and later by Asplund (1963). From these formulations, approximate methods
were developed based on amplifying moments and deflections calculated from classical methods, similar
to those developed for straight, slender members subjected to compression and bending.
Slender arches are also sensitive to stability problems. Chapters on arch stability edited by Johnston (1976)
and Galambos (1998) give comprehensive reviews of the topic in the context of metal structures based on
studies by Gjelsvik and Bodner (1962), Schreyer and Masur (1966), Austin (1971), Dickie and Broughton
(1971), and Kennedy and Aggarwal (1971). Many of these studies on the stability of arches are based on
elastic analysis and thus can be applied to concrete structures with some adjustment for material
properties.
Of particular interest to the objectives of this thesis are studies on the buckling of shallow arches. In shal-
low arches, axial deformations under uniform load cause significant crown deflections. These deforma-
tions cause shear and bending stresses in the arch similar to those in beam-type structures. Thus, perman-
ent loads, which are normally carried in pure compression along the axis of the arch, are partially carried
by shallow arches in bending. Studies on the stability of shallow concrete arches have been done by Wang
et al. (2006) and Cai et al. (2009). The former work includes shrinkage and creep deformations in formu-
lating stability equations to describe the long-term buckling of shallow arches with hinged supports, while
the latter work deals with shallow arches with fixed supports.
Recent publications that review the analysis of concrete arches include works by Menn (1990), Mondorf
(2006), and Benaim (2007). On-going research in the area of arch bridges, including the analysis of arches,
is presented at the International Conference on Arch Bridges, which has been held every three years since
1995.
83
4. Statical Analysis of Arch Bridges
The arch analysis methods proposed in this chapter are based on the works briefly described above. Al-
though the proposed analysis methods do not introduce anything fundamentally different from current
literature, they do attempt to synthesize all the simplified analysis techniques from literature into one con-
sistent method for calculating second-order sectional forces in arch systems. The study of the transition of
shallow arches between pure arch behaviour and pure beam behaviour has not been previously solved or
explained using the force method to the extent presented by the author in Section 4.4.3. This chapter also
provides many original figures and diagrams that are designed to help readers improve their understand-
The main structural components and geometrical quantities of typical concrete arch bridges are illustrated
in Figure 4-1. The structural system consists of a deck girder, an arch, and spandrel columns. The supports
at each end of the arch are referred to as springing lines. The midspan section of the arch is referred to as
the arch crown. The arch span is the horizontal distance between springing lines, and the arch rise is the
vertical distance between the crown and a line intersecting the springing lines. The spans that the deck
Most concrete arches built to date are arranged with the arch supporting the bridge deck from below as il-
lustrated in Figure 4-1. Arches that suspend the deck from above usually require the arch to be split into
two arch ribs that rise from either side of the bridge deck. These ribs are often braced together above the
deck to provide lateral stability. Tension members are used instead of spandrel columns to transmit forces
from the deck to the arch. Arches in this arrangement are usually designed in structural steel or composite
84
4. Statical Analysis of Arch Bridges
concrete-filled steel tubes as in the Lupu Bridge in China (Lin et al. 2004). One exception to this is the re-
cently completed Third Millennium Bridge in Spain (Kremer 2008), in which the deck is supported from
above by a concrete arch. Although advanced construction techniques and high-strength concrete may al-
low for concrete arches to be built economically above the deck, the scope of this thesis is limited to the
Three types of arch systems are used in practice: three-hinged arches, two-hinged arches, and fixed arches.
As their classification names imply, the types of arches are distinguished by the number of hinges in the
system. Fixed arches have no hinges, two-hinged arches have hinges at each springing line, and three-
hinged arches have an additional hinge at the crown, as shown in Figure 4-2.
The relative differences between the three types of arch systems are described in Table 4-1. In broad terms,
these systems differ in terms of degree of indeterminacy, cost of construction and maintenance, flexibility
and stability, and bending moment demand. The degree of indeterminacy is highest in fixed arches, with
three redundant forces. While fixed arches offer the highest level of redundancy, which is favourable in
terms of structural reliability, fixed arches also require the most computational effort when calculating sec-
tional forces. Hinges in concrete structures require details that must be able to safely transmit high shear
and axial stresses while providing a rotational degree of freedom. Fabricating and maintaining mechanical
steel hinges can be costly due to their complexity and high exposure to the environment. Concrete hinges,
which are typically made by reducing the depth of section locally and providing sufficient steel reinforce-
ment, are less complex than mechanical hinges and can be constructed relatively cheaply.
85
4. Statical Analysis of Arch Bridges
Table 4-1. Relative comparison between fixed, two-hinged, and three-hinged arches
fixed two-hinged three-hinged
relative comparison arches arches arches
degree of indeterminacy 3 1 0
cost of construction and maintenance lower medium higher
magnitude of deflections lower medium higher
buckling strength higher medium lower
arch axis geometry same same same
moments caused by nonuniform loads lower medium higher
moments caused by restrained deformations higher lower none
Flexibility of the system is also determined by the fixity of the arch, with the highest flexibility associated
with three-hinged arches. Hence, deflections in hinged arches tend to be larger than those in fixed arches,
which make them more sensitive to long-term deflections. Higher flexibility also leads to higher susceptib-
ility to stability problems. Fixed arches have the highest resistance against flexural buckling, while three-
hinged arches have the lowest. Methods for calculating the buckling load of arches are discussed in the
next section.
Primary flexural buckling modes for each type of arch system are shown in Figure 4-3. These systems are
subjected to vertical, uniformly distributed loads, which cause pure axial compression along the axis of the
arch. As the applied loads are increased, the arches will eventually reach their critical load HE and buckle
according to the deflected shapes shown in the figure. Three-hinged arches may buckle in a symmetric,
snap-through mode or in an antisymmetric, bifurcation mode. Two-hinged and fixed arches buckle only
in antisymmetric, bifurcation modes. Analogous straight column buckling models can be constructed us-
ing the same member length and support conditions as the arch models they describe. Additional lateral
supports at mid-length are needed in the analogous column models to account for the two-wave primary
buckling modes that arches tend to exhibit. In these column models, the axes of the arches are straightened
out, and horizontal reactions of the arch are applied as concentric axial loads on the columns. Using this
86
4. Statical Analysis of Arch Bridges
analogy, the effective arc length factor k of each arch buckling model can be determined using convention-
al stability analysis of straight compression members. Values of k are given below (Austin 1971):
arch model
arch flexural EI
HE / 2 S f
buckling load kS 2
L
L2 £ 4 f
L
16 f ¥
2 2
1 2
arc length S L
16 f 2
ln ² ´ analogous column model
2 8f ¤ L ¦
S
initial shape
deflected shape
HE HE HE HE
k 0.54 k 0.50
HE HE HE HE
k 0.50 k 0.35
Figure 4-3. Arch and analogous column buckling models. Partially adapted from Austin (1971).
The buckling resistance of arches HE is given by the following equation, where EI is flexural rigidity of the
EI
HE = ! 2 Equation 4-2
( kS )2
87
4. Statical Analysis of Arch Bridges
Arc lengths of parabolic arches can be calculated as a function of span L and rise f using Equation 4-3. This
expression is based on the segment of a parabola, as given by Spiegel and Liu (1999).
L2 ! 4 f + L2 + 16 f 2 $
S= 1
2 L2 + 16 f 2 + ln # & Equation 4-3
8f " L %
The effective arc length factors given in Equation 4-1 can be verified by comparing their values with other
stability analyses of arches. Rambøll (1944) used a virtual work calculation method to calculate buckling
loads of three-hinged, two-hinged, and fixed arches. Results from Rambøll’s analysis are shown in Figure
4-4a as a plot of buckling coefficient ) versus rise-to-span ratio f/L. By reworking Rambøll’s buckling coe-
fficient results in terms of effective arc length factors k, the curves shown in Figure 4-4b are obtained. In
comparison to Rambøll’s analysis, the k values given in Equation 4-1 are accurate at low rise-to-span ratios,
but are slightly unconservative at rise-to-span ratios higher than 0.1. Other arch stability studies have been
done analytically by Dinnik (1955) and Stussi (1935), and experimentally by Gaber (1934) and Kolibrunner
(1936), all of which are in good agreement with the analogous column method (Austin 1971). All these
studies, however, neglect the effects of axial shortening, which can reduce buckling resistance considerably
HE = ! EI2 HE = "2 EI 2
L (kS)
! 90 k 1
0.8
3-hinged arch
60 2-hinged arch
0.6
fixed arch k=0.54 effective arc length
k=0.50
factors given by
0.4
k=0.35 analgous column model
30
2-hinged arch
3-hinged arch 0.2 fixed arch
0 0
0 0.1 0.2 0.3 0 0.1 0.2 0.3
f f
(a) L
(b) L
Figure 4-4. Comparison of effective arc length factors with stability analysis results obtained by Rambøll (1944)
88
4. Statical Analysis of Arch Bridges
All arch systems should be designed to carry dead loads in pure compression along their arch axis. In or-
der to achieve this, the shape of the arch should coincide as closely as possible with the pressure line of the
sustained load. For a given set of loads, the shape of the pressure line is the same for all arches, whether
they are hinged or fixed. The shape of the pressure line is the same as the bending moment diagram for a
simply supported beam subjected to the same loading (see Figure 4-5a,b). In order to determine a unique
pressure line, the rise of the arch f must be specified. Once f is specified, the ordinates of the pressure line
y(x) can be calculated using Equation 4-4, where MS(x) is the bending moment diagram in a simply sup-
ported beam caused by the given set of loads, and MS(L/2) is the moment of the midspan section of the
M S (x)
y(x) = f Equation 4-4
M S ( 12 L)
Since most of the sustained loads are applied to the arch through spandrel columns, the loads can be rep-
resented as a set of concentrated loads, as shown in Figure 4-5c,d. Representing the loads in this way res-
ults in a pressure line that is polygonal, rather than smoothly curved. Using a polygonal shape for the arch
simplifies the arch geometry, making it easier to build than a smoothly curved arch. In this form, devia-
tions of the arch from the pressure line are greatest between spandrel columns, since the self-weight of the
arch will always be distributed rather than concentrated at the angle breaks of the arch.
As an alternative to Equation 4-4, a graphical method of calculation can also be used to determine the
pressure line for a set of concentrated loads, as illustrated by Figure 4-5d,e. In this method, forces are
drawn as force vectors, with vector length representing magnitude of force, and vector direction represent-
ing direction of force. Equilibrium about Joint 1 in Figure 4-5d is expressed in Figure 4-5e by three force
vectors arranged in a triangle. Vector ST represents the load from the spandrel column and Vectors TO
and OS represent reacting forces along the pressure line. Assembling triangulated forces from each joint in
the pressure line results in global equilibrium Triangle VOS, where Vectors VO and OS are the right and
89
4. Statical Analysis of Arch Bridges
left support reactions, respectively, and Vector SV is the sum of applied loads. The horizontal length H of
Triangle VOS represents the horizontal support reactions and the constant horizontal component of force
along the pressure line. Alternate states of global equilibrium can be made by moving Point O horizontally.
Moving this point to the left increases the horizontal reaction H and reduces the rise f. Thus, Point O can
(a) bending moment for a simply supported beam (b) parabolic pressure line for
caused by a uniformly distributed load uniformly distributed load
y
f
x
H x H
4f
M S ( L2 ) H u f y(x) (x)(x < L)
L2
(c) bending moment for a simply supported beam (d) polygonal pressure line for
caused by a set of concentrated loads a set of concentrated loads
y
f
1
x H x H
M (x)
y(x) S L f
MS ( 2 )
M S ( L2 ) H u f
H
S
tion
e ac T
tr
1 lef
S O
rig
T ht
re
ac
tio
n
O V
equilibrium of equilibrium of
forces at joint 1 forces at all joints
Figure 4-5. Pressure lines for uniformly distributed and concentrated loads
The true distribution of self-weight, which is needed to determine the pressure line, is not known until the
final geometry of the arch is defined. Thus, the process of finding the pressure line and setting the arch on
it is iterative. For preliminary design, it is sufficient to approximate the sustained load as a uniformly dis-
tributed load, as in Figure 4-5b. With this assumption, the pressure line is described by the parabola y(x)
90
4. Statical Analysis of Arch Bridges
4f
y(x) = (x)(L ! x) Equation 4-5
L2
qL2
H= Equation 4-6
8f
Axial forces along the arch N(x) are given by Equation 4-7, where '(x) is angle of inclination of the arch:
H
N (x) = Equation 4-7
cos ! (x)
The axial force N is equal to the horizontal reaction H at the crown, and is greatest at the springing lines, as
given by:
2
! f$
N (x = 0, L) = H 1 + 16 # & Equation 4-8
" L%
Equations 4-5 through 4-8 were calculated based on the static equilibrium of a three-hinged parabolic
arch. They also apply to statically indeterminate arches because the additional rotational restraints are ir-
relevant to this uniform load condition (only axial compression stresses are caused). This can be verified
Nonuniform loads such as partial live loads cause arches to deflect off the pressure line. At each arch sec-
tion, bending moments are generated equal to the axial force times the eccentricity between the centroid
of the arch and the displaced line of action of force. Thus, nonuniform loads are carried by arches primar-
ily in bending.
Figure 4-6 shows the bending moment diagrams M(x) and bending moment envelopes in fixed, two-
hinged, and three-hinged arches caused by a single, vertical concentrated load Q. Each row of diagrams
represents one position of Q. The last row shows the maximum positive and negative moments caused by
91
4. Statical Analysis of Arch Bridges
applying Q at any point along the arch. Bending moments were obtained from the structural analysis pack-
age sap2000 using simple, elastic arch models. Only first-order analysis was considered, implying that
equilibrium was calculated based on the original undeformed geometry of the structure. All structural ele-
ments were made weightless and inextensible. In order to prevent frame action in the system, the arch was
isolated by modelling deck elements as simply supported between adjacent spandrel columns.
The data presented in Figure 4-6 show that the highest bending moments are carried near the quarter-
points of arches. One exception to this are the large negative moments carried by fixed arches near their
springing lines. Fixed arches are often deepened toward the springing lines to accommodate for these large
moments. Comparing moments within the middle three-quarters of the span, three-hinged arches carry
62% more positive bending moments than fixed arches, and 115% more in negative bending. These premi-
ums in bending moment are also reflected when partial distributed loads are considered, as shown in Fig-
ure 4-7. For partial distributed loads, three-hinged arches carry twice as much as the bending moments
carried in fixed arches near the quarter-points. Again, this reduction in moments in fixed arches comes
92
4. Statical Analysis of Arch Bridges
L L L
a a a
Q Q Q
f f f
M/QL in 10-3 fixed 2-hinged 3-hinged
a/L
-39.4
-2.0 -11.2 -6.2
0.05 3.32
8.6
41.6 42.7
-60.6
-21.6 -12.5
-6.8
0.10 11.3
25.6
67.9 72.0
-67.7
-30.6 -18.7
-12.9
0.15
21.5
42.0
81.6 89.2
-64.0 -37.6
-19.2 -25.0
0.20
31.9
53.6
85.6 96.0
-52.9 -42.1 -31.2
-24.5
0.25
40.8
59.1
82.9 93.7
-44.4 -37.5
-37.0 -28.0
0.30
46.9
59.3
76.3 84.0
-43.7 -43.7
-29.1
-18.9
0.35
49.3
56.0 68.3 68.2
-39.8 -50.0
-28.2 -12.0
-13.4 -7.1
0.40 -0.46
47.5 48.0
51.5 61.1
-56.2
-33.8 -34.9
-15.4 -24.6 -16.1
0.45
16.5
41.2 24.7
47.9 56.0
-62.5 -62.5
-19.9 -19.9 -25.3 -25.3
0.50
30.7 30.7
46.6 54.3
-67.7 -62.5
-29.1 -44.4
-14.3 -12.9 -16.2
Envelope
49.3 25.5
46.6 54.3
59.3
85.6 96.0
Figure 4-6. Bending moments caused by concentrated loads in fixed, two-hinged, and three-hinged arches
93
4. Statical Analysis of Arch Bridges
-17.0 q
-12.4
-6.7 -10.0
q over
four-tenths
of span
9.1 11.4
16.3 18.7
q -15.6 -15.6
-15.5
-8.8
q over
half of span
8.8
15.5 15.6 15.6
q
-11.1 -11.1
-4.9 -4.9 -6.8 -6.8
q over middle
third of span
8.0 5.2 8.0 7.2
-16.4 -18.7
-17.0
-9.3 -7.2
-4.1 -5.4
q envelope
4.1 5.4
9.3 7.2
17.0
16.4 18.7
Figure 4-7. Bending moments caused by distributed loads in fixed, two-hinged, and three-hinged arches
The three partial distributed load arrangements shown in Figure 4-7 are typically used to estimate maxim-
um bending moments in arches. The arrangements of partial distributed loads that cause maximum bend-
ing moments at given sections along the arch can be found using influence lines &, as shown in Figure 4-8.
The positive and negative areas under the influence lines are tabulated in the first column to the right of
each diagram. The second column tabulates the locations where influence lines are zero. Based on these
values, applying distributed loads from the support x=0 to x=0.36L, or x=0 to x=0.46L result in maxim-
um moments in the sections near the quarter-points. Thus, using distributed loads over four-tenths of the
span provides reasonable estimates of maximum first-order moments. Relative to the four-tenths span dis-
tributed load case, half span loading has slightly lesser maximum moments, but has greater moments in
the unloaded half of the arch. Also, the moments and deflections caused by half span loading are symmet-
rical, which better reflect the primary buckling modes illustrated in Figure 4-3. Hence, half span loading is
the most suitable load case when second-order effects and stability are being considered in arches.
94
4. Statical Analysis of Arch Bridges
L L L
x x x
f f f
!=0 at
!/L in 10-3 fixed 2-hinged 3-hinged
± area/L2
x/L
in 10-3
-67.7
49.3
-37.5
-15.7 -22.5
0.07L
0.05 ±8.90 ±6.68 0.34L ±7.37 0.35L
0.44L
26.6
41.6 42.7
-28.0 -40.0
-18.1
0.10 ±4.19 0.17L ±11.6 0.36L ±12.9 0.36L
0.54L
25.6
67.9 72.0
-52.5
-36.7
-14.3
0.15 ±4.66 0.28L ±14.7 0.38L ±16.5 0.37L
42.0
81.6 89.2
-60.0
-41.9
-20.6
0.20 ±7.41 0.36L ±16.3 0.40L ±18.5 0.39L
53.6
85.6 96.0
-62.5
-44.4
-26.4
0.25 ±9.00 0.40L ±16.4 0.43L ±18.8 0.40L
59.1
82.9 93.7
-60.0
-43.7
-29.1
0.30 ±9.34 0.44L ±15.3 0.46L ±17.5 0.42L
59.3
76.3 84.0
-52.5
-28.5 -39.8
0.35 ±8.57 0.48L ±13.1 0.49L ±14.8 0.44L
56.0 68.2
68.3
-33.9 -40.0
-25.4
0.16L
0.40 ±7.08 ±13.2 0.53L ±10.9 0.46L
0.52L
51.5 61.1 48.0
Figure 4-8. Bending moment influence lines for fixed, two-hinged, and three-hinged arches
95
4. Statical Analysis of Arch Bridges
Shape of bending
moment envelope for
concentrated loads
Shape of bending
moment envelope for
distributed loads
note: bridges
not drawn to
same scale Sandö Bridge (1943) Caracas (1953) Salginatobel (1930)
Figure 4-9. Comparison of bending moment envelopes and variations in depth along the arch
Many designers shape their arches to reflect the bending moment demands of the chosen arch type, as
shown in Figure 4-9. For example, the three-hinged arch in Maillart’s Salginatobel Bridge is deepest at the
quarter-points and thinnest near the springing lines and arch crown (Billington 2003). The deepening of
the arch increases the section modulus of the arch ribs where it is needed most, improving the structural
efficiency of the structure. A typical strategy for fixed arch bridges is to increase the depth and width of the
arch gradually toward the springing lines, as done in the Sandö Bridge. Most designers use symmetric arch
sections since positive and negative bending moment demands are usually of similar magnitude.
The effects of restrained deformation increase with the degree of statical indeterminacy of any system. Axi-
al deformations in arches, such as caused by shrinkage, creep of sustained loads, elastic axial stresses, hori-
zontal yielding of supports, and uniform changes in temperature induce bending moments in statically in-
determinate arches. These redundant moments MK are proportional to the imposed axial strain %K and
flexural rigidity EI of the arch and are inversely proportional to rise f. Using the force method (as done in
Section 4.4.2), it can be shown that the redundant moments MK at the ends of fixed, parabolic arches are
given by:
EI
M K = ! 152 " K Equation 4-9
f
96
4. Statical Analysis of Arch Bridges
Expressions for various imposed axial strains %K are given in Table 4-2, where H is horizontal reaction, EA
is average axial rigidity of the arch, ,(t) is creep coefficient, %sh(t) is concrete shrinkage strain, #T is change
Table 4-2. Expressions for axial strains and flexural rigidities in arches
effective
action axial strain %K flexural rigidity
elastic shortening H/EA EI
creep of sustained load ,(t)·H/EA Eadj·I
concrete shrinkage %sh(t) Eadj·I
uniform change in temperature -T/L EI
horizontal movement of supports -L/L EI
Because of the effects of creep, the effective flexural rigidity of the arch can be reduced if the imposed axial
strains are applied gradually over a long period of time. Trost (1967) and Bazant (1972) developed an ap-
proximate method that uses an age-adjusted, effective modulus to calculate long-term stresses caused by
long-term strains. The age-adjusted, effective modulus Eadj is given by Equation 4-10, where Ei is elastic
modulus at the time of initial stress, ,(t) is creep coefficient, and +(t) is aging coefficient. Typical values for
,(t) and +(t), which are functions of time and age at initial loading, are given in Bazant (1972).
Ei
Eadj (t) = Equation 4-10
1 + !" (t)
Figure 4-10 shows bending moment diagrams caused by restrained deformations in different arch types.
Fixed arches have strain-induced moments that are up to four times as large as moments in two-hinged
arches. Three-hinged arches are determinate and so do not develop these strain-induced bending stresses.
0
!·MK
!·MK
Figure 4-10. Bending moments caused by restrained deformations in fixed, two-hinged, and three-hinged arches
97
4. Statical Analysis of Arch Bridges
Since the effects of restrained deformations are proportional to the flexural rigidity of the system, the
effects tend to decrease as critical sections of the arch become nonlinear. As redundant moments in fixed
arches are increased, the arch system should evolve into a two-hinged arch and then into a three-hinged
arch, provided that there is enough plastic rotation capacity to form plastic hinges at the critical sections of
the arch. As the system softens, moments caused by restrained deformation will be limited, and load-in-
duced moments will redistribute themselves toward the quarter-points. Thus, statically indeterminate
arches that are sensitive to restrained deformations will probably not undergo sectional failure due to large
redundant moments alone. These arches, however, can fail due to the formation of plastic hinges, which
will redistribute and increase load-induced moments and increase deflections, and reduce the resistance of
the system against buckling. In other words, the softened system will be more prone to buckling as plastic
Other loads, such as wind loading and earthquake loading, may cause sectional forces in the arch that ex-
ceed those caused by vertical loads (dead loads and live loads) and restrained deformations. These other
loads, along with unbalanced multilane lane loading, also affect the out-of-plane behaviour of the system.
Because this thesis deals primarily with the longitudinal behaviour and proportioning of arch systems,
transverse loading will not be considered. It is therefore assumed that these other loads and their load
combinations do not increase the structural demands imposed on the concept arches proposed in Chapter
6. In places where wind and earthquake loads are severe, appropriate modifications should be made.
Section 4.3 described the primary differences between three-hinged, two-hinged, and fixed arches. Only
fixed arches, however, will be considered in the development of new concepts for arches in Chapter 6. The
primary reasons why only fixed arches will be considered are: (1) fixed arches have higher resistances
against buckling than hinged arches, which allows for lighter and more slender members to be used (2)
fixed arches have higher degrees of redundancy than hinged arches, which is favourable in terms of struc-
98
4. Statical Analysis of Arch Bridges
tural reliability, and (3) fixed arches are more commonly built than hinged arches (as observed later from
the database of concrete arches, Table 5-3 on page 161), which suggests that they may have some intrinsic
economic advantage over hinged arches. The following section will expand on the structural analysis of
fixed arch systems, and discuss the effects of deck-stiffening, shallowness, slenderness, system softening,
When acting alone, fixed arches are statically indeterminate to the third degree. When spandrel columns
and deck girders are integrated with the arch, the degree of indeterminacy of the system increases with the
number of fixed joints between the structural members. Analyzing such highly indeterminate systems pre-
cisely is only practically feasible through the use of computer frame analysis packages. Such packages are
generally ill-suited for preliminary design calculations of arches because: (1) they tend to inhibit creativity
by separating design from insights of structural behaviour drawn from simple manual calculation, (2) the
inter-relationships between redundant moments, distributions of stiffness in the system, and system
softening are not easily observed, and (3) the geometry of an arch is sensitive to the pressure line of the
load, and thus must be updated with each design change to minimize dead load moments. Thus, approx-
imate analysis methods, which allow for the identification of important design parameters and critical load
The degree of indeterminacy of fixed arched systems and the complexity of computation can be signific-
Assumption 1: Spandrel columns are pin-connected to the arch and deck. This assumption reduces the
indeterminacy of the system by removing the redundant moments associated with the ends of each span-
drel column. In some cases, spandrel columns are physically pin-connected to the deck using special bear-
ings. In other cases cases, spandrel columns are monolithically built into the arch and deck. For these
cases, Assumption 1 is still reasonable because the bending stiffness of the column is typically much less
than the connected deck or arch members. According to Hardy Cross’ moment distribution method
99
4. Statical Analysis of Arch Bridges
(Cross 1932), unbalanced fixed-end moments at joints are successively distributed to each connected ele-
ment in proportion to the ratio of its bending stiffness to the sum of bending stiffnesses of all elements
connected to the joint. As an example, moment distribution factors have been calculated for several
column joints in the Colorado River Bridge and are shown in Figure 4-11. Distribution factors were calcu-
lated using equations for non-prismatic members, as derived by Gere (1963). The highest spandrel column
distribution factor is 0.14, which is small. The other spandrel column joints have even smaller distribution
factors at their bases. The distribution factors equal to zero indicate that the joint connections between
0.14 0 0
fixed
connection
0.06
0.48
0.46
fixed
connection
0.03
1
0.5
6
0.4
fixed
connection
0.03
51
0. moment distribution factors
46
0.
Figure 4-11. Moment distribution factors for the Colorado River Bridge.
Drawing adapted from us dot fhwa (2003).
Assumption 2: The flexural rigidity of the arch projected on to the horizontal axis is constant over the
entire span and is equal to the flexural rigidity of the arch at the crown. This assumption is illustrated in Fig-
100
4. Statical Analysis of Arch Bridges
ure 4-12 and is expressed algebraically in Equation 4-11 (Menn 1990), where ' is the angle of inclination of
the arch, and EIarch and EIcrown are flexural rigidities of the arch and crown section of arch, respectively.
y tanƧ
s 1
x
EI arch (s)
flexural rigidity perpendicular to curved axis
EI arch (s)u cosƧ(s) EI crown
horizontal component of flexural rigidity EI arch (x) EI crown
Assumption 2 simplifies deflection calculations by making arches act as straight beams with constant flex-
ural rigidity. By changing the variable of integration (Equation 4-12), deflections $ calculated by the meth-
od of virtual work can be integrated with respect to the horizontal axis x (Equation 4-14), rather than the
curved axis of the arch s (Equation 4-13). In these equations, S is arc length, L is span length, m is virtual
S S M (s)
! = $ m(s) " # (s)ds = $ m(s) " ds Equation 4-13
0 0 EI arch (s)
L M (x)
! = # m(x) " dx Equation 4-14
0 EI crown
To ensure the validity and accuracy of Assumption 2, designers may choose to shape the arch such that
Equation 4-11 is true. Stiffening the arch towards its ends, as required by the assumption, may also come
naturally from the higher sectional force demands at the ends of the arch. If the distribution of flexural ri-
gidity departs significantly from Equation 4-11, then some further redistribution of moments among the
arch and deck will needed, especially near the springing lines.
101
4. Statical Analysis of Arch Bridges
Assumption 3: Vertical deflections of the deck and arch are equal along the span. This compatibility as-
sumption is valid at the locations where the spandrel columns connect with the deck and girder. The ac-
curacy of solutions based on this assumption increases as the number of spandrel columns are increased.
Assuming equal deck and arch deflections, $deck(x)=$arch(x), leads to simple expressions for the distribution
of system moments among arch and deck, as derived by the equations below.
Deflections of the deck are given by Equation 4-15 and Equation 4-16, where c is an integration constant
that result from solving virtual work integrals for given curvature diagrams ,(x), Mdeck is deck moment,
M deck (x) 2
! deck (x) = c(x) " L Equation 4-15
EI deck
M arch (x) 2
! arch (x) = c(x) " L Equation 4-16
EI crown
L
c = # m(x) ! " (x)dx ÷ " (x) ! L2 Equation 4-17
0
Equating arch and deck deflections $deck=$arch results in Equation 4-18 (integration constant c and span L
are the same for both arch and deck). This equation can be rearranged to obtain an expression for deck
EI deck
M deck (x) = M arch (x) Equation 4-19
EI crown
Flexible system moments M are defined as the sum of arch and deck moments (Billington 1973):
102
4. Statical Analysis of Arch Bridges
The flexural rigidity of the system EIsys is defined as the sum of arch and deck flexural rigidities, EIdeck and
EIcrown. Substituting EIsys in Equation 4-20 and rearranging results in expressions for moments distributed
EI crown
M arch (x) = M (x) Equation 4-21
EI sys
EI deck
M deck (x) = M (x) Equation 4-22
EI sys
These expressions show that system moments are distributed to the arch in proportion to the ratio of its
flexural rigidity to the flexural rigidity of the system (the same applies to the deck). One interpretation of
these results is that arch and deck participate together like a pair of parallel beams in carrying moments
In general, there are four ways that flexural rigidity can be distributed in the system, corresponding to the
four classes of fixed arch systems shown in Figure 4-13. The first class is the unstiffened arch system, whose
decks are discontinuous over the spandrel columns. Because of the discontinuity, the global flexural rigid-
ity of the deck can be taken as zero, regardless of the section properties of the deck. In this case, arches act
alone in resisting moments caused by nonuniform loads or restrained deformations. The second class is
self-stiffened arch system, or arch-stiffened system. In this case, the deck is continuous over the spandrel
columns, but its flexural rigidity is much smaller than the flexural rigidity of the arch. Because of this, only
a small portion of the system moments is distributed to the deck. Reversing the flexural rigidity of the arch
and deck results in the third class, called the deck-stiffened arch system. In this system, the flexural rigidity
of the deck is much greater than the flexural rigidity of the deck. This system was pioneered by Robert
Maillart in the 1920s and 1930s, during the latter years of his career as a bridge designer in Switzerland
103
4. Statical Analysis of Arch Bridges
(Billington 1979). His analysis of this system was simple and approximate, assuming that: (1) the arch axis
is taken as the pressure line of the dead weight plus half of the traffic load, (2) the arch carries all uniform
loads by axial forces only, and (3) all bending moments in the vertical plane are carried by the stiffening
deck girder alone (Billington 1973). The fourth class is the partially deck-stiffened arch system, whose deck
and arch have flexural rigidities of similar magnitude. In this case, both deck and arch carry significant
arch system structural model assumed primary system parallel beam analogy
classification of arches and redundant forces
Classification based Without simplification, Only three independent redundant System moments are distributed
on stiffness distribution. these arch models are forces are considered in each system. to deck and arch in proportion to
highly indeterminate. System moments caused by redundant the ratio of their flexural rigidity to
forces are identical among systems. the flexural rigidity of the system.
EI deck 0
unstiffened
arch system x1 x3
x2 EI crown
EI deck EI crown
self-stiffened
arch system
x1 x3 EI crown
x2
(1 < Ơ ) u x1 (1 < Ơ ) u x2
EI deck
partially deck-stiffened
arch system
x3
EI crown
Ơ x1 Ơ x2
x1 x2 EI deck
deck-stiffened
arch system
x3
EI crown EI deck
Figure 4-13. Classification of fixed arch systems, and their respective simplified and analogous structural models
The middle column of diagrams in Figure 4-13 shows the primary systems and redundant forces x1, x2, and
x3 for each class of fixed arch system. The sum of virtual bending moment diagrams caused by each set of
redundant forces is the same among the four classes. Thus, the procedure for calculating bending moments
for any arch system is as follows: (1) loads or restrained deformations are applied to the arch alone, (2) the
104
4. Statical Analysis of Arch Bridges
properties of the arch are assumed to be identical to the arch in the original system, except that its flexural
rigidity includes the flexural rigidity of the deck, (3) uniform loads are carried by the arch along the pres-
sure line in compression, and (4) bending moments are distributed to the arch and deck according to
Equations 4-21 and 4-22. The stiffened arch model used to calculate system moments is shown in Figure 4-
14, along with the virtual moment and virtual axial forces caused by the three redundant forces.
L
redundant virtual bending virtual axial
forces moment diagram force diagram
f
-f
primary system x3
and redundant forces -1
Figure 4-14. Stiffened arch model and moments and axial forces caused by unit redundant forces
Figure 4-15 investigates the validity of Assumptions 1 and 3 described previously. Fixed arch systems sub-
jected to uniformly distributed loading over half the span have been analyzed using the structural analysis
package sap2000. Three parameters have been varied: the flexural rigidities of the arch EIcrown, deck EIdeck,
and spandrel columns EIcol. All flexural rigidities are expressed as fractions of a constant reference rigidity
EI*. Each row of bending moment diagrams in the figure corresponds to a constant value of EIcol, but di-
fferent proportions of EIcrown and EIdeck. Each column of diagrams corresponds to constant proportion of
EIcrown and EIdeck, but different values of EIcol. Spandrel columns were modelled as monolithically connected
to the arch and deck, except for the last row of diagrams, where they are pin-connected. The flexural rigid-
ity of the arch EIarch was varied along the span according to Equation 4-11, such that the horizontal com-
105
4. Statical Analysis of Arch Bridges
span The proportion of bending moment Bending moments in the arch and deck
distributed to the arch is proportional to increase as the flexural rigidity of the
span!= 200 m the ratio of the flexural rigidity of the columns decrease.Thus, it is conservatve
q EIdeck rise != 25 m arch to the flexural rigidity of the to assume that the columns are pin-
q != 35 kN/m system. This is illustrated by connected when designing the deck and
f EIcol I*!= 16.6 m4 comparing diagrams in the same row. arch.This is illustrated by comparing
E != 25000 MPa diagrams in the same column.
class of partially
arch system deck-stiffened deck-stiffened self-stiffened
400
1930
2590
3090
0.25·EI*
3500
3500
3090
2590
EI*
1930
EI*
-4.2 -8.0
-1.9 -3.0
-0.1
0.1·EI*
-7.1 -12.4
-2.9 -5.0
-0.1
Figure 4-15. Effect of arch, deck, and column flexural rigidities on the distribution of bending moment
Two observations can be made from the analytical results presented in Figure 4-15. First, bending mo-
ments in the arch and deck increase as the flexural rigidity of the columns are decreased. This implies that
assuming pin-connected spandrel columns is conservative, since the moments in the arch and deck are
highest when the columns are pin-connected. The moments in spandrel columns, however, increase as
their flexural rigidity increases, and should be checked for strength and ductility once the overall propor-
tions of the system are specified. The second observation is that the moment distribution between arch and
106
4. Statical Analysis of Arch Bridges
deck deduced from Assumption 3 is reasonably accurate, regardless of the fixity and flexural rigidity of the
spandrel columns. This can be seen by comparing the end moments of the arch and deck in the figure.
Assumption 4: When using the method of virtual work for calculating sectional forces in arches, only
bending deformations and their internal work terms need be considered. Figure 4-14 shows the virtual bend-
ing moment and virtual axial force diagrams caused by each unit redundant force. The internal work terms
associated with axial deformation . nN/EA dx can be neglected in non-shallow arches, which are arches
whose crown deflections due to axial shortening are small relative to the initial rise of the arch. Using
high-strength materials in arches may allow for greater shallowness than has been previously achieved.
Thus, Assumption 4, which has typically been valid for arches made from conventional concrete, will not
In this section, the analysis of fixed arches using the force method will be presented using the simplified
statical system and the four assumptions presented in Section 4.4.1. The inclusion of axial deformations in
internal work formulations, which is important for the statical analysis of shallow arches, will be con-
The force method uses compatibility equations to solve redundant forces, as given by:
In this equation, deformation matrix [d0] contains deformations of the statically determinate primary sys-
tem caused by loads or dimensional changes, force vector [F] contains unknown redundant forces, and
flexibility matrix [d] contains deformations caused by unit redundant forces. Total deformations $ are the
sum of bending deformations $M, axial deformations $N, and shear deformations $V:
! = ! M + ! N + !V Equation 4-24
For the type of structural members used in arches, shear deformations are typically small and therefore
will be assumed to be zero ($V=0). Using Assumption 4, axial deformations will be neglected ($N=0). Thus,
107
4. Statical Analysis of Arch Bridges
only bending deformations $M will be considered in the calculation of flexibility coefficients $ij as given by
Equation 4-25, where mi is virtual moment caused by unit redundant force i, and ,j is real curvature
S
! ij = $ mi (s) " # j (s)ds Equation 4-25
0
These quantities are integrated with respect to the curved arch axis s, over the total arc length of the arch S.
Assuming linear-elastic behaviour, real curvature ,j can be expressed in terms of bending moment Mj di-
vided by flexural rigidity EI (Equation 4-26). The resulting expression can be simplified further by apply-
ing Assumptions 1, 2, and 3 and changing the variable of integration to a horizontal variable, using the
S M j (s)
! ij = " mi (s) ds
0 EI (s)
Equation 4-26
L M j (x)
= " mi (x) dx
0 EI sys
According to the redundant forces shown in Figure 4-14, the flexibility matrix [d] of the system is:
" 9 3 15
f !1 %
EI sys $ '
2
[d]!1 = $ 3 9 15
2 f !1 ' Equation 4-28
L $ '
!1 !1
$# 15
f 15
f 45
f !2 '
2 2 4
&
108
4. Statical Analysis of Arch Bridges
Force vector [F] can be solved by using Equation 4-29, where xi are redundant forces, and $i0 are deforma-
tions of the primary system caused by loads or dimensional changes calculated using Equation 4-25.
[F] = ![d]!1[d 0 ]
" x1 % " 9 3 15
f !1 % " )10 %
$ ' EI sys $ 2
' $ ' Equation 4-29
$ x2 ' = ! $ 3 9 15
2 f !1 ' ( $ ) 20 '
$ x ' L $ ' $ '
$# 3 '& $# 15
f !1 15
f !1 45
f !2 ' $# ) 30 '&
2 2 4
&
distributed load q
M (x) = 0
over full span
# q
q
% (
141x 2 ! 60Lx + 4L2 ) 0"x"
L
% 486 3
distributed load q %
over middle third span
M (x) = $ !
q
(
102x 2 ! 102Lx + 23L2 ) L 2
<x" L
% 486 3 3
% q
(
% 486 141x ! 222Lx + 85L
2 2
) 2
3
L<x"L
&
q
# q L
% ! ( 8x ! L ) ( 2x ! L ) 0 " x "
distributed load q % 64 2
M (x) = $
over half span % q ( 8x ! 7L ) ( 2x ! L ) L < x " L
%& 64 2
Q
) 27 " 10 2 % L
+ Q $ x + 6x ! L ' 0(x(
concentrated load Q + 512 # L & 4
M (x) = *
at quarter-point + 1 " 270 % L
x 2 ! 350x + 101L ' <x(L
+ 512 $# L
Q
& 4
,
Q
) Q " 30 2 % L
+ $ x ! 14x + L ' 0(x(
concentrated load Q + 32 # L & 2
M (x) = *
at midspan + Q " 30 2 % L
$ x ! 46x + 17L ' <x(L
+ 32 # L & 2
,
strain
EI # # x & &
2
15 x
axial shortening M (x) = ! K " " % 6 % ( ) 6 + 1(
2 f $ $ L' L '
109
4. Statical Analysis of Arch Bridges
With the redundant forces solved, final bending moment diagrams can be found by superimposing mo-
ment diagrams caused by external loads and redundant forces. These moments can then be distributed
among deck and arch according to Equations 4-21 and 4-22. Table 4-3 summarizes bending moment dia-
grams for six load cases calculated by the author using the force method.
In this section, the analyses of fixed arch systems will be repeated, this time with the inclusion of axial de-
formations in internal work formulations. This more complex analysis reveals the effects of axial rigidity of
The cross sectional area of the arch typically changes along the arch axis. Similar to the assumption made
for flexural rigidity, the axial rigidity of the arch is assumed to vary according to Equation 4-30, such that
its horizontal component is constant and equal to the axial rigidity of the arch at the crown EAcrown (Melan
Flexibility coefficients $ij including axial deformations are given by Equation 4-31, where ni is virtual axial
force caused by unit redundant force i, and %j is real strain caused by unit redundant force j. Real strain is
equal to real axial force Nj divided by arch axial rigidity EAcrown. The first term in Equation 4-31 is identical
to Equation 4-26, for which the variables have already been defined.
L L
! ij = $ mi" j dx + $ ni # j dx
0 0
L Mj L Nj Equation 4-31
= $ mi dx + $ ni dx
0 EI sys 0 EAcrown
The flexibility matrix [d] of the system based on the redundant forces shown in Figure 4-14 is:
110
4. Statical Analysis of Arch Bridges
" 13 1
( 13 f %
" !11 !12 !13 % $ 6
'
$ ' L $ 16 1
( 13 f '
[d] = $ ! 21 ! 22 ! 23 '= $
3
' Equation 4-32
$ ! ' EI sys $ ( 1 f EI sys '
$# 31 ! 32 ! 33 ( 13 f f +
8 2
'& $ 3 15
EAcrown '
# &
# 6
f 2 Acrown + 6I sys 2
f 2 Acrown ! 3I sys fAcrown &
EI sys 30 % 5 5
(
!1
[d] = " % 2
f 2 Acrown ! 3I sys 6
f 2 Acrown + 6I sys fAcrown ( Equation 4-33
L 4 f 2 Acown + 45I sys % 5 5
(
% fAcrown fAcrown 2 Acrown (
3
$ '
When uniformly distributed load q is applied over the entire span, the primary system deformations [d0]
are given by Equation 4-34, where $10 and $20 are rotations at each end of the arch, and $30 is horizontal
displacement:
"!10 % " %
T
qL3 qL3 qL3
[d 0 ] = $$! 20 '' = $ ( 241 ( 1
24
1
15 f ' Equation 4-34
$ EI sys EI sys EI sys '
$#! 30 '& # &
Solving for force vector [F], the redundant forces shown in Equation 4-35 are obtained, where x1 is mo-
ment at the left support ML, x2 is moment at the right support MR, and x3 is horizontal reaction H:
! x1 $ ! $
T
( ( qL2
[F] = ## x2 && = # ' 121 qL2
1
' qL1 2 1
& Equation 4-35
1+ ( 1 + 45 ( f 1 + 45 ( &
12 8
#" x3 &% #" %
These expressions have been simplified by introducing a non-dimensional ratio ", which is defined as:
45 EI sys
!= " Equation 4-36
4 f 2 EAcrown
111
4. Statical Analysis of Arch Bridges
Non-dimensional ratio " will be referred to as the arch-beam parameter because it describes the transition
of shallow arches between fixed arch and fixed beam behaviour. At the limit "=0, fixed arch systems exhib-
it pure arch behaviour, responding to loads and deformations as described in Section 4.4.2. At this limit,
arches are not shallow, since crown deflections are small relative to the initial rise of the arch. At the limit
"=!, fixed arch systems exhibit pure beam behaviour, responding to loads and deformations as would
straight fixed beams. At this limit, arch crown deflections are large relative to the rise of the arch.
The force method was repeated for several different load cases, including: partial distributed load over half
span and middle third, concentrated loads at midspan and quarter-point, and restrained axial deforma-
tion. The resulting support reactions are tabulated in Figure 4-16. In addition to showing support reac-
tions, the figure also shows bending moments and bending deflections at critical locations (x=a) for the
112
L moments caused by axial forces caused by
! ! redundant forces ! redundant forces ! flexural rigidity EI = EI sys ! axial rigidity EA = EAcrown
f -H·f -H 45 EI
! ! ! ! arch-beam parameter ! = " = 45 # $2
f !
fixed arch system 4 f 2 EA
2
ML $ H ' $ L 3
'
! ! ! ! axial deflections at crown " n! dx !f " & 64
+ # ) * & 15 + 4 f)
% EA ( % f (
ML MR
primary system H MR
and redundant forces
load case! full span q ! middle third q ! half span q ! midspan Q ! quarter-point Q ! shrinkage
q q q Q Q
strain
" 2
1 " 398 # 1 ! 53 " 1 " 4# 27
1 + 83 # EI 1
left moment reaction, ML! ! 121 qL2 ! + 243 qL2 ! ! ! 641 qL2 ! + 321 QL ! ! ! 512 QL " ! ! 152 " # #
1+ " 1+ # 1+ " 1+ # 1+ # f 1+ $
113
" 2
1 " 398 # 1 + 113 ! 1 " 4# 21
1 " 87 # EI 1
right moment reaction, MR! ! 121 qL2 ! + 243 qL2 ! ! + 641 qL2 ! + 321 QL ! ! + 512 QL ! ! ! 152 " # #
1+ " 1+ # 1+ ! 1+ # 1+ # f 1+ $
L L L L L L
critical location x=a! ! ! ! ! !
2 2 4 2 4 2
57 32
! 5
1 + 10 " 1
1 + 373 ! 1 + 83 " 243
1 + 27 " EI 1
moment at x=a ! + 241 qL2 ! + 972 qL2 ! ! + 128 qL2 ! + 643 QL ! ! + 4096 QL ! ! + 154 ! " "
1+ ! 1+ " 1+ ! 1+ " 1+ " f 1+ #
2
! 49 qL4 $ 2
! 11 qL4 $ 1
! 1 QL3 $ 19
! 9 QL3 $
bending deflections " m! dx at x=a! 0 ! 49 #" 31104 EI &% ! 11 #" 12288 EI &% ! 16 #" 192 EI &% ! 64 #" 4096 EI &% ! 0
Another important system parameter is modified slenderness ratio (f, which has been used in stability
models for shallow arches by Cai et al. (2009) and Bradford et al. (2007). Modified slenderness ratio is de-
fined as two times the arch rise f divided by radius of gyration r of the arch (Equation 4-37). Because the
deck contributes to the bending stiffness of the systems being considered, it is appropriate to replace r with
EAcrown 2 f
!f = 2 f = Equation 4-37
EI sys rsys
EI deck + EI crown
rsys = Equation 4-38
EAcrown
Because they share the same parameters, arch-beam parameter " can be expressed explicitly in terms of
! = 45 " #2
f Equation 4-39
The transition between pure arch and pure beam behaviour is best illustrated by comparing moments car-
ried by fixed arch systems with those carried by fixed beams subjected to the same loading. From Equation
4-35, arch end moments ML and MR caused by uniform load q distributed over the full span L are given by:
"
M arch (x = 0, L) = ! 121 qL2 Equation 4-40
1+ "
Under the same loading, end moments in fixed beams are given by:
In terms of arch-beam parameter " and modified slenderness ratio (f, the ratio between fixed arch and
114
4. Statical Analysis of Arch Bridges
M arch (x = 0, L) !
=
M beam (x = 0, L) 1 + !
Equation 4-42
45 " #2
=
f
1 + 45 " #2
f
The horizontal reaction in shallow arches H caused by uniform loading q is given by:
qL2 1
H= !
8 f 1+ "
Equation 4-43
qL2 1
= !
8 f 1 + 45 # $2
f
The first factor in Equation 4-43 is the horizontal reaction in non-shallow arches (Equation 4-6). The
second factor 1/(1+") is a ratio between 0 and 1 that represents the amount of arch action that is present in
the system, 0 being no arch action (pure beam action), and 1 being pure arch action.
The ratio of horizontal reactions (Equation 4-43) and ratio of end moments (Equation 4-42) are plotted in
Figure 4-17 as functions of modified slenderness ratio (f. The transition point, where arch systems behave
equally as arch and beam occurs when (f =9.49. Although it would be possible to design arch systems with
(f =9.49, the system would not be structurally efficient. The primary advantage of the arch form is that per-
manent loads can be carried in pure compression, which requires less material than carrying permanent
loads in flexure. This advantage is lost when the modified slenderness ratio of the system is low enough
such that significant bending moments are present even under permanent loads whose pressure line fol-
lows the shape of the arch. Such moments would better be resisted using girder-type systems rather than
115
4. Statical Analysis of Arch Bridges
0
0 20 40 60
Modified slenderness ratio !f
To ensure that given arch designs behave efficiently, a minimum modified slenderness ratio (f of 20 is pro-
posed (see Figure 4-17). This threshold ensures that dead load moments in fixed arches are less than 10% of
those in fixed beams, and that more than 90% of uniform loads are carried in axial compression along the
axis of the arch. As (f is decreased beyond the proposed threshold, there is a rapid increase in bending mo-
ment and a rapid decrease in axial force in the arch, which are both undesirable.
The transitional behaviour between arch and beam is not limited to the case of uniformly distributed load-
ing. All applied load cases shown in Figure 4-16 have moments that are functions of the factor "/(1+").
These factors account for the effects of axial deformation on the system. When "=0, axial deformations do
not cause additional bending moments in the system above that of non-shallow arches. As " is increased
above 0, axial deformations introduce additional sectional forces that move the system toward fixed beam
behaviour.
116
4. Statical Analysis of Arch Bridges
Crown deflections #f, which will later be used to calculate second-order moments, can be calculated using
Equation 4-44, where H is horizontal reaction caused by dead and live loads, EAcrown is axial rigidity of the
arch crown, % is strain caused by creep, shrinkage, and uniform drops in temperature, L is span, f is rise,
$ H ' $ L2 3 '
!f " & + # ) * & 15 +4 f) Equation 4-44
% EAcrown ( % f (
64
This equation is calculated by integrating ∫n(x)·%(x) dx, where n is virtual axial force caused by a unit load
applied at the crown, and % is real axial strain in the system. This equation is similar to the approximate
H $ 1 L2 3 '
!f " # 4 +4 f) Equation 4-45
EA &% f (
The last four rows of diagrams and equations in Figure 4-16 depict the differences of the limiting cases of
pure arch behaviour "=0 and pure beam behaviour "=!. For each type of load case, including distributed
load q, concentrated load Q, and restrained deformation, bending moment diagrams for arches with "=0
and arches with "=! are drawn to the same scale. By inspection, pure arch systems have significantly
lower moments than pure beam systems. Bending deflections in the systems are calculated using virtual
work: ∫m(x)·,(x) dx, where m is virtual moment caused by a unit load applied at the location of and in the
same direction as the desired bending displacement and , is real curvature caused by the given loading.
The results show that pure arch systems have significantly lower bending deflections than pure beam sys-
tems. The case of distributed loading over half the span, for example, causes deflections in pure arches that
are 18% of those in pure beams. These deflections, however, exclude vertical deflections caused by axial de-
formations of the arch. Thus, the expressions given in the figure for bending deflections at midspan must
117
4. Statical Analysis of Arch Bridges
The flexural buckling of arches has previously been discussed in Section 4.3.1. Recent analytical studies by
Cai et al. (2009) have shown that the buckling resistance of shallow arches can be significantly lower than
those predicted by the effective arc length method presented in Section 4.3.1. Using virtual work, Cai et al.
formulated nonlinear equilibrium and compatibility equations, and then solved them to obtain solutions
for symmetric snap-through and antisymmetric bifurcation buckling loads. In their formulations, they as-
sumed that: (1) the form of the arch is parabolic, (2) the square of slopes of the parabola are much less than
unity: (dy/dx)2<<1, and (3) the change in horizontal displacements of the arch with respect to the hori-
The distinction between when arches are shallow and when they are not shallow is not specified by Cai et
al. (2009). An inspection of their second assumption (dy/dx)2<<1 gives some insight into when their stab-
ility analysis is valid. The square of the slope (dy/dx)2 of parabolic arches at their springing lines is given by
'2
! dy(x = 0, L) $ ! L$
2
Figure 4-18, which illustrates Equation 4-46, shows that the assumption of (dy/dx)2<<1 is only valid at the
springing lines at high span-to-rise ratios. Arch springing lines with span-to-rise ratios of 12:1 or higher
square of slope 1
at springing line
d2y
dx2
0.5
12:1
0.1
0
0:1 4:1 8:1 12:1 16:1 20:1
span-to-rise ratio L/f
118
4. Statical Analysis of Arch Bridges
Cai et al. (2009) found that modified slenderness ratio (f (Equation 4-37) was the governing parameter of
their virtual work analysis. They identified three domains of buckling behaviour, determined by the value
of (f. Their results (adjusted according to the erratum by Cai and Feng (2009)) are summarized below.
Domain 1: When (f >19.8, shallow arches buckle anti-symmetrically at critical buckling loads HE, f :
$ "2 '
H E, f = & 0.6 ± 0.4 1 ! 30.7 2 ) * H E # f > 19.8 Equation 4-47
% #f (
HE is the second mode, buckling resistance of fixed columns under uniform compression, as given by
Equation 4-48. When arc length S is substituted for L, HE is equal to the buckling resistance of non-shallow
EI
HE = ! 2 2
Equation 4-48
# L&
%$ 0.7 " ('
2
Domain 2. When 9.87 $ (f $ 19.8, shallow arches buckle symmetrically at critical buckling loads:
( )
H E, f ! 0.00079 " 2f # 0.009 " f + 0.314 $ H E 9.87 % " f % 19.8 Equation 4-49
Cai et al. (2009) wrote a computer program to solve two equations iteratively to obtain symmetric buck-
ling loads for given values of modified slenderness. Because the iterative procedure was complex and did
not result in closed form solutions, Cai et al. used a least squares approximation of the results to calculate
the quadratic coefficients in Equation 4-49. Their solution showed good agreement with results from
numerical buckling analyses done in the finite element analysis package ansys.
Domain 3. When (f < 9.87, shallow arches do not buckle because they do not carry axial forces large
enough to cause buckling. Thus, these shallow arches behave more like curved beams with small initial
119
4. Statical Analysis of Arch Bridges
no buckling
symmetric buckling
antisymmetric buckling
q
HE,f
1
H E, f H E, f
HE
0.8
0.4
q
0.2
H E, f H E, f
0
0 9.87 19.8 60
symmetric buckling mode
Modified slenderness ratio !f
Figure 4-19. Buckling of shallow arches. Adapted from Cai et al. (2009).
Equations for buckling loads, or buckling resistances HE, f for shallow arches are plotted in Figure 4-19. The
figure shows that the efficiency threshold of (f > 20 proposed in Section 4.4.3 has relevance to the buckling
mode of shallow arches. Proportioning arches such that their modified slenderness is greater than 20 en-
sures that their primary buckling mode remains anti-symmetric and that their buckling resistance is at
least 80% of HE. As (f is decreased beyond the proposed threshold, there is a rapid decrease in arch buck-
ling resistance due to a distinct change in the primary buckling mode of the system.
An additional effective arc length factor kf can be used in Equation 4-2 on page 87 to account for the re-
duction of buckling resistance in arches due to the effects shallowness. The shallowness factor can be cal-
culated using Equation 4-50, or can be conservatively taken as 1.12, if modified slenderness ratio is greater
than 20.
' 12
! H E, f $
kf = # Equation 4-50
" H E &%
The force method solutions presented in Section 4.4.2 and Section 4.4.3 do not account for arch displace-
ments in formulations of equilibrium, and are thus considered first-order analyses. Because arches carry
loads in compression and bending, they are susceptible to additional bending moments that are propor-
120
4. Statical Analysis of Arch Bridges
tional to the eccentricities between the centroid of the arch and line of action of load. Second-order ana-
lyses, in which these additional moments are considered, have been previously discussed in Section 3.1 for
eccentrically-loaded columns. The second-order behaviour of arches is very similar to the second-order
behaviour of straight columns. Vianello’s deflection magnification method (1898) will be used to calculate
second-order deflections and moments approximately. Second-order arch deflections w(x) are given by
Equation 4-51, where w0(x) is first-order arch deflection, H is horizontal reaction, and HE is buckling resist-
1
w(x) = w0 (x) Equation 4-51
H
1!
HE
First-order deflections at critical locations along the arch w0(x=a) can be calculated using the equations
shown in Figure 4-16 on page 113. When considering crown deflections w0(x=L/2), deflections caused by
both flexural deformations ∫m·, dx and axial deformations #f =∫n·% dx (Equation 4-44) need to be con-
sidered. The component of #f that is caused by sustained loads can be excluded if the arch is properly
cambered during construction. In other words, the camber of the arch should be designed such that the
pressure line of the sustained load and the axis of the arch are concentric once the arch is closed, eliminat-
ing any eccentricity that would cause additional moments. When considering quarter-point deflections
w0(x=L/4), deflections caused only by flexural deformations ∫m·, dx need be considered, since axial short-
ening of the arch causes only minimal eccentricities of load near the quarter-points.
Horizontal reactions of the arch H can be calculated by summing all applicable equations for H in Figure
4-16, including the effects of restrained deformations and axial shortening which can reduce compression
in the arch. The largest portions of H are typically contributed by permanent dead loads qdead and uni-
formly distributed live loads qlive, which can be calculated using Equation 4-43.
The critical buckling load of fixed arches HE can be calculated using Equation 4-2 on page 87, in which
effective arc length factor k is taken as 0.35, and EI is taken as the flexural rigidity of the system EIsys. If the
121
4. Statical Analysis of Arch Bridges
span-to-rise ratio of the system is 12 or higher then the buckling resistance of shallow arches HE, f (Equa-
At the critical section, the eccentricity e(x=a) between the centroid of the arch and the deformed line of
action of the load is proportional to the second-order deflection w(x=a), as given by Equation 4-52, where
Eccentricity constants & can be assumed to be the same as those in straight columns with similar deflected
shapes. Arch deflections caused by uniformly distributed loading over half the span are similar to the
second mode shape of straight, fixed-fixed beams, as shown in Figure 4-20 (Menn 1990). In this case, the
eccentricity between the deformed line of action of load and the centroid of the critical section of the de-
formed member is 0.733 times the maximum deflection of the member (& = 0.733). For load cases in which
the crown section is critical, arch deflections are similar to the first mode shape of straight, fixed-fixed
beams, as shown in Figure 4-20. In this case, the eccentricity between the deformed line of action of load
and centroid of the critical section of the deformed member is 0.5 times the maximum deflection of the
q q
H H H H
Z = second-order deflection
Z Z at critical section
H
H H H H undeformed line of action
H H H deformed line of action
e=ƦÃZ e = second-order eccentricity
e=ƦÃZ
e=0.733ÃZ of load at critical section
e=0.5ÃZ
Figure 4-20. Analytical model for second-order analysis of arches. Partially adapted from Menn (1990).
The total second-order moment M at the critical section (x=a) is the sum of first-order moment M0 and
122
4. Statical Analysis of Arch Bridges
Vianello’s method for calculating second-order moments provides a simple and rational means of account-
ing for geometric nonlinearity. The nonlinear response of the material caused by the cracking of concrete,
the yielding of steel, and the pull-out of fibres can be accounted for by replacing flexural rigidity EI with
secant stiffness EI!, which is a secant to the moment-curvature diagram calculated for a given point as giv-
en by Equation 4-54, where M is bending moment, , is curvature of the section, and N is axial force
(Menn 1990):
M
EI !(N ) = Equation 4-54
"
The calculation of EI! is iterative because the second-order moment M is a function of deflection, and
deflection is a function of EI!, and EI! is a function of M. To simplify calculations, secant stiffness can be
taken as one of three discrete values: secant stiffness at cracking EIcr = Mcr ÷ ,cr, secant stiffness at yielding
of reinforcing steel EIy = My ÷ ,y, and secant stiffness at ultimate state EIu = Mu ÷ ,u.
The various limit states corresponding to cracking, yielding, and ultimate have previously been discussed
in Section 3.5.1 for conventional reinforced concrete sections and in Section 3.5.2 for ultra high-perform-
ance fibre-reinforced concrete sections. These limit states were examined within the context of two ex-
ample hollow box sections whose geometries were calibrated so that their maximum axial forces in pure
axial compression were about the same. From this study, it was observed that high-strength concrete sec-
tions do not exhibit a distinct yield point corresponding to distinct changes in flexural stiffness. This is due
to the difference in form factor of reinforcing steel as fibres, rather than bars. It was also observed that
effective secant stiffnesses of high-strength concrete sections at ultimate limit states are typically 50% to
75% of the gross flexural rigidity EI0 of the section. This stiffness is quite high compared to the 25% of EI0
For the statical analysis of ultra high-performance fibre-reinforced concrete systems, the effective secant
stiffness for serviceability limit states can be assumed to be equal to the secant stiffness at cracking EIcr,
123
4. Statical Analysis of Arch Bridges
which is equal to, or slightly less than the gross flexural rigidity of the section EI0. The effective secant stiff-
ness at ultimate limit states EIu can be approximated using the following equation, where N! is maximum
axial force of the section in pure axial compression (same as Equation 3-25):
+ # N &
- 1
EI 0 % + 1( 0 ) N ) 12 N "
-
2
$ N" '
EI u ! , Equation 4-55
- # N&
1
EI 0 % 2 * ( 1
N" ) N ) N"
- 2
$ N"'
2
.
System flexural rigidity of systems EIsys is defined as the sum of the rigidities of the deck and arch. Simil-
arly, system secant stiffness EI!sys is defined as the sum of secant stiffnesses of the arch and deck:
! = EI deck
EI sys ! + EI crown
! Equation 4-56
The value of EI!sys at ultimate limit states can conservatively be taken as the sum of effective secant stiff-
nesses EIu of both the arch and deck. In some cases, only one of the two elements will have exceeded its
cracking moment under a given set of factored loads. For example, in a deck-stiffened arch, it is possible
that the deck girder has exceeded its cracking moment, while the arch remains uncracked at ultimate limit
states. In this case, it is appropriate to calculate the effective system secant stiffness as:
! = EI u,deck + EI 0,crown
EI sys Equation 4-57
Using secant stiffnesses to account for nonlinear moment-curvature behaviour also affects how flexible
system moments are distributed to the arch and deck. Updating Equations 4-21 and 4-22 on page 103 result
!
EI crown
M arch (x) = M (x) Equation 4-58
!
EI sys
!
EI deck
M deck (x) = M (x) Equation 4-59
!
EI sys
124
4. Statical Analysis of Arch Bridges
For convenience, a system stiffness factor κ can be defined as the ratio between the effective system secant
"
EI sys
!= Equation 4-60
EI sys
Hence, all calculations that use system flexural rigidity EIsys can be modified by multiplying EIsys by system
stiffness factor κ. When the arch and deck remain uncracked, such as under service loads, system stiffness
There are four important quantities that depend on the system stiffness factor κ: (1) deflections, which are
inversely proportional to κ, (2) buckling resistances, which are proportional to κ, (3) system radii of gyra-
tion, which are proportional to the square root of κ, and (4) moments caused by restrained deformation,
which are proportional to κ. Using estimates of system stiffness factors κ that are lower than the value ob-
tained more precisely through iteration (Equation 4-54) is conservative for the first two quantities: deflec-
tions are increased and buckling resistances are reduced. Using low estimates of system stiffness factors κ is
unconservative for the last two quantities. Decreases in system radii of gyration rsys causes increases in
modified slenderness ratios (f, which moves the system toward arch behaviour (as seen in Figure 4-17 on
page 116), causing moments in the system to be reduced. Moments caused by restrained deformation are
highly sensitive to the assumed value of system flexural rigidity. Significant underestimates of κ would res-
ult in significant underestimates of these moments. Thus, particular attention must be given to the calcula-
tion of these moments at serviceability limit states, where structures typically remain uncracked (κ=1), and
at ultimate limit states, where the extent of system softening (κ<1) is governed by the rotation capacities of
Arch-beam parameter " (Equation 4-36) can be reformulated to account for material nonlinearity by using
125
4. Statical Analysis of Arch Bridges
45 # EI sys
!= " Equation 4-61
4 f 2 EAcrown
This formula assumes that the axial rigidity EA of the member remains unchanged even at ultimate limit
state. This assumption is valid as long as the axial forces in the member are kept below 70% of the maxim-
um compressive strength of the section: N < 0.7N!. Below this limit, mid-depth strains along the member
will remain in the linear-elastic range of the material, as shown by the stress-strain curve in Figure 2-6 on
page 24.
Up to this point, only moments arising from the global deformations of arch systems have been con-
sidered. These moments are collectively known as flexible system moments (Menn 1990). Fixed system
moments, which are defined as the moments that arise from local deformations of the deck and arch
between spandrels columns, must be added to the flexible system moments to obtain the total bending
moment demand. The magnitudes of fixed system moments are related to interior span lengths. Decks
must safely carry both dead and live loads to supporting spandrel columns in bending before loads are car-
ried by the system as a whole. If the arch form is polygonal in shape, then self-weight moments of the arch
will also cause local bending between the angle breaks of the polygon. As a conservative approximation,
fixed system moments can be accounted for by increasing the positive and negative flexible system mo-
ment demands Mpos and Mneg by the maximum moments caused by distributed load q and concentrated
load Q fixed-fixed beams with spans equal to the interior span length Ln:
126
4. Statical Analysis of Arch Bridges
Section 4.4 presented methods for calculating second-order moments in fixed arches caused by dead loads,
live loads, and restrained deformations applied individually. The following sections will describe methods
for which these loads can be combined to estimate maximum sectional forces in the system. Five load
combinations are considered: three that cause maximum sectional forces at midspan and two that cause
maximum sectional forces near the quarter-points, as shown in Figure 4-21. When the effects of restrained
deformation (i.e. creep, shrinkage, and changes in temperature) are non-negligible, maximum sectional
forces in fixed, shallow arch systems tend to occur at the springing lines and at midspan. Calculations for
these load combinations are summarized in Table 4-4. When the effects of restrained deformation are neg-
ligible, maximum sectional forces tend to occur near the quarter-points. Calculations for these load com-
short-term
live load
moments
temperature
induced moments
concrete shrinkage
induced moments
short-term
and additional
long-term secondary
moments
127
4. Statical Analysis of Arch Bridges
128
4. Statical Analysis of Arch Bridges
Horizontal reactions H and bending moments M are divided into two categories: forces caused by long-
The response of fixed arches to long-term loads and short-term loads will be discussed respectively in Sec-
tion 4.5.1 and Section 4.5.2. Calculations are presented in the context of live load 2 (partial distributed load
over middle third span) and live load 3 (concentrated load at midspan) from Figure 4-21, combined with
dead load, creep, shrinkage, and uniform temperature drop to cause maximum sectional forces at mid-
span. Section 4.5.3 describes the calculation of total second-order moments, which requires that long-term
deflections are modified as equivalent short-term deflections. The last section deals with live load 4 (partial
distributed load over half the span) and live load 5 (concentrated load at the quarter-point) from Figure 4-
21. These calculations are less complex because creep, shrinkage, and temperature are not considered.
In all these calculations, the effects of axial deformation and arch shallowness are considered. Thus, equa-
tions in Figure 4-16 on page 113, which were determined using the force method, are used in calculating
bending moments and deflections at critical sections of the system. These calculations are outlined in
Tables 4-4 and 4-5. Load factors are also given in these tables according to the types of action being con-
sidered. The symbols used for load factors are !D for dead load, !L for live load, and !K for restrained de-
formations. As discussed in Section 4.4.6, material nonlinearity is accounted for by using system stiffness
factor κ which accounts for the softening of the system at ultimate limit states. When appropriate, age-ad-
justed effective modulus of elasticity Eadj is used to account for creep effects.
The middle four-fifths of the span can be designed for the greater of the of the calculated quarter-point or
midspan moments. The remaining outer portions of the span will have larger sectional force demands than
the middle four-fifths of the span. Sectional forces at the arch springing lines are not explicitly described in
the rest of this chapter because they can be estimated from the maximum sectional forces at the quarter
point and midspan. For preliminary design, bending moments at the springing lines can be taken as twice
129
4. Statical Analysis of Arch Bridges
the maximum moments found at the quarter-points or at midspan. Among the five critical live load cases
considered in Figure 4-21, end moments are less than or equal to two times the maximum quarter-point or
midspan moment. Alternatively, a more detailed calculation can be made by using the various redundant
moment equations given in Figure 4-16 on page 113. Once the deck and arch have been designed to meet
the force demands at the quarter-point and midspan, their proportions can be gradually increased as
needed toward the springing lines. The additional amount of concrete needed to increase the capacity of
the deck and arch at the springing lines is typically small relative to the total weight of the superstructure.
Long-term moments Mlong in arch systems are caused by dead loads Mdead, by concrete creep Mcreep, and by
M long (x,t) = M dead (x) + M creep (x,t) + M shrink (x,t) Equation 4-65
The moments caused by dead loads Mdead are greatly influenced by the camber of the arch and method of
construction. Arches should be cambered such that their geometries after elastic deformation are aligned
as closely as possible with the pressure line of the permanent loads, adjusted to the desired arch rise. Even
if arches are perfectly cambered, bending stresses will still occur if the system is restrained while the arch
shortens elastically. Such stresses occur when dead loads are transferred to arches by the systematic lower-
ing of falsework. In this case, axial stresses and axial deformations are introduced while arches are already
in their final, indeterminate state. The bending moments that arise due to uniformly distributed dead load
# " &
M dead (x = 12 L) = +! D % 241 qdead L2 Equation 4-66
$ 1 + " ('
According to Bazant (1972), the effect of concrete creep can be accounted for by using Equation 4-67,
which gives the ratio of long-term stress /(t) to initial applied stress /(t0) as functions of creep coefficient
130
4. Statical Analysis of Arch Bridges
! (t) # (t)
= 1" Equation 4-67
! (t 0 ) 1 + $ (t) % # (t)
This ratio represents the reduction by creep of stresses caused by restrained deformation. Combining the
second term in Equation 4-67 with Equation 4-66 leads to an expression for the reduction by creep of mo-
# (t)
M creep (x = 12 L,t) = !" K M dead (x = 12 L) Equation 4-68
1 + $ (t) % # (t)
Moments caused by the autogenous and drying shrinkage of concrete are given by Equation 4-69, where
An improved method for decentering arches from their forms was developed by Freyssinet for the con-
struction of the Praireal bridge in 1907 (Ordóñez and Antonio 1979). Freyssinet used flat jacks at the arch
crown joint to gradually introduce axial stresses into the arch before it was closed. As hydraulic pressure
was increased in the jacks, abutting sides of the joint pressed apart, causing the two sides of the arch to
shorten while they were still statically determinate. Eventually, the arch would lift off of its centering and
become self-supporting (Mondorf 2006). This method of transferring dead loads to the arch has been ad-
opted in the construction of most large concrete arch bridges ever since. In the construction of the Rio
Parana Bridge in 1965, for example, twenty-eight hydraulic jacks (see Figure 4-22) were activated at the
crown of the arch to introduce a force of 100 MN (Stellmann 1966). This caused the joint to open by 60
mm. Once the thrust in the arch was fully developed, the open joint was wedged so that the jacks could be
deactivated. Then the joint was filled in with a new placement of concrete.
131
4. Statical Analysis of Arch Bridges
Figure 4-22. Flat hydraulic jacks at crown of Rio Parana Bridge. Adapted from Stellmann (1966).
Using this method of releasing the arch from its forms by jacking significantly reduces the bending stresses
in the arch. If the procedure is executed with high precision and control, it is possible to eliminate virtually
all bending moments at closure Mclose (Favre and de Castro San Román 2001):
Applying a concentrated jacking force at the tip of a curved cantilever normally causes bending in the
member. The shape of the two cantilevered arch halves and the arrangement of loads and reactions, how-
ever, allow the unclosed arch to remain free from moment. Free body diagrams of one-half of the unclosed
arch during the crown jacking procedure are shown in Figure 4-23. The weight of the arch q is assumed to
be uniformly distributed. At the initial stage, the weight of the arch q is carried entirely by the falsework.
As the force in the jacks Fjack is slowly increased, the reacting pressure of the falsework qfalse is decreased.
The line of action of the jacking force is deviated downward by the portion of arch weight that has been
unloaded from falsework, resulting in a state of pure axial compression along the arch. When the jacking
force Fjack reaches a value of qL2/8f, the arch breaks free from the falsework and becomes self-supporting.
q ZLSM^LPNO[<+3 q q q
qL qL qL
R 101 R 104 R
2 2 2
Figure 4-23. Free body diagrams of cantilevered arch during crown jacking
132
4. Statical Analysis of Arch Bridges
The change of system at closure, from statically determinate to statically indeterminate, causes a redistribu-
tion of stresses with time. The sum of moments caused by dead loads and creep will be somewhere
between the moments at closure Mclose, and the moments that would have occurred had the bridge been
built on falsework, and released by the lowering of the falsework Mfalse (Favre and de Castro San Román
2001):
" (t)
M dead (x) + M creep (x,t) = ! D M close (x) + ! K $ ( M false (x) % M close (x))
1 + # (t) $ " (t)
Equation 4-71
" (t)
= +! K $ M false (x)
1 + # (t) $ " (t)
Equation 4-71 shows that although the moments at closure can be virtually eliminated through the use of
hydraulic jacks, indeterminate moments in the final system are inevitably reintroduced by creep. Delaying
the closure of the arch crown after the jacks have been activated has the benefit of decreasing the amount
of creep and shrinkage that remains after the system is made indeterminate.
The horizontal reaction Hdead caused by uniformly distributed dead load qdead is given by:
qdead ! L2 1
H dead = + 18 ! Equation 4-72
f 1+ "
The total long-term horizontal reaction Hlong is given by Equation 4-73. The terms representing the effects
of creep Hcreep and shrinkage Hshrink are negative because these actions reduce the compression in the arch.
The long-term, first-order crown deflection of arches $long,0 is given by Equation 4-74. Here, elastic, creep,
and shrinkage deformations all cause the crown of the arch to move downward.
133
4. Statical Analysis of Arch Bridges
If the crown jacking method is used and the arch camber is equal and opposite to the crown deflection
caused by elastic deformations of the arch under dead load Δfcamb=–Δfdead, then Δfdead can be taken as zero
in Equation 4-74. In this case, only creep deflections Δfcreep and shrinkage deflections Δfshrink need to be con-
sidered in Equation 4-74, since only they cause the arch to move off the pressure line. It is possible to in-
crease the camber of the arch to account for the long-term deflections caused by creep and shrinkage. This
would ensure that the arch maintains some nominal clearance below the crown. The additional camber,
however, will not eliminate the bending moments caused by creep and shrinkage, since they are primarily
The short-term moments Mshort that will be considered are those caused by live load Mlive and by uniform
The maximum positive live load moment at midspan can be estimated by applying distributed live load
qlive over the middle third of the span and a concentrated live load Qlive at midspan, as given by:
1 + 10
57
# 1 + 83 #
M live (x = 12 L) = +! L 5
qlive L2 " + !L 3
Qlive L " Equation 4-76
1+ # 1+ #
972 64
The temperature-induced moment at midspan is given by Equation 4-77, where !T is coefficient of thermal
expansion, and ΔT is uniform change in temperature. A drop in temperature results in a positive moment
at midspan.
% EI sys 1
M temp (x = 12 L) = +! K 154 ( "! T # T ) $ $ Equation 4-77
f 1+ &
134
4. Statical Analysis of Arch Bridges
The short-term horizontal reaction Hshort is taken as the sum of horizontal reactions caused by live load
Hlive and temperature change Htemp, as given by Equation 4-78, where a drop in temperature ΔT causes a re-
The first and second terms in Equation 4-79 correspond to bending deformations, while the third term
In the calculation of additional moments caused by second-order effects, it is overly conservative to dir-
ectly superimpose long-term and short-term deflections. The method for combining these deflections de-
scribed in the following section is based on a method proposed by Menn (1990) for straight columns.
The second-order, long-term deflection $long can be calculated using Vianello’s method as given by Equa-
tion 4-80, where $long,0 is first-order, long term deflection (Equation 4-74), Hlong is long-term horizontal re-
action (Equation 4-73), and HE,long is long-term buckling resistance (Equation 4-81).
1
! long (x,t) = ! long,0 (x,t) Equation 4-80
H long (t)
1"
H E, long
135
4. Statical Analysis of Arch Bridges
In Equation 4-81, ,c is resistance factor of concrete, kf is shallowness factor (Equation 4-50), kS is effective
arc length, κ is system stiffness factor, and Eadj is age-adjusted, effective modulus of elasticity. For rein-
forced concrete members, Menn (1990) uses a different effective secant stiffness than shown in the nomin-
For long-term buckling resistance, Menn (1990) uses a long-term effective secant stiffness EI, equal to the
short-term secant stiffness at yield EIy times a creep stiffness factor k,, which accounts for the reduction of
The creep stiffness factor k% is given by Equation 4-83, where Mu, and ,u, are long-term moment of resist-
ance and long-term curvature at ultimate limit state, and My and ,y are short-term moment resistance and
short-term curvature at yield. These quantities vary with the axial force N in the section, and can be solved
using sectional analysis. Design charts of creep stiffness factor k, for typical circular and rectangular solid
M u! !y
k! (N ) = " Equation 4-83
M y !u!
Long-term sectional forces can be calculated using concrete stress-strain diagrams whose strains are
factored by 1++(t),(t). This results in “horizontally-stretched” long-term stress-strain diagrams that are
softer than those for short-term loading. Compared to short-term analysis, using this creep modified
stress-strain curve only slightly reduces the ultimate moment of resistance of the section Mu,≈Mu≈My be-
cause: (1) the assumed compressive strength of the concrete remains the same, and (2) the yield force in re-
inforcing steel is unaffected by creep. This means that the first factor in Equation 4-83 is equal to, or
slightly less than unity. Curvatures at ultimate limit state, however, do increase significantly ,u,>,y, which
in turn results in creep stiffness factors k, that are less than one. Hence, the long-term effective secant stiff-
ness at ultimate EI, is less than short-term secant stiffness at yield EIy.
136
4. Statical Analysis of Arch Bridges
For ultra high-performance fibre-reinforced concrete sections, the effective secant stiffness κ·Eadj·Isys shown
in Equation 4-81 will be used to calculate long-term buckling resistances of arches. Using this effective
stiffness is equivalent to multiplying short-term effective stiffnesses by creep stiffness factors k,, as in
Menn’s method. This is possible to do for ultra high-performance fibre-reinforced concrete members be-
cause: (1) the material is homogenous (i.e. reinforcement is smeared), (2) secant stiffnesses at ultimate limit
state EIu are already being used (as opposed to secant stiffness at yield EIy in reinforced concrete mem-
bers), and (3) using both short-term and long-term stress-strain curves produce the same ultimate mo-
Long-term deflections $long (Equation 4-80) can be expressed as equivalent short-term deflections $eq,0 us-
ing Equation 4-84 (Menn 1990), where Hlong is long-term horizontal reaction (Equation 4-73) and HE,short is
$ " EI sys
H E,short = !c " # 2 Equation 4-85
(k )
2
f " kS
The total first-order deflection $0 is then the sum of the equivalent short-term deflection $eq,0 (Equation 4-
The total first-order deflection $0 can be magnified using Vianello’s method to obtain the total second-or-
1
! (x,t) = ! 0 (x,t) Equation 4-87
H long (t) + H short
1"
H E,short
137
4. Statical Analysis of Arch Bridges
As discussed in Section 4.4.5, the eccentricity e between the line of action of force and centroid of critical
sections (x=a) of a member can be approximated by multiplying the second-order deflection $ of the crit-
ical section by eccentricity constant &, as given by Equation 4-52 on page 122. When sectional forces are
greatest at midspan $(x=0.5L), eccentricity constant & is equal to 0.5, which is the same as in the analysis
The additional moments caused by geometric nonlinearity are the product of eccentricity e and total hori-
zontal reaction H. Adding this quantity to the long-term and short-term moments results in an expression
(
M (x,t) = M long (x,t) + M short (x) + e(x = a,t) ! H long (t) + H short ) Equation 4-88
These second-order moments are the moments carried by the flexible system. Once calculated, the distri-
bution of these moments among deck and arch can be determined using Equations 4-58 and 4-59 on page
124.
To evaluate the adequacy of given arch designs, second-order moments March and corresponding axial
forces N(x=0.5L)=H can be plotted on to member capacity N-M* interaction diagrams calculated using the
general analysis method or simplified design method described in Sections 3.2 and 3.5. The same can be
done for trial deck girder cross-section, where axial force N is taken as the long-term effective prestressing
Only dead and live loads will be used to estimate maximum sectional forces at the quarter-points because
the effects of restrained deformations are near zero at the quarter-points. This method is more or less the
same as the method presented by Menn (1990), except that his method neglects the effects of arch shallow-
The horizontal reaction H caused by uniformly distributed dead load qdead, distributed live load qlive over
half the span, and concentrated live load Qlive applied at the quarter-point is given by:
138
4. Statical Analysis of Arch Bridges
H = H dead + H live
qdead " L2 1 qlive L2 1 Qlive L 1 Equation 4-89
= +! D 1
" + !L 1
" + !L 135
"
1+ # 1+ # 1+ #
8 16 1024
f f f
The bending moment Mlive at the quarter-point caused by the assumed arrangement of live load is given by:
1 + 373 # 1 + 27
32
#
M live (x = 41 L) = +! L 128
1
qlive L2 " + !L 243
Qlive L " Equation 4-90
1+ # 1+ #
4096
The first-order deflection $0 at the quarter-point caused by these live loads is given by:
qlive L4 Q L3
! 0 (x = 41 L) = +1.63"10 #4 + 6.53"10 #4 live Equation 4-91
$ EI sys $ EI sys
The second-order eccentricity e between the line of action of force and centroid of the quarter-point sec-
tion can be calculated using Equation 4-92, where short-term buckling resistance HE,short is calculated from
Equation 4-85 and eccentricity constant & is 0.733, which is the same as in the analysis of straight fixed-pin
e(x = 41 L) = ! " # (x = 41 L)
1
e(x = 41 L) = 0.733" # 0 (x = 41 L) Equation 4-92
H
1$
H E,short
The total second-order moments M are the sum of first-order moments Mlive and additional secondary mo-
ments caused by the maximum displacement of the arch off the pressure line H·e as given by Equation 4-93
(Menn 1990). These second-order moments are the moments carried by the flexible system, and can be
distributed among the deck and arch using Equations 4-58 and 4-59 on page 124.
For arches with low span-to-rise ratios L/f, axial forces N at the quarter-points can be significantly greater
than the horizontal reaction H. Axial forces N at the quarter-point can be calculated by:
139
4. Statical Analysis of Arch Bridges
2
!2f $
N (x = L) = H 1 + # &
1
Equation 4-94
4
" L %
To evaluate the adequacy of given arch designs, second-order moments March and corresponding axial
forces N(x=0.25L) can be plotted on to member capacity N-M* interaction diagrams calculated using the
general analysis method or simplified design method described in Sections 3.2 and 3.5. The same can be
done for trial deck cross-section, where axial force N is taken as the long-term effective prestressing force
In this chapter, a large amount of technical material has been presented regarding the statical analysis of
concrete arch bridges. The most important topics and insights are summarized below:
• Three-hinged, two-hinged, and fixed arches were compared in terms of: bending moment diagrams
caused by concentrated load, distributed loads, and restrained deformations (Figures 4-6, 4-7, and 4-
10 on pages 93-97), bending moment influence lines and envelopes (Figures 4-8 and 4-9 on pages 95-
96), and primary buckling modes and buckling resistance (Figure 4-3 on page 87).
• Only fixed arch systems will be considered for the development of new concepts in Chapter 6, be-
cause: (1) fixed arches have higher resistances against buckling than hinged arches, which allows for
lighter and more slender members to be used, (2) fixed arches have higher degrees of redundancy
than hinged arches, which is favourable in terms of structural reliability, and (3) fixed arches are more
commonly built than hinged arches, which suggests that they may have some intrinsic economic ad-
• A simplified statical system consisting of a deck girder, spandrel columns, and an arch was defined
using the following three assumptions: (1) spandrel columns are pin-connected to the arch and deck,
(2) the flexural rigidity of the arch projected on to the horizontal axis is constant over the entire span
and is equal to the flexural rigidity of the arch at the crown, and (3) vertical deflections of the deck
140
4. Statical Analysis of Arch Bridges
and arch are equal along the span. Assumption 1 is shown to be conservative, based on the results of a
simple parametric study on systems with varying distributions of flexural rigidity among deck, arch,
and spandrel columns (Figure 4-15 on page 106). Assumption 2 simplifies the bending behaviour of
arches under pure antisymmetric loading to that of constant depth beams with pin supports at mid-
span. Assumptions 3 leads to expressions for distributing flexible system moments among arch and
deck:
EI crown EI
M arch (x) = M (x) and M deck (x) = M (x) deck
EI sys EI sys
• The force method was used to solve the three redundant forces in fixed arch systems. The method of
virtual work was used to calculate both bending and axial deformations. Analytical solutions for vari-
ous critical load cases are summarized in Figure 4-16 on page 113. From these results, a non-dimen-
sional ratio ", called the arch-beam parameter was identified as the parameter that governs the trans-
ition of shallow arch behaviour between pure arch action and pure beam action:
45 EI sys
!= "
4 f 2 EAcrown
• By comparing sectional forces in shallow arches with those in a fixed beams, a minimum modified
slenderness ratio (f threshold value of 20 was proposed. Fixed arch systems with (f >20 carry more
than 90% of permanent loads in pure axial compression along the pressure line:
EAcrown 2 f
!f = 2 f =
EI sys rsys
• Modified slenderness ratio (f also affects the buckling behaviour of shallow arches. Shallow arches
mode when 9.87 $ (f $ 19.8. The buckling load HE, f for the anti-symmetrical mode is:
141
4. Statical Analysis of Arch Bridges
$ "2 '
H E, f = & 0.6 ± 0.4 1 ! 30.7 2 ) * H E # f > 19.8
% #f (
• Geometric nonlinearity is accounted for by using Vianello’s deflection magnification method. Eccent-
ricities e between the deformed line of action of load and centroid of critical section are calculated
from second-order deflections. Eccentricity constants & are assumed to be the same as those in
straight columns with similar deflected shapes, as shown in Figure 4-20 on page 122.
1
e(x = a) = ! " # 0 (x = a)
H
1$
HE
• Material nonlinearity is accounted for by substituting system flexural rigidity EIsys in linear-elastic
calculations with effective system secant stiffness EI!sys=κ·EIsys. At serviceability limit state, stiffness
factor κ is taken as unity. At ultimate limit state, stiffness factor is a function of axial force and is in
the range: 0.5 $ κ $ 0.75. These values are based on the sectional analysis of a hollow, ultra high-per-
• Flexible system moments are defined as the moments that arise from the global deformations of the
integrated arch system. These moments are distributed to the arch and deck according to their relat-
ive flexural rigidities. Fixed system moments, which are defined as the moments that arise from local
deformations of the deck and arch between spandrels columns, must be added to flexible system mo-
• Critical load cases causing maximum sectional forces at midspan, at the quarter-points, and over the
springing lines are illustrated in Figure 4-21 on page 127. Step-by-step design calculations for two of
these load combinations are presented in Section 4.5. Calculations are separated into short-term and
long-term loads effects, which are then combined by expressing second-order, long-term deflections
142
4. Statical Analysis of Arch Bridges
Appendix D.
The simple arch analysis methods presented in this chapter are consistent with the methods presented in
previous works. Based on the author’s review of current literature, there is no single publication that covers
all the material presented in this chapter in a comprehensive way. The significance of this chapter lies in its
utility, consistency, and illustration of fundamental arch behaviour. The concepts and equations presented
in this chapter cover all the major design checks required for the preliminary design of concrete arches. As
such, this chapter provides engineers with a complete and consistent means of learning how to design con-
crete arches. Based on the author’s limited experience, the analysis of arches is not a topic that is com-
monly taught to undergraduate students in civil engineering. Based on what is taught, graduates of civil
engineering would probably approach the design of an arch by creating a numerical model in a frame ana-
lysis program, and then use the resulting force demands to proportion the structure. Because sectional
forces in arches are intrinsically linked to the assumed geometry, distribution of stiffness, and nonlinear
behaviour of arches, a computer model-driven design approach is both inefficient and flawed. The design
of an arch, including the selection of form, layout of geometry, and proportioning of members, should aim
to control the flow of forces, rather than respond to a predefined structural analysis. This chapter, thus,
provides the necessary means for understanding how arches work at an intuitive level, and how they
should be designed.
Before concluding this chapter on the statical analysis of arch bridges, a simple parametric analysis on the
effects of deck-stiffening will be discussed in the next and last section. This analysis attempts to determine
whether or not there are any advantages to using the deck girder to stiffen the arch.
143
4. Statical Analysis of Arch Bridges
In his book Robert Maillart’s Bridges: The Art of Engineering, Billington (1979) discusses the differing views
of the analyst and the designer with regard to the use of deck-stiffening in concrete arch bridges. As an ex-
ample of the viewpoint of the analyst, Billington cites the doctoral work of El-Arousy at the ETH in Zurich
(1942), whose objective was to further improve and simplify the analytical calculation of stiffened arches.
El-Arousy’s work contained two ideas: (1) that generality is the essential basis for application, and (2) that
the structural design of forms is based upon forces predetermined by a prior structural analysis. According
to this viewpoint, the analysis of a structure precedes its design. Billington (1979) contrasted El-Arousy’s
analytical views, which were shared by many experts at the time (and to this day), with those of structural
designer Robert Maillart, whose interest was in the creation of specific bridges rather than the discovery of
general theories of nature. Maillart believed that the dimensions set by the designer controlled the forces,
and that analysis and calculation were subservient to design (Billington 1979). For many of his arch
bridges, Maillart used deck-stiffening to enable him to design very slender, flexible arches. In his design
approach, all nonuniform live loads are carried in bending by the deck girder and all permanent loads are
To illustrate the validity of Maillart’s design approach to deck-stiffened arches, Billington (1979) presents a
simple analysis of flexural stresses in arch ribs. The objective of the analysis is to determine the optimum
arch-to-deck moment of inertia ratio needed to produce the lowest flexural stresses for a given flexible sys-
tem moment. For the analysis, the moment of inertia of the deck girder is held constant, while the depth of
the arch is increased from zero to infinite. The changing geometry is illustrated in Figure 4-24a. Assuming
linear-elastic behaviour, the flexural stress of an arch σarch caused by the arch portion of the flexible system
moment March is given by Equation 4-95, where y is distance away from the centroid of the section and Iarch
M arch " y
! arch = Equation 4-95
I arch
144
4. Statical Analysis of Arch Bridges
b b b
flexural
(b)
stress per !
unit M
bending stress in deck
moment
adapted from
Billington (1979)
stress in arch
2. Broadway
1. Schwand- Bridge (1932)
bach Bridge
(1933)
0
0.001 0.01 0.1 1 10 100
arch-to-deck Iarch
moment of Iarch 1
inertia ratio maximum arch = Ideck
stress occurs when Ideck 2
Figure 4-24. Influence of deck-stiffening on flexural stresses in the arch and deck
Figure 4-25. Schwandbach Bridge and Broadway Bridge. Adapted from Bill and Maillart (1955) and Ostrander and
Oliver (1987). Elevation views are drawn at 1:1000 scale and sections views are drawn at 1:500 scale.
145
4. Statical Analysis of Arch Bridges
Assuming a rectangular arch section with width b and depth darch, distances to extreme fibres ytop and ybot,
darch
ytop = ybot = Equation 4-96
2
3
bd arch
I arch = Equation 4-97
12
Substituting Equations 4-20 (page 103), 4-96 and 4-97 into 4-95, and simplifying gives:
6darch
! arch = Equation 4-98
bd + 12I deck
3
arch
The derivative of arch flexural stress (/arch)! with respect to darch is given by:
" 6 ( bd arch
3
+ 12I deck ) # 6darch $ 3bd 2arch
( arch )
! = Equation 4-99
(bd arch + 12I deck )
3 2
Setting (/arch)! to zero and rearranging gives the following critical point (Billington 1979):
I arch 1
= Equation 4-100
I deck 2
The critical point is labelled in Figure 4-24b, corresponding to maximum flexural stresses in arches caused
by given flexible system moments M. The vertical axis measures flexural stress and increases upwards from
zero at the origin. The scale of the axis is unspecified because only the shapes of the stress curves is relev-
ant to this analysis. The horizontal axis measures arch-to-deck moment of inertia ratio Iarch/Ideck, which in-
Billington (1979) observed that the change in arch stress (solid curve in Figure 4-24b) with respect to the
ratio of arch-to-deck moment of inertia depends on the initial starting form of the design. If a flexible arch
(Iarch/Ideck < 0.5) is envisioned by the designer, then flexural stresses in the arch are reduced by decreasing the
146
4. Statical Analysis of Arch Bridges
depth of the arch. This is the case for Maillart’s deck-stiffened Schwandbach Bridge, built in Switzerland in
1933 (see Figure 4-25). If a stiff arch (Iarch/Ideck > 0.5) is envisioned by the designer, then flexural stresses in
the arch are reduced by increasing the depth of the arch. This is the case for the Broadway Bridge built in
Saskatchewan in 1932 (Figure 4-25). There are practical limitations on both the minimum and maximum
depth of arches. Minimum depths are governed by bucking of the arch between spandrel columns. Max-
Although Billington’s (1979) analysis gives some validity to the deck-stiffened arch form, it is problematic
in several ways. The first problem is in his visual display of results, as shown in Figure 4-26. Although he
does not explicitly state that deck-stiffened arches are inherently more efficient than partially deck-sti-
ffened or self-stiffened arches, it is not difficult for readers to erroneously draw this conclusion due to the
way the results are presented. By omitting stress values between 0 and 6 kg/cm2, Billington’s graph overem-
phasizes the differences in arch stress between Maillart’s Schwandbach Bridge and “typical American
designs of of 1931” (the Canadian Broadway Bridge is substituted here as an example). Also, the horizontal
axis is cut-off prematurely at a arch-to-deck moment of inertia ratio of 1, which dismisses the self-stiffened
arch form as a viable solution. These concerns of visual presentation are addressed in Figure 4-24b by
8.5
8.0
7.5
7.0
6.5
Schwandbach
Bridge design
6.0
0.06 0.20 0.50 1.00
Stiffness ratio Iarch / Ideck
Figure 4-26. Billington’s graph of arch stress versus stiffness ratio between arch and deck.
Adapted from Billington (1979).
147
4. Statical Analysis of Arch Bridges
The second problem with Billington’s analysis is that the minimization of arch stress is assumed to be a val-
id measure of efficiency. Billington’s analysis presupposes that: (1) the minimization of arch stress will res-
ult in the direct minimization of arch weight, and that (2) the minimization of arch weight is favoured over
the minimization of deck weight. The first assumption is only valid when only solid sections are con-
sidered. For solid sections, the only way to increase flexural strength (i.e. reduce bending stress) is to make
arches deeper and hence to add weight. In this case, the increase in weight is equal to the change in depth
times the total width of the section. If more efficient hollow box sections are considered, then it is possible
to increase flexural strength (i.e. reduce bending stress) by adding much less weight. In this case, the in-
crease in weight is equal to the change in depth times the sum of web thicknesses.
Regarding the second assumption, the minimization of arch weight is most critical for cast-in-place on
falsework methods of construction: the lighter the arch, the lighter the required falsework. This method re-
quires that falsework must be designed to carry 100% of the weight of the forms and concrete before the
arch cures and starts to carry its own weight. Hence, there are clear economic incentives for reducing the
weight of the arch. Maillart was aware of these incentives, and designed many of his arches with the intent
of reducing load demands on falsework (Billington 1979). If modern precast, cantilever methods of con-
struction are considered, these economic incentives of minimizing arch weight (as opposed to the total
weight of structure) are less apparent. Precasting allows for stresses to be applied to the arch as soon as
they are erected, reducing the demands on temporary works relative to cast-in-place on falsework. Minim-
izing the flexural rigidity of the arch reduces its resistance against buckling during construction, which
might increase demands on temporary works. Hence, the optimal distribution of flexural stresses, weight,
and stiffness among the deck and arch cannot be determined through the optimization of any one single
Billington’s analysis can be extended by calculating stresses in the deck. If the deck girder is assumed to be
rectangular and to have the same width as the arch, flexural stresses in the deck can be calculated expli-
citly. Deck stresses are shown in Figure 4-24b as a dashed curve. This curve gives a new perspective on the
148
4. Statical Analysis of Arch Bridges
efficiency of the system, since deck stresses become larger as the moment of inertia of the arch is reduced.
Thus, there is a penalty associated with reducing stresses in the arch through the use of deck-stiffening.
These increases in flexural stress in the deck girder were acknowledged by contemporaries of Maillart,
such as Ketchum (1934). Maillart’s contemporaries saw these relatively high deck stresses as being an in-
herent inefficiency of the deck-stiffened arch form. In doing so, they failed to see the potential gain in con-
struction efficiency that Maillart saw in minimizing the weight of the arch.
The third problem with Billington’s analysis is that the effects of dead load are not considered. Dead loads
are designed to be carried in pure compression along the arch, causing uniform compressive stresses.
These stresses act as a form of effective prestress for the arch, which in turn increases the flexural capacity
of the arch. Flexural capacity is increased because precompression stresses in the arch delay the onset of
tensile stresses along the flexural tension face of the arch. Thus, the optimization of system geometry is
achieved not by minimizing arch stress, but by maximizing the use of the decompression force to carry
flexural stresses. The same can be done for the deck girder using prestressing steel. The amount of
prestressing steel needed to offset flexural stresses in the arch, however, may in some cases be overly large
and impractical. Deck prestressing requires additional materials, detailing, and labour. The natural
prestress in the arch, on the other hand, is obtained essentially for free. For these reasons, it is more effi-
cient to carry flexible system moments in arches rather than in deck girders.
This design strategy of using the arch prestress and prestressing the deck girder can be observed in some of
Menn’s Swiss concrete arch bridges built in the 1960s. Although he was aware of Maillart’s innovative and
economical deck-stiffened arch designs, Menn moved in a different direction, proportioning some of his
arches to be partially deck-stiffened, rather than fully deck-stiffened. For the design of the Nanin and
Rhine Bridges (see Figure 4-27), Menn chose to prestress the deck girder and proportioned the arch to
149
4. Statical Analysis of Arch Bridges
Figure 4-27. Swiss concrete arches designed by Menn. Adapted from Vogel and Marti (1997).
Elevation views are drawn at 1:2000 scale and sections views are drawn at 1:400 scale.
As discussed above, Billington’s arch stress analysis has limited applicability because its conclusions draw
upon assumptions that are not always valid. To broaden its applicability, a modified version of the analysis,
shown in Figure 4-28, is proposed. First, the variation in assumed geometry is modified (see Figure 4-28a).
Hollow box sections are chosen to better reflect the type of sections considered in modern concrete arch
systems. Arch depth is increased as deck depth is decreased, such that (1) the sum of cross-sectional areas
of arch and deck remains constant, and (2) the widths and thicknesses of both top and bottom slabs of arch
and deck remain the same. Effectively, the webs of the arch and deck sections are linearly exchanged
between the two sections. This improves the comparison between self-stiffened and deck-stiffened arch be-
cause the total volume of concrete is kept constant. In Billington’s analysis, the total volume and weight of
the system were rapidly increasing as the arch was made deeper. The second modification is that the hori-
zontal axis is expressed as the ratio of arch moment of inertia to system moment of inertia with a theoret-
ical range between 0 and 1. These lower boundary does not actually exist, since it implies that the arch has
zero stiffness. The upper boundary can exist if the deck girder is simply supported between spandrel
150
4. Statical Analysis of Arch Bridges
b b b
flexural
(b)
!
stress per
unit M
bending
moment
stress in deck stress in arch
adapted from
Billington (1979)
0
0 0.2 0.4 0.6 0.8 1
arch-to-system
moment of Iarch
inertia ratio Isys
The modified stress analysis results show that arch stresses are minimum when Iarch/Isys & 0 and are maxim-
um when Iarch/Isys & 0.85. Deck stresses are minimum when Iarch/Isys & 1 and are maximum when Iarch/Isys &
0.15. These results still do not provide any definitive conclusions regarding the optimal distribution of
flexural rigidity among the arch and deck. Some broad conclusions can be drawn if the relative magnitudes
• If compressive stresses in the arch caused by permanent load (i.e. arch prestress) are much higher
than flexural tensile stresses caused by maximum bending moments, then it is probably best to move
towards self-stiffened arch systems. This is because: (1) arch flexural stresses will be offset by precom-
pression in the arch, and (2) flexural stresses in the deck will be minimized. The penalty of increased
151
4. Statical Analysis of Arch Bridges
arch weight during construction is much less pronounced since efficient hollow, box sections are
• If compressive stresses in the arch caused by permanent load are much lower than flexural tensile
stresses caused by bending moments, then it is probably best to move towards deck-stiffened arch
systems. Because of the low prestress, the flexural capacity of the arch will be low. Thus, it is advant-
ageous to protect the arch by distributing most of the flexible system moments to the deck. For the
deck, flexural capacity can be increased as needed by the use of prestressing steel.
152
Chapter 5. Comparative Study of 58 Concrete Arch Bridges
This chapter describes a comparative study of fifty-eight concrete arch bridges. These bridges were com-
piled to characterize a baseline of existing concrete arch technology. These bridges represent designs that
have been feasibly built using conventional reinforced concrete as the primary building material. These
bridges respond to a variety of geometrical site requirements, and represent design solutions that satisfy
the serviceability, strength, and stability requirements imposed on highway traffic bridges. Trends of sec-
tional properties and global geometries among bridges in the study will be compared with new concepts
The primary objectives of this study are (1) to describe and summarize the state-of-the-art of concrete arch
bridge technology in the form of a consistent and comprehensive database, and (2) to identify and ration-
alize trends among the geometries of the concrete arch bridges being studied.
The compiled database describes each concrete arch bridge graphically, quantitatively, and qualitatively.
These descriptions along with trends observed from them can be used as references and design tools for
designers. This database can also serve as a tool for evaluating structural efficiency and feasibility. New
design concepts for arch bridges, such as those presented in Chapter 6, can be compared to bridges in the
153
5. Comparative Study of 58 Concrete Arch Bridges
Databases of concrete arch bridges in current literature have limited utility for designers because of their
lack of information on the sectional properties and geometric ratios of each bridge. Most publications re-
viewed on this topic were concerned with record span lengths of arches and innovative construction tech-
niques. For example, Chen (2007) discusses concrete arch bridges built in China and provides a table with
the names, years of completion, spans, structural types, and construction methods of thirteen arch bridges.
Ewert (1999) provides a table with the names, countries, years of completion, spans, rises, arch depths, and
methods of construction of twenty-eight concrete arch bridges with spans over 200 metres. Further, di-
mensioned drawings of selected arch cross-sections accompany short descriptions of each bridge. Šavor
and Bleiziffer (2008) discuss long span arch bridges in Europe and provide a table with the names, coun-
tries, years of completion, spans, deck widths, and structural types of thirty-nine of the spanning arch
bridges in the world. Mondorf (2006) describes the construction of specific concrete arch bridges and
provides dimensioned drawings for many of them. Selected data from these sources are compiled in Table
5-1.
The information presented in Table 5-1 does not inform the reader on how concrete bridges are typically
proportioned, or on their structurally efficiency relative to one another. These kinds of comparative analys-
is require a rigorous database that considers the sectional properties and geometric ratios of each bridge.
154
5. Comparative Study of 58 Concrete Arch Bridges
Table 5-1. Concrete arch bridge data compiled from reference databases
bridge country year span bridge country year span
in m in m
• bullets mark bridges that are part of the database that is described in Section 5.2
155
5. Comparative Study of 58 Concrete Arch Bridges
This section describes how details and properties of concrete arch bridges from references were processed
for use in the database. The main sources of information were journal articles, conference proceedings,
construction drawings, and textbooks. The name, location, year of completion, key dimensions, construc-
tion method, and drawings were collected for each bridge accepted into the database. A succinct integrated
version of the database, from which all figures in this chapter are adapted from, is provided in Appendix A.
Technical engineering journals report on new and innovative developments in the field, and are con-
sidered primary sources of information for the profession because of the high standards of peer review and
evaluation. In order to produce a database of arch bridges that is accurate, reliable, and representative, a
comprehensive search of authoritative English and German structural engineering journals was carried
out. Table 5-2 lists the journals and volumes that were searched and identifies articles that were found to
have information on the design and construction of concrete arch bridges. In addition to keyword searches
of journal indexes, the table of contents of each issue was inspected. Because the author and assistant were
not fluent in German, each German journal volume was visually skimmed through in search of drawings
A second approach used for finding information on concrete arch bridge references involved inspecting
reference lists of journal articles, conference papers, and textbooks, especially those mentioned in Section
5.1. These lists often included references to old journal articles, which were in some cases retrievable. An-
other major source of information was the proceedings from several International Conferences on Arch
Bridges, which contained many papers dedicated to recently completed projects around the world. Several
construction drawings of Canadian and American concrete arch bridges were obtained from contacts of
the author.
156
5. Comparative Study of 58 Concrete Arch Bridges
ASCE. Journal of Bridge Engineering English 1996-2009 Baxter and Balan 2008: 13(5)
Liu et al. 2002: 7(1)
ASCE. Journal of Structural Engineering English 1983-2009 Laffranchi and Marti 1997: 123(10)
ASCE. Journal of Structural Design and Construction English 1996-2009 none
CSCE. Canadian Journal of Civil Engineering English 1974-2009 Ostrander and Oliver 1987: 14(4)
Proceedings of the ice. Bridge Engineering English 2003-2009 Radic et al. 2006: 159(3)
Structural Engineering International English 1991-2009 Tanner and Bellod 2005: 15(3)
Fonseca 2005: 15(2)
Blinkov et al. 2001: 11(3)
Wang and Au 2001: 11(2)
Cheng 1994: 4(4)
Der Bauingenieur German 1985-2009 Hünlein and Ruse 1985: 60(8)
Bautechnik German 1985-1996, 2004-2009 Reintjes 2005: 82(11)
Beton- und Stahlbetonbau German 1985-2009 Freytag et al. 2009: 104(3)
Viet Tue et al. 2005: 100(11)
Eilzer et al. 2005: 100(3)
Zimmermann 2004: 99(4)
Werschnick 2000: 95(2)
von Wölfel 1999: 94(12)
Ewert 1999: 94(9)
Dajun and Weiqing 1999: 94(4)
Standfuß 1990: 85(5)
Schweizer Ingenieur und Architekt German 1985-1996 Koppel and Walser 1991: 109(11)
Herzog 1990: 108(26)
Bridges were accepted into the database based on their structural system and the amount of information
• bridge uses reinforced concrete as the primary building material (composite decks also accepted),
• bridge deck is supported by the arch from below (half-through and tied arches not accepted), and
• references had scaled drawings with dimensions (elevation and cross-section views).
157
5. Comparative Study of 58 Concrete Arch Bridges
Many of the longest concrete arch bridges in the world have been included in the database. According to
Table 5-1, eighteen of the longest spanning concrete arch bridges in the world have been included in the
database. Each included bridge has been marked with a bullet in the table.
Figure 5-1 shows the distribution of span lengths of bridges accepted into the database. Each bridge is rep-
resented by one mark and is assigned a bridge id (1 being the longest spanning arch and 58 being the
shortest spanning arch in the database). Having reference bridges from all span ranges is important be-
cause observations in Section 5.3 will be made primarily as functions of span length.
bridge ID 51 41 31 21 11 1
Drawings of bridges in the database are shown in Figure 5-2, ordered by length of main span. These draw-
ings were traced from their original sources in accordance with all specified dimensions. Elevation views
are shown at 1:15000 scale, while section views are shown at 1:1500 scale (see Appendix A to view these
drawings at larger scale). Bridges are identified by name, year of completion, and bridge id number. Arch
section views are cut perpendicular to the arch axis, and are shown with dimensions averaged from sec-
tions at the crown and springing line. Section views show outlines of spandrel columns along with single
column sections shown on the right. In general, all variable dimensions of sections are drawn using av-
erage values. In some cases, dimensions of spandrel columns were inferred from elevation and section
views. Unless otherwise specified in references, spandrel columns with thicknesses of 1.5 m or greater were
assumed to be hollow with wall thicknesses of 305 mm. Original sources of information for each bridge in
158
5. Comparative Study of 58 Concrete Arch Bridges
typical section
column section
159
5. Comparative Study of 58 Concrete Arch Bridges
1.! Dajun, D., and L. Weiqing. 1999. Neue Entwicklungen Bei Hochhäusern Und Groβen Brücken Aus Beton in China. Beton- und Stahlbetonbau 94(4): 178-85.
2/11.!Radić, J., J. Bleiziffer, and D. Tkalčić. 2005. Maintaining Safety and Serviceability of Concrete Bridges in Croatia. Bridge Structures 1(3): 327-44.
3.! US Dept. of Transportation FHWA. 2003. AZ-NV HPP 93(3) Hoover Dam Bypass Colorado River Bridge: 60% Submittal. April 4.
4.! Baxter, J. W., A. F. Gee, and H. B. James. 1965. Gladesville Bridge. Proceedings of the Institute of Civil Engineers.! !
5.! Stellmann, W. L. O. 1966. Brücke Über Den Rio Paraná in Foz Do Iguaçú, Brasilien. Beton- und Stahlbetonbau 61(6): 145-49.
6.! Fonseca, A., and F. Mato. 2005. Infant Henrique Bridge Over the River Douro, Porto. Struct. Eng. Int. 15(2): 85-87.
7.! Liebenberg, A.C., and M.G. Latimer. 2001. Bloukrans Bridge. Arch'01: troisième conférence internationale sur les ponts en arc, Sept. 19-21, Paris.
8.! Fritzell, G. 1960. Deflection Measurements on the Sandö Bridge 1942-1958. Sixth Congress of the IABSE.
9.! Kamimura, K., M. Kouno, K. Okada, and T. Tsuka. 2001. Design and Construction of the Tensho Concrete Arch Bridge. Arch'01: Third Int'l Conf. on Arch Bridges, Paris.
10.! Radić, J., Z. Šavor, J. Bleiziffer, and I. Kalafatić. 2008. Ŝibenik Bridge - Design, Construction and Assessment of Present Condition. Chinese-Croatian Joint Colloquium:
Long Arch Bridges, July 10-14, Brijuni Islands, China.
12.! von Wölfel, R. 1999. Die Talbrücke Über Die Wilde Gera. Beton- und Stahlbetonbau 94: 546-51.
13.! State of Nevada Department of Transportation. 2000. B-1948-N/S Galena Creek Bridge. April 19.
14.! Šavor, Z., N. Mujkanović, and G. Hrelja. 2008. Design and Construction of Krka River Arch Bridge. Chinese-Croatian Joint Colloquium Long Arch Bridges, Brijuni Islands.
15.! Radić, J., Z. Šavor, V. Prpić, M. Friedl, and Ž. Žderic. 2008. Design and Construction of the Maslenica Highway Bridge. Chinese-Croatian Joint Colloquium, Brijuni Islands.
16.! Šavor, Z., N. Mujkanović, G. Hrelja, and J. Bleiziffer. 2008. Reconstruction of the Pag Bridge. Chinese-Croatian Joint Colloquium: Long Arch Bridges, Brijuni Islands.
17.! Wang, J. J., and F. T. K. Au. 2001. Modong Hongshui River Bridge, China. Structural Engineering International 11(2): 101-03.
18.! Aigner, F. 1968. Stahlbeton Bogenbrücken Auf Der Österreichischen Brenner Autobahn. Der Bauingenieur.
19.! Igarashi. 1976. Die Hokowazu-Brücke in Japan, Ein Stahlbetonbogen Im Freien Vorbau. Der Bauingenieur (8).
20.! Anonymous. 1978. Communications Techniques Belges 8. Annales des Travaux Publics de Belgique, No. 1-2.
21.! Cheng, K. M. 1994. The Pitan Bridge, Taiwan. Structural Engineering International 4(4): 231-34.
22.! Standfuβ, F. 1990. Nationalbericht: Brücken in Der Bundesrepublik Deutschland. Beton- und Stahlbetonbau (5): 106-13.
23.! Wöβner, K., H. Gebhardt, R. Schnabel, and H. Wörner. 1979. Die Talbrücke Rottweil-Neckarburg Im Zuge Der A81. Beton- und Stahlbetonbau (10-11).
24.! Bouchet. 1964. L'Echangeur La Araña Du Croisement Des Autoroutes Nord-Sud Et Est-Ouest À Caracas (Venezuela). La Technique des Travaux (Belgique).
25.! Hünleim, W., and P. Ruse. 1985. Ein Neues Verfahren Für Den Bau Von Bogenbrücken Dargestelit Am Bau Der Argentobelbrücke Würgau. Der Bauingenieur (12).
26.! Žderic, Ž., A. Runjić, and G. Hrelja. 2008. Design and Construction of Cetina Arch Bridge. Chinese-Croatian Joint Colloquium on Long Arch Bridges, Brijuni Islands.
27.! Koppel, A. J., and R. Walser. 1991. Hundwilertobelbrucke: Ein Bemerkenswerter Neubau. Schweizer Ingenieur und Architekt 109(11): 250-56.
28.! Taiwan Area National Expressway Engineering Bureau, MOTC. 1999. (In Mandarin) The Second Freeway: An Anthology of Particular Bridges.
29.! BC Ministry of Transportation and Highways. 1995. R6-V433-3051 Big Qualicum River Upstream Bridge No. 3051 (Drawings).
30.! Goodyear, D., N. Smit, J. Heacock, and S. Starkey. 2001. Respect for Tradition: The New Crooked River Gorge Bridge. Arch'01: Third Int'l Conf. on Arch Bridges, Paris.
31.! Podolny, W., and J. M. Muller. 1982. Construction and Design of Prestressed Concrete Segmental Bridges. New York: Wiley.
32.! Inaudi, D., A. Rufenacht, B. Arx, H.P. Noher, and S. 2002. Monitoring of a Concrete Arch Bridge During Construction. Proc. SPIE International Society for Optical Eng.
33.! Vogel, T, and P. Marti. 1997. Christian Menn, Brückenbauer. Basel, Switzerland: Birkhäuser Verlag.
34.! State of California Department of Public Works Division of Highways. 1932. Bridge Across Bixby Creek (as Built Plans) Doc. No. 50001001.
35.! Vogel, T, and P. Marti. 1997. Christian Menn, Brückenbauer. Basel, Switzerland: Birkhäuser Verlag.
36.! Eilzer, W., W. Schmidtmann, and R. Jung. 2005. Die Wirrbachtalbrücke. Beton- und Stahlbetonbau 100(3): 236-40.
37/55/57/58. Bill, M., and R. Maillart. 1955. Robert Maillart. Translated by W. P. M. K. Clay. Zürich, Switzerland: Girsberger.
38/50/56. Sinotech Engineering Consultants, Ltd. Personal correspondence, August 2007.
39.! Mörsch, E. 1909. Die Gmündertobel-Brücke Bei Teufen Im Kanton Appenzeil. Schweizerische Bauzeitung 53 (7): 81.
40.! Schwaab, E., and A. Gattner. 1953. Straβenbrücke Über Das Tiefe Tal Bei Rosshaupten, Allgau. Beton- und Stahlbetonbau (11).
41.! Tonello, J., and P. Dal-Palu. 2001. Le Pont Du Triple Saut (the Triple Jump Bridge). Arch'01: troisième conférence internationale sur les ponts en arc, Sept. 19-21, Paris.
42.! State of California Department of Public Works Division of Highways. 1996. Bridge Across Russian Gulch, Earthquake Retrofit Project No. 714.0.
43.! Petrangeli, M.P. 2007. Prefabrication of Medium Span Arch Bridges. Arch'07: 5th Intʼl Conference on Arch Bridges, Sept. 12-14, at Madeira, Portugal.
44.! Zimmermann, W., and N. im Gailtal. 1999. Der Bau Der Stampfgrabenbrücke. Beton- und Stahlbetonbau 99(4): 304-10.
45.! State of California Department of Public Works Division of Highways. 1932. Bridge Across Rocky Creek (as Built Plans) Doc. No. 50001081.
46.! Baxter, D. J., and T. A. Balan. 2008. Design of the Fulton Road Bridge Precast Segmental Concrete Arches. Journal of Bridge Engineering: 476-82.
47.! Blinkov, L. S., E. Cosolo, and S. N. Valiev. 2001. Rehabilitation of a Bridge Over the Matsesta River, Russian Fed. Structural Engineering International 11(3): 181-83.
48.! Ostrander, J. R., and D. C. Oliver. 1987. Construction of the Broadway Bridge At Saskatoon, Saskatchewan, in 1932. Canadian Journal of Civil Engineering 14: 429-38.
49.! State of California Department of Public Works Division of Highways. 1937. Bridge Across Big Creek (as Built Plans) Doc. No, 50001093.
51.! Viet Tue, N., F. Dehn, T. Schliemann, K. H. Reintjes, and F. Tauscher. 2005. Anwendung Von Hochleistungsbetonen Bei Der Bogenbrücke Wölkau Im Zuge Der Bab A17.
Beton- und Stahlbetonbau 100(11): 931-38.
52.! Tanner, P., and J. L. Bellod. 2005. Widening of the Elche De La Sierra Arch Bridge, Spain. Structural Engineering International (3): 148-50.
53.! Tandon, M. 1995. Arch Bridge At Dodan Nallah. Proceedings of the First International Conference on Arch Bridges, Bolton, UK.
Table 5-3 shows the year of completion, location, construction method, and global geometry of bridges in
the database. Geographical locations, expressed as latitudes and longitudes, were found by searching satell-
ite images online using Google Maps (2009). Measurements of span, rise, height over arch crown, and total
spandrel column length measurements were read from drawing dimensions and textual descriptions, or
scaled from drawings. The height over arch crown is measured from mid-depth of arch crown to the top of
roadway. Total spandrel column length is the sum of all column lengths between springing lines, excluding
those directly above the springing lines. Arch fixity refers to the number of hinges present in the structural
system.
160
5. Comparative Study of 58 Concrete Arch Bridges
G means value not t timber scaffolding / centering 3-hinged arch k=0.54
applicable for s steel scaffolding / centering 2-hinged arch k=0.50
comparison cantilever using cable-stayed IB0/,<.35
cantilever using effective truss k=effective arc length factor
? means value pre-erected reinforcement arch
unknown rotation method
161
5. Comparative Study of 58 Concrete Arch Bridges
Table 5-4 shows the section types and geometric properties of the decks, arches, and spandrel columns of
bridges in the database. Deck and arch sections were categorized as one of the following: rectangular solid
sections (slabs or solid ribs), voided slab sections, tee or slab-on-girder sections, single or multi-cell box
girder sections, or composite concrete and steel sections. Slab-on-girder sections were assumed to act
compositely, as references did not usually provide enough information to suggest otherwise. Deck widths
were measured from edge to edge of the section, including pedestrian corridors and traffic shoulder. This
was done for uniformity in data, as not all sources marked the location of traffic lanes. Cross sectional
areas and moments of inertia of deck and arch sections were calculated by discretizing sections into sets of
simple rectangular elements. For variable depth arches, the depth used for calculation was the average
between maximum and minimum depths. For transformed moments of inertia of composite sections, a
modular ratio of 8 was assumed. Deck continuity refers to the support condition over the spandrel
columns. Discontinuous deck refers to decks that are simply supported between spandrel columns. In these
bridges, all flexible system moments are carried by the arch. Continuous deck refers to decks that are con-
tinuous over spandrel columns, which allows for their participation in resisting flexible system moments.
Table 5-5 shows calculated values of various span-to-rise ratios, span-to-depth ratios, slenderness ratios,
moment of inertia ratios, equivalent slab thicknesses, and quantities related to the effects of dead and live
load. All these quantities are related to some aspect of arch structural behaviour. Each quantity will be de-
162
5. Comparative Study of 58 Concrete Arch Bridges
@ means value not rectangular section -- discontinuous —average of max and min depth
applicable for v voided slab section over columns — max depth = depth + var. of spandrel columns used to
comparison T tee or slab-on-girder section — continuous — min depth = depth var. calculate area and moment of inertia
n n-cell box girder section over columns — depth measured to arch axis
? means value composite steel box section — depth used to calculate area and moment of inertia
unknown modular ratio of 8 assumed for transformed moment of inertia
163
5. Comparative Study of 58 Concrete Arch Bridges
normalized
equiv. system moment
slab thickness of inertia
by system inertia
equiv. slab thickness by volume
column inertia : system inertia
arch inertia : system inertia
164
5. Comparative Study of 58 Concrete Arch Bridges
This section describes trends observed from the recorded data presented in Section 5.2. In the following
analyses, arch span length is treated as the primary independent variable. Bridges in the database are inter-
preted as unique, feasible design solutions that respond to given span length requirements.
This study is biased toward concrete arch bridges that were sufficiently detailed in references and accessible
to the author. Because the database draws extensively from journal articles and international conferences
on arch bridges, it can be argued that the collected body of data represents the state-of-the-art of concrete
arch bridge technology, especially in North America and Europe. Generally speaking, most new bridges
are designed with proportions that do not depart appreciably from past completed works. In other words,
new bridges are typically designed in reference to or with knowledge of designs that have been previously
lightness, or method of construction, then it is likely that information on its design would have been publ-
ished, and thus included in this database. Based on these observations, it is assumed that the selective
sample of 58 concrete arch bridges in this study is representative of nearly all concrete arch bridges built in
modern times.
165
5. Comparative Study of 58 Concrete Arch Bridges
The geographic locations of bridges included in the database are shown in Figure 5-4. The number of
bridges in the database by country is tabulated in Table 5-6. Swiss, German, Croatian, and Taiwanese con-
crete arch bridges are well represented in the database, as well as those along the west coast of the United
States. Arch bridges in China are poorly represented in the database. According to a study by Chen (2007),
151 concrete arch bridges with spans greater than or equal to 100 metres were built or under construction
as of March 2006. Detailed information about Chinese arch bridges were inaccessible to the author, except
ÇÇ Ç
Ç Ç
Ç Ç Ç
Ç ÇÇ
A
Ç Ç ÇÇ ÇÇÇ
ÇÇÇ Ç ÇÇ ÇÇ ÇÇÇÇ Ç Ç
Ç Ç Ç
Ç B ÇÇ Ç
Ç ÇÇ
Ç Ç
Ç
Ç B
ÇÇ
Ç Ç Ç
Ç
Taiwan Europe
Switzerland 9 Brazil 1
Germany 8 France 1
United States 8 India 1
Croatia 7 Italy 1
Taiwan 5 Portugal 1
Canada 3 Russia 1
Austria 2 South Africa 1
China 2 Spain 1
Japan 2 Sweden 1
Australia 1 Venezuela 1
Belgium 1
166
5. Comparative Study of 58 Concrete Arch Bridges
bridge ID 51 41 31 21 11 1
2010 s ⩚ Colorado
s ⩚
⦿ ⩚ ☒
s
2000 s s ⩚ ⦿ ⩚ ⩚ ☒
s ⩚ ⦿
s
s
1990 ⦿
Wanxian-Yangtze
Argentobel ⦬
⩚
1980 ☒ ☒ Krk I
⩚ s
☒
1970
t s⩚
⩚ s s
? Gladesville
1960
t ⩚
1950
t timber scaffolding / centering t Sandö
s steel scaffolding / centering 1940 ?
⩚ cantilever using cable-stayed t?
?
☒ cantilever using effective truss t t t ⦿
⦿ pre-erected reinforcement arch 1930 tt t
⦬ rotation method ⦿
t Elche de la Sierra
1920
Figure 5-5. Year of completion and method of construction versus span length
Figure 5-5 shows the years of completion and methods of construction of bridges in the database versus
span length. Bridges with the longest span for each method of construction have been labelled. Most con-
crete arches in the first half of the 20th century were built cast-in-place on timber falsework. This falsework
would support timber centering, which is the curved temporary framing that is used to support the form-
work for the concrete arch. This labour intensive construction technique limited the lengths that concrete
arches could span economically. The longest spanning arch in the database using timber scaffolding is the
Sandö bridge in Sweden, with a span of 264 metres. Figure 5-6 shows the complex geometry and large
volume of throwaway materials needed to construct the Sandö arch. This method of construction causes
Figure 5-6. Sandö Bridge construction using timber scaffolding. Adapted from Mondorf (2006).
167
5. Comparative Study of 58 Concrete Arch Bridges
According to trends in Figure 5-5, steel replaced timber as the preferred material for scaffolding and cen-
tering after 1960. In 1965, the Gladesville Bridge in Australia was constructed using structural steel to form
temporary piers and centering as shown in Figure 5-7. At the time, this bridge held the world record for
longest concrete arch span ever built. The arch is composed of four arch ribs that were built-up from
precast segments.
Figure 5-7. Gladesville Bridge construction using steel centering. Adapted from Seegers (1963).
Instead of using steel arch ribs solely as temporary centering, they can also be designed to remain in place
as permanent reinforcement for the arch. An early example of a bridge that used this technique is the
Elche de la Sierra Bridge in Spain, built in 1927 (Tanner and Bellod 2005). During construction, a pilot
truss composed of steel members were used to support formwork for the arch as shown in Figure 5-8. In
the final stages of construction, the steel pilot truss was embedded in the concrete, making a composite
arch section. This technique reduced the complexity of falsework and the amount of throwaway material.
A similar technique, albeit at a much larger scale, was used 70 years later to construct the Wanxian-
Yangtze bridge, which is currently the longest spanning concrete arch bridge in the world. Instead of using
falsework to support the steel pilot truss as in the Elche de la Sierra Bridge, towers were built above the
springing lines from which cables were stayed. These stays supported the cantilevered steel trusses as they
168
5. Comparative Study of 58 Concrete Arch Bridges
The cantilevering technique used in the Wanxian-Yangtze Bridge can also be used to erect precast or cast-
in-place concrete arch segments, although the weight of concrete cantilevers are normally heavier than
steel ones. This cantilevering method of construction using towers and stays was recently used in the con-
struction of the Colorado River Bridge as shown in Figure 5-9 (US DOT FHWA 2003). Stays emanating from
the towers support the arch cantilevers at several locations along the arch. The forces in these stays are
equilibrated by reactions from the temporary towers, and back stays that are anchored into the rock. Based
on photos of the actual construction of this bridge, all stays were directed upwards to the top of the tower
Figure 5-9. Colorado River Bridge construction using tower and stay method. Adapted from usdot-fhwa (2003).
169
5. Comparative Study of 58 Concrete Arch Bridges
Another variation of cantilevering in the construction of arches involves creating an effective truss, in
which the uncompleted arch ribs form the compression chord of the truss. This was done in the construc-
tion of the Krk Bridges in Croatia in 1980. For these bridges, steel ties were used to carry tensile forces
along the top chord and diagonals of the truss, as shown in Figure 5-10 (Šram 1982). Also shown in the fig-
ure is an overhead cableway, which supports cranes that can be used to transport materials to the free ends
of the cantilevers.
In the construction of the Infant Henrique Bridge in Portugal, a similar trussed cantilever method was
used. The tensile forces for the top chord of the truss, however, were carried by the prestressed deck girder,
which was being built simultaneously with the cantilevered arch (Fonseca and Mato 2005). Compared to
using only steel ties, using the deck as part of the truss improves the stiffness and stability of the overall
170
5. Comparative Study of 58 Concrete Arch Bridges
The last method of construction commonly used for concrete arches is the rotation method. The Argento-
bel Bridge in Germany built in 1986 used this method (Hünleim and Ruse 1985). As shown in Figure 5-11,
each concrete arch half-rib was cast in an upright position using slip-forming. The ribs were then rotated
66.5° downward to their final positions. To achieve the rotation, special temporary hinges were built at the
springing lines, and were later fixed with closure pours. Forming the arch upwards, as opposed to out-
wards over the span, gives the advantage of having a smaller construction footprint and better access to the
advancing end of the cantilever. A variation of this method has the arch half-ribs rotated horizontally
rather than vertically. Where the topography of the site allows, this method allows the arch to be built over
land, over the banks at either side of the crossing. These rotation methods of construction are typically
limited to short and medium span arches because of the high degree of movement and bending moments
Figure 5-11. Argentobel Bridge construction using the cantilever rotation method.
Adapted from Hünleim and Ruse (1985).
Figure 5-12 shows the structural depths of arches and decks of bridges in the database versus span length.
Nearly all arches with depths of 2 metres or greater are box girder sections. Of these arches, 19 out of 27
(70%) have box girders with multiple cells. Only 3 out of 57 arches in the database (5%) have tee-shaped
sections. This low quantity is probably because arches must resist moments of similar magnitude in both
positive and negative bending. Tee-shaped sections are typically better suited for systems that are biased
171
5. Comparative Study of 58 Concrete Arch Bridges
bridge ID 51 41 31 21 11 1
3⧈
⧈
6
5 ⧈
3⧈
2⧈
⧈
4 3⧈ 2⧈ 3⧈ 3⧈ 3⧈
⧈
⧈
3⧈
3
⧈ 2⧈
2⧈ 2⧈ 2⧈
2⧈
4⧈ 2⧈ 3⧈
2⧈ ⧈
2⧈ −
2 −
− −− − T
−
−T ⧈ v
deck depth in m 6
5 ⧈
⧈
4
⧆
3 2⧈
⧆
⧈ T
2 T T T T
T ⧆ T
T 4⧈ 4⧈
T
T 2⧈
T T v
− rectangular solid section
T
T ? T T T
T
v voided slab section T TT ⧈ v
T tee or slab-on-girder section 1 ⧈ v
T T T T
n⧈ single or n-cell box-girder section T T TT T − T
−
v
T T
⧆ composite steel-concrete section −
T v
−
−
0
Figure 5-12. Section type and structural depths of arches and decks versus span length
Figure 5-13 shows arch depths and deck depths superimposed on the same plot. The arch depths of the
bridges in the database tend to increase linearly with span length. The two bridges that depart from this
trend are the Pitan and Infant Henrique bridges. The Pitan bridge has no spandrel columns and must resist
large bending moments resulting from the frame action between its fused arch and deck. Thus, deep sec-
172
5. Comparative Study of 58 Concrete Arch Bridges
tions are needed to carry these moments. The Infant Henrique bridge achieves a thin arch by minimizing
bridge ID 51 41 31 21 11 1
Infant-Henrique
0
Figure 5-13. Structural depth of arch and deck versus span length
The deck depths of bridges in the database are less correlated with span length than arch depth. In most
cases, the deck depth is lower than arch depth. This shows that there is a preference among bridges in the
database toward self-stiffened arch systems. Decks of self-stiffened and discontinuous deck arch bridges
tend not to be deeper than 2.5 metres because they do not participate in resisting global flexible system
moments.
Figure 5-14 shows span-to-depth ratios of bridges in the database versus span length. The ratios calculated
for the top graph only consider the depth of the arch. In this case, 48 out of 57 bridges in the database
(84%) have span-to-arch-depth ratios in the range 39 to 77. Thus, this range constitutes the typical range
for span-to-arch-depth ratios in concrete arch bridges. This range includes all bridges with decks that are
simply supported between spandrel columns. The Schwandbach and Infant Henrique Bridges have span-
to-arch-depth ratios of 187, which are nearly twice that of the next highest ratio in the database. These two
bridges achieve thinness in the arch by having stiff decks that attract most of the flexible system bending
moments.
173
5. Comparative Study of 58 Concrete Arch Bridges
bridge ID 51 41 31 21 11 1
max ratio
average ratio 100
min ratio 77
50
39
span-to-system-depth ratio 80
with continuous deck
Schwandbach Infant-Henrique
with discontinuous deck
60
system = arch + deck
max ratio 48
average ratio 40
min ratio
24
20
The ratios calculated for the bottom graph in Figure 5-14 consider the sum of the arch and deck depths, or
system depth, in order to capture overall system behaviour. In this case, 46 out of 55 bridges in the data-
base (84%) have span-to-system-depth ratios in the range 24 to 48. Thus, this range constitutes the typic-
ally range for span-to-system-depth ratios in concrete arch bridges. The total amount of bridges between
the top and bottom graphs is different because the deck depth of two bridges are not known. When the
deck depth is considered, the span-to-depth-ratios for the Schwandbach and Infant Henrique Bridges are
not exceptional compared to the ratios of other bridges in the database. These observations suggest that
span-to-system-depth ratios are better measures of arch bridge proportions than are span-to-arch-depth
ratios.
The moments of inertia of arch, deck, and columns are indicators of flexural rigidity, which is the product
of the elastic modulus and moment of inertia. The flexural rigidities of the arch, deck, and columns de-
termine the distribution of flexible system moments, as previously illustrated in Figure 4-15 on page 106.
Because the elastic modulus of concrete is not known for each bridge in the database, flexural rigidities
174
5. Comparative Study of 58 Concrete Arch Bridges
among bridges cannot be directly compared. Instead, moments of inertia and ratios between moments of
inertia will be used as an approximate means to compare distributions of flexural rigidity of bridges in the
database. Effectively, this is equivalent to assuming that the elastic modulus is more or less equal among
bridges in the database. A typical range for the elastic modulus of conventional concrete is between 25000
and 35000 MPa. This means that the flexural rigidities of sections with equal moments of inertia may in
Figure 5-15 shows the arch and deck moments of inertia of bridges in the database versus span length. For
purposes of comparison, these values are normalized by deck width. Trends seen on the top graph for
these values plotted on a log scale are similar to those in Figure 5-13, which shows structural depths of
arches and decks plotted on a linear scale. The general trend is that arch moments of inertia increase faster
than deck moments of inertia, except for those bridges with stiffening decks.
bridge ID 51 41 31 21 11 1
100
10 1
10 2
10 3
10 4
normalized system 6
moment of inertia in m linear regression for :
Pitan y = 0.0109 x + 0.337
= 3 12 system inertia
r2 = 0.906
deck width Tensho
4
y is normalized system inertia in m
excluded from regression analysis x is span length in m
Colorado r2 LVFRHIÀFLHQWRIGHWHUPLQDWLRQ
2
Figure 5-15. Moment of inertia of arch, deck, and system versus span length
175
5. Comparative Study of 58 Concrete Arch Bridges
System moment of inertia Isys is the sum of arch and deck moments of inertia. A means of normalizing this
quantity is needed to make meaningful comparisons among bridges in the database. The moment of iner-
By setting I=Isys and rearranging, the definition for normalized system moment of inertia hI given by Equa-
12I sys
hI = 3 Equation 5-2
bdeck
Quantities of hI versus span length are plotted in the bottom graph of Figure 5-15. A linear regression ana-
lysis was performed on the data, excluding the Pitan Bridge, which is clearly an outlier. The analysis resul-
Based on this trend line, the data shows that there is a strong correlation between normalized system iner-
tia and span length. The coefficient of determination R2 was found to be 0.906. The two bridges that depart
most from the trend line are the Tensho and Colorado River Bridges. The former has high span-to-rise ra-
tio and probably requires high flexural rigidity to prevent excessive deflections. The latter has relatively
high concrete strength and also has a composite deck. These factors may hinder direct comparison with
the other concrete bridges in the database with respect to bending stiffness in the system.
176
5. Comparative Study of 58 Concrete Arch Bridges
bridge ID 51 41 31 21 11 1
Figure 5-16 shows system radius of gyration of bridges in the database versus span length. The expression
for system radius of gyration rsys, previously given in Equation 4-38 on page 114, is modified as:
I arch + I deck
rsys = Equation 5-4
Aarch
In this modification, modulus of elasticity E is assumed to be constant among deck and arch, and arch
cross sectional area Aarch is taken as an average value rather than the value at the crown Acrown. A linear re-
gression analysis was performed on the data, excluding the Pitan Bridge, which again is clearly an outlier.
The analysis resulted in the trend line given by Equation 5-5, where L is span length in metres:
Based on this trend line, the data shows that there is a strong correlation between system radius of gyra-
tion and span length. The coefficient of determination R2 was found to be 0.913.
177
5. Comparative Study of 58 Concrete Arch Bridges
bridge ID 51 41 31 21 11 1
0.6
18 partially
deck-
0.4 stiffened
0.2
4 deck-
stiffened
0
Figure 5-17. Ratio of arch inertia to system inertia versus span length
Figure 5-17 shows the ratios of arch moment of inertia to system moment of inertia of bridges in the data-
base versus span length. This ratio is used to estimate how flexible system moments are distributed among
arch and deck. As discussed in Section 4.4.1, flexible system moments are distributed to arch and deck in
proportion to their relative flexural rigidity. When the arch inertia to system inertia ratio equals one, the
arch is effectively unstiffened by the deck. This ratio also equals one if the deck is discontinuous over the
columns. Of the bridges in the database, 32 out of 54 (71%) have arch inertia to system inertia ratios greater
than or equal to 0.8, which implies that most are self-stiffened or have discontinuous decks. When this ra-
tio equals zero, the system is stiffened completed by the deck, which carries all flexible system bending
moments. The Infant Henrique bridge is the only deck-stiffened bridge in the database with a span greater
than 100 metres. This design raises the question: why are more long span bridges not built with slender
Figure 5-18 shows the ratios of column moment of inertia to system moment of inertia of bridges in the
database versus span length. Column moments of inertia are calculated based on average dimensions
among spandrel columns. This ratio of inertias affects the magnitude of flexible system bending moments
carried by the deck and arch. As shown in Figure 4-15 on page 106, lower column flexural rigidities corres-
pond with higher flexible system moments in the arch and deck. When this ratio is increased to one and
spandrel columns are monolithically connected to the arch and deck, the system behaves as a Vierendeel
truss, reducing bending moments in the arch and deck, but also increasing moments in the columns.
178
5. Comparative Study of 58 Concrete Arch Bridges
When the ratio is low, the columns behave like pin-ended columns regardless of their connection type,
since very little bending moment can be distributed into the columns. In the database, 39 out of 49 bridges
(80%) have column moment of inertia to system moment of inertia ratios of 0.1 or less. In conceptual
design, spandrel columns are often assumed to be pin-ended, even if they will eventually be detailed as
fixed or monolithic connections. The generally low flexural rigidity of the spandrel columns relative to that
of the system makes this preliminary design assumption acceptable in most cases.
0.1
0.01
0.001
Figure 5-18. Ratio of average column inertia to system inertia versus span length
Arches are subjected to combined compression and bending and are thus susceptible to second-order
effects. The severity of these additional deflections and bending moments depend on the slenderness of the
system. In general, slenderness is a function of unbraced length, support conditions, and radius of gyration
kS
!= Equation 5-6
rsys
Effective arc length kS has been previously discussed in Section 4.3.1 on page 86. Effective arc length
factors k are determined based on the number of hinges present in the arch, and are given by Equation 4-1.
Arc lengths S is given by Equation 4-3, which assumed that the arch axis forms a parabola. System radius
of gyration rsys is given by Equation 5-4, which includes the effect of deck-stiffening.
179
5. Comparative Study of 58 Concrete Arch Bridges
60
40 38
20
Figure 5-19 shows system slenderness ratios of bridges in the database versus span length. In the database,
43 out of 54 bridges (81%) have system slenderness ratios in the range of 38 to 80. Thus, this range consti-
tutes a typical range for system slenderness ratios in concrete arch bridges. The Russian Gulch and Nanin
Bridges have the highest system slenderness with ratios of 99. Bridges with high slenderness, such as these,
are susceptible to second order effects, and thus require structural analyses that account for material and
geometric nonlinearity.
The choice of span to rise ratio is strongly influenced by the geometric requirements of given bridge sites.
Elevations of roadways are often determined by the start and end points of existing or planned transporta-
tion networks or navigation clearances. The location of arch foundations are constrained by site accessibil-
ity and the ability of the ground material to provide adequate reactions. Along with these considerations,
potential arch bridges should respond naturally to the surrounding landscape from a visual perspective.
Figure 5-20 shows the ground and arch profiles for selected bridges in the database. In most of these cases,
the elevation of the arch crown is chosen to be close to the elevation of the higher, shallower slopes at each
end of the bridge site. The springing lines of the arch are positioned to traverse deep valleys and avoid
steep slopes, such as in the Colorado River bridge site. In some cases, the springing lines are placed adja-
cent to edges of bodies of water, while in others, they are not. In the case of the Neckarburg Bridge, mov-
ing the springing lines closer to the edges of the water surface would result in span-to-rise ratios less than
180
5. Comparative Study of 58 Concrete Arch Bridges
3, which is possible to build but lacks in visual appeal. This aesthetic judgement can be made if the viewer
holds the following values: (1) that longer spans are more visually appealing than shorter spans because
they are more daring and impressive, and (2) that less material is more visually appealing than more ma-
terial. In terms of the second value, the ratio of arc length S to span length L increases rapidly as span-to-
rise ratios are decreased below 3. This vertical elongation of the arch may be perceived as an inefficient use
of material.
Figure 5-20. Ground and arch profiles of selected bridges in the database.
All drawn at 1:20000 scale. Inverted triangles mark the surfaces of water.
To illustrate the visual effect of different span-to-rise ratios, variations of the Colorado River Bridge arch
are shown in Figure 5-21. Variation A has a very deep arch with a starting inclination angle of about 60°.
This angle makes the parabolic form similar to an equilateral triangle, accentuating the vertical rising of
the arch rather than its horizontal spanning. In this case, the arch would be difficult to build because the
left slope is steep and not easily accessed. If this arch was built using cast-in-place construction, the high
angle would require top and bottom forms, which is more costly and more complex than casting with bot-
tom forms only. Variations B and C move away from the triangular form, and accentuate the horizontal
spanning of the arch. According to the aesthetic values previously stated, Variations B and C are superior
to Variation A. Visually, Variations B and C are not that different from the original, and respond reason-
ably well to the landscape. Although Variation D spans much farther than the original and may seem more
structurally daring, its aesthetic quality is compromised by the impractical positioning of the left springing
181
5. Comparative Study of 58 Concrete Arch Bridges
line. Variation D overshoots the natural cliff edge on the left side, which makes the economic premium of
Figure 5-21. Variations of span-to-rise ratio based on the Colorado River Bridge. All drawn at 1:20000 scale.
Span-to-rise ratios of bridges in the database versus span length are shown in the top graph of Figure 5-22.
55 out of 58 bridges in the database (95%) have span-to-rise ratios in the range of 2.3 to 8.0. As previously
discussed, the low end of the range is governed by awkward aesthetics and material efficiency. The high
end of this range is probably governed by precedence, economy, and complexity. The Infant Henrique,
Elche de la Sierra, and Wölkau Bridges show that span-to-rise ratios greater than 8.0 are possible to build.
Three possible reasons why concrete arch bridges with high span-to-rise ratios are not commonly built are:
(1) there are fewer bridge sites that are geometrically suitable for arches with high span-to-rise ratios than
for conventional span-to-rise ratios, (2) in terms of economy, variable depth girders become increasingly
competitive in spaces suitable for arches with high span-to-rise ratios, (3) arches with high span-to-rise ra-
tios require large horizontal reaction forces and thus foundations are costly or impractical, and (4) con-
struction requires a high standard of care, given the potentially severe consequences of misalignments of
182
5. Comparative Study of 58 Concrete Arch Bridges
span-to-rise ratio 16
12
Elche de la Sierra Infant-Henrique 11.2
Wölkau
8 8.0
4
2.3
Bronte Creek
0
2000 =4
1000
Krk-II Colorado
0
The bottom graph of Figure 5-22 shows quantities of span squared divided by span length of bridges in the
database versus span length. This quantity is also referred to as the static coefficient. According to Mondorf
(2006), bridge builders used static coefficients to gauge how demanding the construction of a prospective
arch would be. Higher coefficients indicated higher complexity of construction. This coefficient is present
in the expression for horizontal reaction H, as given by Equation 5-7, where q is uniformly distributed
qL2
H= Equation 5-7
8f
For the same uniformly distributed load, the Krk II and Colorado River Bridges have about the same hori-
zontal reaction, even though they differ in span length by 79 m. The Infant Henrique Bridge, has a static
coefficient 2.5 times that of the Krk II and Colorado River Bridges, which is large considering that it spans
183
5. Comparative Study of 58 Concrete Arch Bridges
As discussed in Section 4.4.3 on page 110, modified slenderness ratio (f, which is given by Equation 5-8
(Cai et al. 2009), governs the transition of shallow arches between pure arch behaviour and pure beam be-
haviour. A minimum value of 20 was previously identified as a threshold for efficient fixed arch behaviour.
Shallow fixed arches with (f >20 carry more than 90% of permanent loads in compression along the pres-
sure line, and less than 10% in bending. Decreases in (f cause increases in bending moment demand and
2f
!f = Equation 5-8
rsys
Modified slenderness ratios of bridges in the database versus span length are shown in Figure 5-23. Of the
fixed arches in the database, only 3 out of 51 bridges (6%) have modified slenderness ratios less than 20.
This subset includes the Nan Ke and Pitan Bridges, which do not have spandrel columns and are thus in-
herently more beam-like than other bridges in the database. Most of the other bridges in the database have
ratios much greater than 20, which indicates that they are not shallow and that their crown deflections are
PRGLÀHGVOHQGHUQHVVUDWLR 160
2 rise
=
V\VWHPUDGLXVRIJ\UDWLRQ
120
À[HGDUFKHV
KLQJHGDUFKHV
80
IRUÀ[HGDUFKHV
40 increasing arch action
Nan Ke Pitan 90% arch action
0 Bohlbach increasing beam action
A strong correlation was previously observed between system radius of gyration and span length in Sec-
tion 5.3.4. Using the linear regression trend line given by Equation 5-5, thresholds for efficient arch beha-
184
5. Comparative Study of 58 Concrete Arch Bridges
viour can be expressed in terms of span-to-rise ratios. By substituting Equation 5-5 into 5-8, and imposing
that (f is greater than 20, the following inequality is obtained, where L is span in metres:
L ! L $
< 16.1 # Equation 5-9
f " L + 12.4 m &%
Equation 5-9 gives threshold values for span-to-rise ratio L/f as a function of span length L and is plotted
in Figure 5-24. The figure shows that arches with spans less than 50 m have particularly low span-to-rise
ratio threshold values. For spans greater than 50 m, the L/f threshold values lie around 12:1 and 14:1. Be-
cause Equation 5-9 is based on average system radius of gyration values, the actual threshold curve should
span-to-rise ratio 16
Threshold based on
12 linear regression trend
line of system radius
of gyration.
8
4
Arches with span-to-rise ratios less
than the threshold curve exhibit
0 greater than 90% arch action.
0 100 200 300 400 500
span length in m
Figure 5-24. Efficiency threshold curve based on average system radius of gyration trend line
The interplay of section capacity and demand compounds the problem of arch shallowness and beam-like
behaviour. For a given span length, arches with high span-to-rise ratios have higher sectional forces than
arches with lower span-to-rise ratios. Thus, the former requires deeper and stiffer sections than the latter.
Choosing stiffer sections will tend to increase system radius of gyration rsys, unless flexural rigidity EI and
axial rigidity EA are increased at the same rate. Increases in system radius of gyration rsys cause decreases
in modified slenderness ratio (f, thus moving the system toward beam behaviour. Increased beam beha-
viour will increase overall bending moments in the arch, which may then require increases in flexural ca-
pacity and flexural rigidity. In other words, proportioning shallow arches to be stronger and stiffer in
bending will inherently make them more beam-like. Based on these reciprocative effects, it is likely that
185
5. Comparative Study of 58 Concrete Arch Bridges
systems with high span-to-rise ratios require system radii of gyration that are greater than those predicted
A more conservative span-to-rise ratio threshold curve can be obtained by increasing the assumed values
for system radius of gyration. A subjective high estimate trend line, which serves as an approximate upper
envelope of system radii of gyration, was chosen by increasing the vertical intercept in Figure 5-16 on page
177 until all but three data points were below the line, while keeping the slope of the line constant. The se-
lected high estimate trend line is given by the following equation, where L is span in metres:
Using Equation 5-10, the resulting span-to-rise ratio threshold inequality becomes:
L ! L $
< 16.1 # Equation 5-11
f " L + 48.4 m &%
This new threshold curve (dashed line) is plotted in Figure 5-25, along with the less conservative threshold
curve (solid line) based on the linear regression trend line (Equation 5-9). The curves on the diagram show
that it is always inefficient to design fixed concrete arches with span-to-rise ratios greater than 16:1. The
shaded area on the diagram represents arch systems that exhibit greater than 90% arch action, based on
the subjective high estimate. The shaded area shows that fixed arches with spans less than 50 m are partic-
ularly sensitive to moments caused by shallowness effects, even in the normal range of span-to-rise ratios
of 3:1 and 8:1. Thus it is important that systems with high values of rsys are not used in this range of spans.
For fixed arches with spans greater than 100 m, the threshold curve ranges between 12:1 and 14:1. The data
point for the Infant Henrique Bridge, which is among the flattest concrete arches in the world, lies just be-
low the threshold curve. Thus there still remains an opportunity to design even flatter, efficient arches that
186
5. Comparative Study of 58 Concrete Arch Bridges
Threshold based on
linear regression trend
line of system radius
of gyration.
span-to-rise ratio 16
12 Threshold based on
Infant Henrique Bridge subjective high
estimate trend line.
8
bridges in database
4
Arches with span-to-rise ratios less
than the threshold curve exhibit
0 greater than 90% arch action.
0 100 200 300 400 500
span length in m
Figure 5-25. Efficiency threshold curve based on subjective high estimate of system radius of gyration
It could be argued that span-to-rise ratios greater than the proposed threshold curves are possible, given
that the radius of gyration of the system could sufficiently be reduced. However, this means that the cross
sectional area of the arch would have to be increased faster than the sum of the moments of inertia of the
deck and arch. Effectively, this requires that material be added close to the centroid of the arch, which is
inherently inefficient for carrying bending moments caused by nonuniform live loads and restrained de-
formations. A more precise determination of threshold values for span-to-rise ratios is not possible
without doing detailed calculations of capacity and demand for specific design concepts. Thus these span-
to-rise threshold curves should be considered to be efficiency guidelines for designers, rather than rigid
Equivalent slab thickness hV based on volume of concrete is defined as the total volume of the system V di-
V
hV = Equation 5-12
bdeck ! L
Estimates of concrete volume V can be made by multiplying the average cross sectional areas of arches,
decks, and spandrel columns by their respective element lengths. The expression used to calculate concrete
volume V is given by Equation 5-13, where Adeck, Aarch, and Acol are cross sectional areas of the deck, arch,
187
5. Comparative Study of 58 Concrete Arch Bridges
and columns, L is span length, S is arc length, and hi is the height of each spandrel column. Because the
volume of diaphragms, transverse ribs, and stiffeners are not considered, this estimate is a lower-bound es-
n
V = Adeck ! L + Aarch ! S + Acol ! " hi Equation 5-13
i=1
Figure 5-26 shows equivalent slab thicknesses of bridges in the database versus span length. This quantity
gives some measure of efficiency in terms of concrete consumption. Based on the data, the Gmundertobel
and Tensho Bridges consume more than two times the amount of concrete used in arch bridges of similar
span. The Gmundertobel Bridge was built in 1909 and is likely influenced by the proportions of heavier
masonry arches. The Tensho Bridge uses a single-cell box girder arch rib with thick slabs and webs, result-
ing in a very heavy arch. The assumption that its spandrel walls are solid may also have caused a moderate
Gladesville
1
Sibenik
Fulton Road Pag
Bohlbach
0
Bridges with the lowest equivalent slab thicknesses for various span ranges are labelled in Figure 5-26. Data
points from these bridges were used to form an envelope of minimum concrete consumption, representing
the highest level of material efficiency achieved by existing concrete arch technology. This envelope can be
used to evaluate the relative efficiency of new concepts for concrete arch bridges, as will be done in
Chapter 6. In equation form, the minimum equivalent slab thickness hV envelope is given by the following
188
5. Comparative Study of 58 Concrete Arch Bridges
# L
% + 0.569 m 0 ! L ! 193m
% 1276
hV = $ Equation 5-14
% L2 L
" + 2.21m L > 193m
% 22800 m 61.7
&
Another important aspect to consider is the historical evolution of concrete arches in the database. There
has been a tendency to believe that improvements in the quality and strength of conventional concrete
over the past century have resulted in significant increases in structural efficiency. There has been no exist-
ing evidence, however, to support this claim. Figure 5-27 shows the same data as in Figure 5-26, but adds
the dimension of time. Equivalent slab thickness data points have been labelled with the years of construc-
tion of each bridge. The data shows that there is no apparent relationship between year of construction and
equivalent slab thickness. This suggests that structural efficiency is more a product of the talent and ambi-
tion of a particular designer, rather than the strength of conventional reinforced concrete.
equivalent slab 3
thickness by 2000
volume in m
1983
1965
1993
2 1999
2003
1974
1932 1943
2000 2002 1997
2007 1980
2003 1932 ?
? 1978 1979 1964
1932 ? 1991
1963 1986 1999
1939 ? 2000
1 1967 2011 1966
1937
1952
? 1995 2007 1996 1953
1924 1934 ? 1938 1968
1933 1927 1932
1932
0
0 100 200 300 400 500
span length in m
Figure 5-27. Equivalent slab thickness, span length, and year of construction
189
5. Comparative Study of 58 Concrete Arch Bridges
Dead load due to the self-weight of concrete can be approximated as a uniformly distributed load q, as giv-
en by the following equation, where V is concrete volume (Equation 5-13), # is unit weight of concrete, and
L is span length:
!V
q= Equation 5-15
L
By substituting Equation 5-15 into Equation 5-7, and normalizing the result by deck width bdeck and unit
weight of concrete #, estimates of horizontal reactions H can made using the following equation:
H V "L
= Equation 5-16
! " bdeck 8 f " bdeck
100 L2 + 10 m2
y
182.7
y is normalized
10 horizontal reaction
Figure 5-28 shows estimates of normalized horizontal reactions of bridges in the database versus span
length. An overall exponential trend can be observed from the data. A Levenberg-Marquardt algorithm
was used to fit the data into the form: y=mx2+b. The resulting trend found for normalized horizontal reac-
H L2
in m 2 # + 10 m 2 Equation 5-17
! " bdeck 182.7
190
5. Comparative Study of 58 Concrete Arch Bridges
With this empirical equation, designers can estimate lower bound values for axial forces in the arch using
the deck width and unit weight of concrete specified in their design. These estimates can then be increased
by some factor to account for the additional weight of diaphragms, transverse ribs, and stiffeners.
The effects of live load on fixed arch bridges in the database will be compared in this section. A simple stat-
ical analysis will be used to estimate the eccentricity of the resultant sectional force at the quarter-point of
the arch. Given that the arch is set directly on the pressure line under dead load, the resultant axial forces
along the arch are assumed to be collinear with the centroidal axis of the arch (see Figure 5-29a). The
effects of elastic shortening of the arch are neglected in this simplified analysis. The addition of nonuni-
form live loads causes both axial forces and bending moments in the arch. Live load axial forces are neg-
lected in this simplified analysis, since they are typically small in comparison to axial forces caused by dead
load. Live load bending moments are expressed as displacements of the resultant axial force away from the
centroid of the arch section. Figure 5-29 illustrates the progression of stress profiles and resultant axial
d State 1:
2 GHDGORDGRQO\
S N XQLIRUPFRPSUHVVLYHVWUHVV
UHVXOWDQWIRUFHLQOLQHZLWKFHQWURLG
d State 2:
2 N e GHDGORDGSOXVQRQXQLIRUPOLYHORDG
S WUDSH]RLGDOFRPSUHVVLYHVWUHVV
UHVXOWDQWIRUFHHFFHQWULFWRFHQWURLG
d N State 3:
2 e GHDGORDGSOXVPRUHOLYHORDG
S QRQOLQHDUFRPSUHVVLYHVWUHVV
UHVXOWDQWIRUFHKLJKO\HFFHQWULFWRFHQWURLG
Figure 5-29. Eccentricity of resultant compressive force caused by live load bending moments
The displacement of resultant axial force will be referred to as live load eccentricity elive. Estimates of max-
imum live load eccentricity can be calculated using Equation 5-18, where Mtruck and Mlane are moments at
191
5. Comparative Study of 58 Concrete Arch Bridges
the quarter-point of the arch caused by multilane truck and lane loads, and N is axial force at the quarter-
Quarter-point axial force N of fixed arches caused by self-weight can be estimated using the following
equation, where # is unit weight of concrete, V is volume of concrete (Equation 5-13), L is span, and f is
rise:
2
!V " L%
N (x = L) =
1
+4 Equation 5-19
8 $# f '&
4
Quarter-point bending moments Mtruck of fixed arches caused by multilane truck loading can be estimated
using Equation 5-20, where Qlive is the weight of a single heavy truck, n is number of design traffic lanes, m
is multilane modification factor, kdyn is dynamic amplification factor, and Iarch and Isys are moments of iner-
(
M truck (x = 41 L) = 59.3!10 "3 !Qlive ! L ! n ! m ! 1 + kdyn !) I arch
I sys
Equation 5-20
Quarter-point bending moments Mlane of fixed arches caused by multilane lane loading can be estimated
using Equation 5-21, where qlive is the uniformly distributed weight of a single lane of traffic:
( )
M lane (x = 41 L) = 59.3!10 "3 ! 108 Qlive ! L + 9.1!10 "3 ! qlive ! L2 ! n ! m !
I arch
I sys
Equation 5-21
Load parameters #, Qlive, qlive, kdyn, n and m are given values according to the Canadian Highway Bridge
Design Code CAN/CSA S6-06, as listed in Table 5-7. All other quantities are retrieved from the arch
database.
192
5. Comparative Study of 58 Concrete Arch Bridges
For purposes of comparison, live load eccentricities are normalized by half the arch depth 0.5darch. This
normalizing value is chosen because most bridges in the database have symmetric arch sections, whose
distances between centroid and extreme fibres are equal to 0.5darch. When resultant axial forces have dis-
placed to the extreme fibre at the quarter-point of the arch, normalized live load eccentricities equal one.
bridge ID 51 41 31 21 11 1
0.5
Normalized live load eccentricities of fixed arch bridges in the database versus span length are plotted in
Figure 5-30. From the figure, it can be observed that the effects of live load diminish as span length in-
creases. In three cases, fixed arch bridges with spans less than 100 metres have normalized live load eccent-
ricities greater than one. This suggests that significant nonlinear bending stresses in the arch are expected
at serviceability limit states. Bridges in the database with spans greater than 200 metres have small live
load eccentricities, suggesting that their arch stresses are mostly compressive at serviceability limit.
The diminishing live load effects in arches can be interpreted as a progression from the governing of live
loads in short-span arches to the governing of dead loads in long-span arches. As dead loads become large
193
5. Comparative Study of 58 Concrete Arch Bridges
relative to live loads, arches benefit from an increase in their natural pre-compression under dead load. In
other words, arches with large resultant axial forces N (Figure 5-29) require smaller eccentricities elive than
arches with small resultant axial forces, given the same live load bending moment demand. A similar beha-
viour occurs in suspension bridges, where main suspension cables in tension equilibrate nonuniform live
loads through deformation. Tensile forces in the main cables are caused primarily by dead load. As long as
cables remain in the elastic range, greater tensile forces require smaller deflections to equilibrate live loads.
This behaviour is sometimes called the “gravity stiffness” of suspension bridges because live load perturba-
tions of the system are trivial when added to the state of equilibrium caused by gravity load.
Based on the data in Figure 5-30, designers may be misled to think that all bending moments in long-span
arches are diminished. The simplified analysis performed here considers only the effects of traffic live
loads. The effects of restrained deformation, including temperature, creep, and shrinkage tend to increase
with span length. These effects will also cause the resultant axial force of the arch to displace away from the
centroid of the arch. In many cases, these effects will be more severe than those caused by traffic live loads.
Thus, it is important to understand that the observations made in this section are limited to the effects of
• 46 out of 57 bridges in the database (84%) have span-to-arch-depth ratios in the range of 39 to 77.
• 46 out of 55 bridges in the database (84%) have span-to-system-depth ratios in the range of 24 to 48.
Span-to-system-depth ratios are better measures of the overall proportions of arch bridges than are
• Normalized system moments of inertia hI and span length L were found to be well-correlated. A lin-
194
5. Comparative Study of 58 Concrete Arch Bridges
• System radius of gyration rsys and span length L were found to be well-correlated. A linear regression
• 32 out of 54 bridges in the database (59%) have ratios of arch inertia to system inertia Iarch/Isys greater
than or equal to 0.80, implying that most arches are self-stiffened or have decks that are discontinu-
• 39 out of 49 bridges in the database (80%) have ratios of column inertia to system inertia Icol/Isys less
than 0.1. Because spandrel columns generally have low bending stiffness relative to the arch and deck,
• 43 out of 54 bridges in the database (81%) have system slenderness ratios ( in the range of 38 to 80.
This ratio includes the effects of deck-stiffening through the use of system radius of gyration rsys.
• 55 out of 58 bridges in the database (95%) have span-to-rise ratios in the range of 2.3 to 8.0.
• 48 out of 51 fixed are bridges in the database (94%) have modified slenderness ratios greater than 20,
which means that these arches exhibit greater than 90% arch action under permanent load.
• A threshold curve of maximum span-to-rise ratios for efficient arch behaviour based on the correla-
tion between system radius of gyration rsys and span length L was calculated to be:
L ! L $
< 16.1 #
f " L + 12.4 m &%
• A threshold curve of maximum span-to-rise ratios for efficient arch behaviour based on a more con-
servative, subjective high estimate envelope of system radius of gyration rsys with respect to span
L ! L $
< 16.1 #
f " L + 48.4 m &%
195
5. Comparative Study of 58 Concrete Arch Bridges
• Equivalent slab thicknesses by concrete volume hV were calculated for each bridge in the database.
# L
% + 0.569 m 0 ! L ! 193m
% 1276
hV = $
% L2 L
" + 2.21m L > 193m
% 22800 m 61.7
&
• There is no apparent relationship between year of construction and equivalent slab thickness, sug-
gesting that increases in structural efficiency are not direct outcomes of historical increases of con-
• An overall exponential trend was observed from the data of normalized horizontal reaction versus
span length. The resulting empirical trend, which excludes the weight of diaphragms, transverse ribs,
H L2
in m 2 # + 10 m 2
! " bdeck 182.7
• A simple statical analysis was used to estimate the effects of live load in fixed arch bridges in the data-
base. Comparisons of live load eccentricities, or displacements of the resultant axial force away from
the centroid of the arch, showed that the effects of live load diminish with increasing span length.
196
Chapter 6. New Concepts for Arch Bridges
This chapter describes new preliminary design concepts for arch bridges using ultra high-performance
fibre-reinforced concrete. The stress-strain, creep, and shrinkage models assumed in the design of these
concepts are based on material tests described in Chapter 2. The sectional and member capacities of decks,
spandrel columns, and arches are calculated using the general analysis method described in Chapter 3, and
are compared with the simplified design method. Maximum forces at critical sections in the system are es-
This chapter begins with a description of the objectives, assumptions, and design strategies of a parametric
design study of arch systems using ultra high-performance fibre-reinforced concrete in Section 6.1. The
three parameters considered are: span length, arch rise, and distribution of flexural rigidity. The resulting
design concepts, each responding to different sets of design parameter values, are described in Section 6.2
in terms of their geometries, proportions, and structural demands. In Section 6.2.3, the preliminary design
for one of the design concepts is further developed. In Section 6.3, comparisons of structural efficiency
among design concepts are discussed. Further comparisons are also made with existing concrete arch
bridges. In the last section, general recommendations regarding the potential opportunities offered by
197
6. New Concepts for Arch Bridges
The primary objectives of this parametric design study are: (1) to determine if it is possible to design arches
using ultra high-performance fibre-reinforced concrete that reduce the consumption of concrete by 50% as
compared with current concrete arch technology, (2) to produce preliminary design concepts that respond
to different span length and arch rise requirements, using different distributions of flexural rigidity among
deck and arch, and (3) to quantify and compare the structural efficiency of these preliminary design
concepts.
The three primary design parameters considered are: span length, arch rise, and distribution of flexural ri-
gidity. Parameters are each discretized into a finite set of values. Eight spans between 50 and 400 metres
are considered, each an increment of 50 metres from the previous value. The longest span considered is not
intended as an upper limit of concrete arches, but is instead chosen to be just beyond the range of spans
that have been previously achieved using current concrete arch technology. Four span-to-rise ratios: 4:1,
8:1, 12:1, and 16:1 are considered. The lower values are representative of the range of span-to-rise ratios
commonly used in concrete arch design, while the higher values are greater than the highest ever built. For
each combination of span and span-to-rise ratio, three classes of fixed arch systems are designed: self-sti-
ffened arch, deck-stiffened arch, and partially deck-stiffened arch, each corresponding to different distribu-
tions of flexural rigidity among deck and arch (see Section 4.4.1 on page 99). The combination of 8 span
lengths, 4 span-to-depth ratios, and 3 arch classes results in a design space of 96 potential design concepts.
198
6. New Concepts for Arch Bridges
In order to keep comparisons among design concepts consistent, other geometrical parameters are kept
constant. The specified transverse width of deck for all concepts is 16 metres, which represents a moder-
ately wide bridge that typically holds up to three highway traffic lanes (this width corresponds to four
design lanes of traffic). The specified number of interior spans for all concepts is nine, which ensures that
localized fixed system moments remain small in comparison to global flexible system moments shared by
deck and arch. Increasing the number of interior spans would further reduce fixed system moments, but
would require an increase in the number of spandrel columns, which may result in increases in dead load.
The assumed material properties are kept constant among all design concepts. All sectional analysis calcu-
lations are based on the nonlinear material models presented in Chapter 2. A design compressive strength
of 120 MPa is used, representing a nominal strength that can readily be achieved by many different ultra
high-performance fibre-reinforced concrete mixes, without the use of specialized heat or steam treatment.
A design tensile strength of 14.6 MPa is used, based on the test data presented in Section 2.1.4. This high
tensile strength requires a high steel fibre content similar to the 5.5% fibre content used in the University of
Toronto mix. Figure 6-1 shows the stress-strain curves used for design. Post-peak tensile stresses, which
compressive fc
stress in MPa
120
84
tensile ft
stress in
MPa
14.6
Other important material properties used for design are summarizes in Table 6-2. The amounts of creep
and shrinkage strains imposed on the completed indeterminate arch system are determined by the method
and schedule of construction. Creep strains do not occur until elastic stresses are applied to the arch. Mo-
199
6. New Concepts for Arch Bridges
ments caused by these creep strains do not occur until systems are made indeterminate by the closure of
the arch. In order to simplify the design process, it is assumed that flat, hydraulic jacks are used to intro-
duce elastic stresses into the system while it is determinate. It is also assumed that the arch crown is made
continuous, or closed, a minimum of 28 days after the jacks are activated. Based on the progression of
creep with time shown in Figure 2-11 on page 31, only about 50% of the total long-term creep strain will
cause redundant forces in the system after closure, corresponding to a creep coefficient ,(t) design value of
0.4. The ageing coefficient is a function of the long-term creep coefficient and of the age of concrete at ini-
tial loading (Bazant 1972). Because this quantity is not easily calculated during preliminary design, it is
usually reasonable to assume a typical ageing coefficient +(t) design value of 0.8 (Menn 1990). Increasing
+(t) to its maximum value of 1 causes relatively small changes in the overall effects of creep.
creep coefficient ,(t) 50% of 0.8 Interim Recommendations (adgc 2002), and Graybeal (2006)
shrinkage strain %sh(t) 20% of 550·10-6 Interim Recommendations (adgc 2002), and Habel et al. (2008)
modulus of elasticity E 42000 MPa based on cylinder and prism tests from Chapter 2
density of concrete #/g 2600 kg/m3 based on mix design quantities presented in Chapter 2
unit weight of concrete # 25.5 kN/m3 based on mix design quantities presented in Chapter 2
Using the previous assumption that the closure of the arch is delayed by at least 28 days, the minimum age
of concrete in the arch at closure is 28 days. Based on the progression of shrinkage with age after casting
(see Figure 2-10 on page 30), only about 20% of the total long-term shrinkage strain will cause redundant
forces in the system, corresponding to a shrinkage strain %sh(t) design value of 110·10-6.
The delay of arch closure may not be required in all cases. If the effects of restrained deformations are
small relative to the effects of dead and live load, then the delayed closure requirement may be ignored. If
the effects of restrained deformation are large, they can be reduced by further postponing the closure of
200
6. New Concepts for Arch Bridges
the arch, or by using heat- and pressure-treated precast segments. The selected design values of ,(t), +(t),
and %sh(t) represent a low to moderate amount of creep and shrinkage that can be expected in the system
All other design material properties, including coefficient of thermal expansion, Poisson’s ratio, modulus
of elasticity, density, and unit weight of concrete are based on the material tests presented in Chapter 2 and
on design values given by the the French Interim Recommendations on ultra high-performance fibre-rein-
All concepts are designed according to loads and load factors given in the Canadian Highway Bridge
For each concept, permanent dead loads consisting of the total self-weight of the structure, the weight of
traffic barriers, and the weight of wearing surface are approximated as uniformly distributed loads. The
distributed load corresponding to the self-weight of the structure qdead is approximated using Equation 6-1,
where # is unit weight of concrete, L is span, and V is the total volume of concrete in the system.
V
qdead = ! "
L
Equation 6-1
# % n
(
=! " V A
'& deck " L + A " S + Acol $ hi *
"
)
crown
L i=1
The volume of concrete that participates in the longitudinal behaviour of the system V can be estimated by
multiplying the deck, arch crown, and column cross sectional areas Adeck, Acrown, and Acol by their respective
lengths: span L, arc length S, and column height hi. The additional weight of transverse ribs, diaphragms,
stiffeners, and prestressing steel is accounted for by extra weight factor !V, which is calculated for all con-
cepts individually as the total weight of all structural components divided by the weight of all continuous,
201
6. New Concepts for Arch Bridges
The design superimposed dead load qsdead is calculated based on the geometry of a standard Performance
Level 3 (pl-3) traffic barrier wall and a 40 mm layer of concrete pavement, as shown in Figure 6-2. The as-
sumed traffic barriers are intended to be conservative, since they are among the heaviest that would be
needed in typical Canadian bridge projects. Performance level requirements depend on factors such as
speed and volume of traffic, and distance from edge of paved roadway (see S6-06 §12.4.3.2.1). The barriers
are made from conventionally reinforced concrete, which suggests that with future research and develop-
ment it may be possible to reduce the barrier weight by using ultra high-performance fibre-reinforced
concrete.
The road surface is normally composed of an asphalt wearing surface and waterproofing membrane,
Swiss bridge using ultra high-performance fibre-reinforced concrete, Denarié et al. (2005) used only 40
mm of bituminous concrete pavement without a waterproofing membrane to top the road surface. They
were able to significantly reduce the road surface thickness due to the high resistance of ultra high per-
formance fibre-reinforced concrete against degradation and water penetration. This pavement thickness
was also used by Spasojević (2008) in her design study of thin ultra high-performance fibre-reinforced
concrete deck slabs. The same 40 mm reduced thickness is assumed for all concepts in the parametric
design study. Based on two PL-3 traffic barriers and a 40 mm pavement, the total superimposed dead load
202
6. New Concepts for Arch Bridges
Truck and lane loads used for design are based on S6-06 §3.8.3.1 and §3.8.3.2. In the calculation of flexible
system moments, truck loads are idealized as single concentrated loads of 625 kN applied in each design
lane, instead of the five-axle CL-625-ont truck specified in S6-06 §A3.4. The simplification of five axles to
one axle is conservative, but not overly conservative since most of the span lengths considered in the study
are much larger than the 18 metre model truck length. The lane load is taken as a 9 kN/m uniformly dis-
tributed load combined with a 500 kN concentrated load. In accordance to S6-06 §3.8.4, multi-lane live
loads are obtained by multiplying single-lane live loads by the number of design lanes, the multi-lane
modification factor, and one plus the dynamic amplification factor, as given below:
live load symbol single lane number of multi-lane modi- 1+dynamic ampli- multi-lane
value design lanes fication factor fication factor value
Design values for uniform drops in temperature are based on S6-06 §3.9.4. The minimum effective tem-
perature for Toronto, Ontario is -33°C as given by S6-06 Figure A3.1.2 and S6-06 Table 3.7. Based on an as-
sumed effective construction temperature of 15°C and on the temperature modifications for arch depth d
(given in S6-06 Figure 3.5), the uniform drops in temperature ΔT in °C used for design are given below:
$ "48 0 # d # 0.4 m
&
! T = % 6.25d " 50.5 0.4 m < d # 2.0 m Equation 6-2
& "38 2.0 m < d
&'
Four combinations of load factors are considered: two at serviceability limit states, and two at ultimate lim-
it states. For both limit states, the effects of restrained deformations will first be excluded (young bridge
condition) and then included (old bridge condition). The load factors used for design are shown in Table
6-4. The effects of wind, seismic, and unbalanced multilane loading will not be considered.
203
6. New Concepts for Arch Bridges
Material resistance factors are used to account for the inherent variability of material strength. The materi-
al resistance factor for prestressing steel ,p is taken as 0.95, in accordance with S6-06 §8.4.6. The material
resistance factor for ultra high-performance fibre-reinforced concrete ,c is assumed to be 0.75, which is
the same as the factor for conventional concrete in S6-06. A comprehensive material characterization is
needed to verify the validity of this assumed resistance factor for ultra high performance fibre-reinforced
concrete. Of the cylinders tests described in Table 2-3 on page 22, a maximum sample size of only 3 cylin-
ders were tested under the same age of concrete after casting and curing regime, which is insufficient for
any form of statistical analyses. In her study of composite behaviour of ultra high-performance fibre-rein-
forced concrete and reinforced concrete members, Habel (2004) used the same partial safety factors for
both conventional and ultra high-performance fibre-reinforced concrete. In the French Interim Recom-
mendations (AFGC 2002), §2.2 gives a partial safety factor of 1.3 for ultra high-performance fibre-rein-
forced concrete in tension. The reciprocal of this safety factor is 0.77, which can roughly be taken as an
equivalent resistance factor. Additional strength reduction factors are proposed in Interim Recommenda-
tions to account for localized effects of fibre orientation and methods of placement. These additional
Only design quantities relevant to the longitudinal behaviour of the system are considered in the paramet-
ric study. For selected concepts, proportions of transverse structural members, diaphragms, and stiffeners
are estimated, but not rigorously designed. The following three sections will describe the strategies used in
204
6. New Concepts for Arch Bridges
Concepts 1 through 32 are designed as self-stiffened arch systems. In these systems, arches are designed to
be stiffer than decks so that bending moments caused by nonuniform loads and restrained deformations
are carried primarily by the arch. To investigate the limiting case where arches carry all flexible system
bending moments, deck girders are assumed to be simply supported between spandrel columns and are
Deck girders are designed as double-tee sections prestressed with internal, bonded post-tensioning, as
shown in Figure 6-3. As an alternative, external, unbonded post-tensioned deck systems are also possible,
but are considered outside the scope of this thesis. One advantage of this system is that much thinner webs
205
6. New Concepts for Arch Bridges
The quantities to be designed for the internal, bonded post-tensioned double-tee deck girder are: (1) depth
of deck, (2) thickness of webs, and (3) amount of prestressing steel. In all cases, the deck slab thickness is
specified to be 50 mm. Transverse ribs are used to carry forces transversely to the longitudinal tee-beams
(not shown in Figure 6-3). As a first estimate, the deck depth is chosen such that the interior span-to-deck-
depth ratio is 20:1. The prestressing steel is given a parabolic profile such that the centroid of prestressing is
100 mm from the bottom fibre at midspan. The flexural capacity MR of the beam at midspan is taken con-
servatively as the effective prestressing force after all losses Fp! times the internal lever arm z between the
centroid of tendons and centroid of top slab, as given in Equation 6-3, where ,p is material resistance
factor of prestressing steel, Ap is total area of prestressing steel, fp! is effective prestress after all losses, fpu is
M R = Fp! ! z
(
= "p ! Ap ! fp! ! z ) Equation 6-3
= (" p )
! Ap ! 0.6 fpu ! ( d # 125 mm )
Bending moment demands are calculated based on dead and live loads. Dead load calculations are based
on the weight of the deck girder, traffic barriers, concrete pavement, stiffeners, ribs, and floorbeams. Be-
cause the interior span lengths are smaller or of similar size to the model truck length of 18 m, it is overly
conservative to model truck loads as single concentrated loads. Thus bending moment demands are calcu-
lated using the cl-625 ont truck model (S6-06 §A3.4) with up to five axles.
Arches for these systems are designed as thin-walled single-cell box sections without permanent longitud-
inal prestressing, as shown in Figure 6-3. The quantities to be designed are: (4) depth of arch and (5) thick-
ness of top and bottom slabs. In all cases, web thickness is specified to be 50 mm. As a first estimate, the
arch depth is chosen such that the span-to-arch-depth ratio is 50:1. The thickness of top and bottom slabs
is first taken as 50 mm, and then increased in increments of 50 mm until the axial force-moment N-M*
member capacity interaction diagram of the arch exceeds demands. Member capacity interaction diagrams
for the arches are calculated using the general analysis method in Section 3.2. The model column length is
206
6. New Concepts for Arch Bridges
taken as the effective arc length kS of the bridge. Axial force and bending moment demands are calculated
using the methods described in Section 4.5. If no satisfactory design is found when slab thickness is in-
creased to 500 mm, then the depth of section is increased by decrementing the span-to-arch-depth ratio
from 50:1 to 45:1, and then to 40:1, as shown in Table 6-5. The ratio of 40:1 is chosen because it is near the
lower boundary of span-to-arch-depth ratios observed in existing concrete arch bridges, as shown in Fig-
Table 6-5. Sequence of trial dimensions used in the design of hollow box sections
Concepts 33 through 64 are designed as deck-stiffened arch systems. In these systems, decks are designed
to be stiffer than arches so that bending moments caused by nonuniform loads and restrained deforma-
tions are carried primarily by the deck. In order to enforce this type of behaviour, trial designs are required
to have ratios of deck flexural rigidity to system flexural rigidity that are 0.95 or greater. Bending moment
demands in the arch and deck are calculated using the methods described in Section 4.5.
Deck girders are continuous over spandrel columns and are designed as thin-walled single-cell box sec-
tions with external, unbonded post-tensioning, as shown in Figure 6-4. The quantities to be designed are:
(1) depth of deck, (2) thickness of top and bottom slabs, and (3) amount of prestressing steel. Prestressing
tendons are anchored to diaphragms above each springing line, and are arranged inside the cavity of the
207
6. New Concepts for Arch Bridges
box such that they provide concentric prestress. The width of the box girder is taken as half the width of
the deck (8000 mm). Transverse ribs are used to carry forces transversely to the 50 mm webs of the
girders. As a first estimate, the deck depth is chosen such that the span-to-deck-depth ratio is 50:1. Trial di-
mensions for the deck girder follow the sequence given in Table 6-5. Only the portion of the deck slab
between the girders are thickened with each design iteration. The thickness of the cantilever portions of
the deck slab are kept constant at 50 mm. Member capacity interaction diagrams for the arch ribs are cal-
culated using the general analysis method in Section 3.2. The model column length is taken as the effective
arc length kS of the bridge (as opposed to the effective span length kL). The effects of prestressing steel is
neglected in member capacity calculations and are instead represented as concentric, externally applied
axial loads. Thus, the axial force N in the deck is taken to be the effective prestressing force after all losses
208
6. New Concepts for Arch Bridges
Arches in deck-stiffened concepts are designed as solid or hollow twin ribs. Because they are stiffened by
the deck, the arch ribs can be designed to be very slender, as shown in Figure 6-4. The quantities to be de-
signed are: (4) depth of arch ribs, and (5) total width of arch ribs. If the ribs are made hollow, web and slab
thicknesses are assumed to be 200 mm. As a first estimate, the depth of ribs is chosen such that the span-
to-arch-depth ratio is 200:1. This trial ratio is chosen to be higher than 187, which is the highest ratio
among existing arch bridges in the database (see Figure 5-14 on page 174). The second trial span-to-arch-
depth ratio is 180:1. The total width of ribs is taken as 4 m (or one quarter of the deck width). If 4 m is un-
single wave arch buckling mode multiple wave arch buckling mode
Because these arches are stiffened by the deck, the spandrel columns act as lateral braces for the slender
arches, preventing global buckling. The arches, however, may buckle between the spandrel columns as
shown in Figure 6-5. Both single and multiple wave buckling modes are checked. If the arch section is kept
constant along the entire arch, then the arch legs nearest to the springing lines have the most critical
single-wave buckling modes. This is because these arch legs have the highest unbraced lengths and highest
axial forces among all legs in the arch polygon. Arches, however, are typically designed to be stiffer toward
the springing lines. Thus, single-wave buckling loads are calculated assuming that the most critical arch
legs are those nearest to the quarter-points. The effective length of the arch leg (kL)arch, then, is taken as
effective length factor k times the interior arc length S/n divided by the cosine of the angle of inclination '
at the quarter-point:
209
6. New Concepts for Arch Bridges
S 1
( kL )arch = k ! !
n cos "
2 Equation 6-4
S # f&
= k ! ! 4% ( +1
n $ L'
The effective length factor k for the fixed-fixed arch leg is taken as 0.5 and the number of interior spans n is
taken as 9. With these values, (kL)arch is used as the model column length for calculating member capacity
interaction diagrams for deck-stiffened arch ribs using the general analysis method.
Concepts 65 through 96 are designed as partially deck-stiffened arch systems. In these systems, decks and
arches are designed to have about the same flexural rigidity, and thus share flexible system moments.
Bending moment demands are calculated for the arch and deck using the methods described in Section
4.5.
210
6. New Concepts for Arch Bridges
As shown in Figure 6-6, the deck girders for this system are designed as thin-walled box girders in the
same way the deck girders are designed in Section 6.1.3. Accordingly, the quantities to be designed are: (1)
depth of deck, (2) thickness of top and bottom slab, and (3) amount of prestressing steel. The first trial deck
depth is chosen such that the span-to-deck-depth ratio is 100:1. The arch ribs are also designed as thin-
walled box girders and are constrained to be identical to the deck girders, except that they are not perman-
ently prestressed. Thicknesses of top and bottom slabs of both deck and arch are first taken as 50 mm.
Amount of prestressing steel in the deck is increased until bending moment demands are within the axial
force-moment N-M* member capacity interaction diagram, as calculated by the general analysis method
in Section 3.2.
To account for the mutual stiffening of the arch and deck, the model column length is taken as 0.707 times
the effective arc length kS, which has the effect of doubling the flexural rigidity of the deck or arch section,
whichever is being analyzed. If member capacities of the deck or arch are unsatisfactory, even with the aid
mm. If the design is still unsatisfactory, then the depths of deck and arch are increased such that their
span-to-depth ratios are 90:1, then 80:1, and then 70:1. Web thicknesses are kept constant at 50 mm.
211
6. New Concepts for Arch Bridges
The longitudinal proportioning of the deck and arch sections among all concepts are summarized below:
self-stiffened arch deck double tee (1) depth of deck L/n:d = 20:1
(2) thickness of webs 500 mm, 1000 mm
(3) amount of prestressing steel number of 0.6” strands
arch hollow box (4) depth of arch L:d = 50:1, 45:1, 40:1
(5) thickness of top/bottom slabs 50 mm to 500 mm
deck-stiffened arch deck hollow box (1) depth of deck L:d = 50:1, 45:1, 40:1
(2) thickness of top/bottom slabs 50 mm to 500 mm
(3) amount of prestressing steel number of 0.6” strands
partially deck-stiffened arch deck/arch hollow box (1) depth of deck and arch L:d = 100:1, 90:1, 80:1, 70:1
(2) thickness of top/bottom slabs 50 mm to 500 mm
deck hollow box (3) amount of prestressing steel number of 0.6” strands
As webs and slabs are made thinner than previously built in conventionally reinforced concrete structures,
local stability problems become increasingly important. The design of hollow rectangular components is
dealt with in S6-06 §8.8.5.8. In this clause, wall slenderness ratios, which are defined as the wall length b
divided by wall thickness t, are used to determine whether a given wall will buckle locally before it can
reach its maximum compressive stress. According to this standard, wall slenderness ratios greater than 15
are subject to decreases in compressive strength by up to 25%. This clause originates from an analytical and
experimental study done by Taylor et al. (1995) on hollow thin-walled piers made from conventionally re-
inforced concrete. Their analysis treats compression flanges of hollow boxes as a thin, rectangular plates
subjected to compressive uniformly distributed loads along two edges, as shown by the diagrams on the
212
6. New Concepts for Arch Bridges
left in Figure 6-7. The support conditions along the edges parallel to the applied load are assumed to be re-
strained elastically against rotation (fixed). The bottom-left diagram shows the buckled shape of the plate.
According to Timoshenko’s Theory of Elastic Stability (1936), critical buckling stresses fE of thin plates can
be calculated using Equation 6-5, where kE is plate buckling coefficient, E is elastic modulus, * is Poisson’s
! 2E $ t '
2
/2E £ t ¥
2 thin-walled ribbed ultra high-
fE kE reinforced performance fibre-
12 1 < p 2 ¤ b ¦ concrete box reinforced concrete box
longitudinal
stiffener
transverse rib
t b t
b
pinned
buckled shape
pinned pinned
213
6. New Concepts for Arch Bridges
The value of the plate buckling coefficient kE depends on the aspect ratio, or width-to-length ratio, of the
plate and the support conditions along the edges of the plate. The minimum plate buckling coefficient for
plates that are simply supported along the edges perpendicular to the load and rotationally restrained
In their study, Taylor et al. (1995) use the tangent modulus Etan=d//d% in place of the initial modulus of
elasticity to account for the material nonlinearities of concrete at high stresses. This relatively simple tan-
gent modulus approach was shown by Swarts and Rosebraugh (1974) to adequately model the buckling be-
haviour of thin concrete plates. By imposing that the critical buckling stress fE of the plate is greater than or
equal to the compressive strength of concrete fc! and rearranging Equation 6-5, an expression for maxim-
b #2 E
! kE " " tan Equation 6-6
t (
12 1 $ % 2
) fc&
The top-right diagram of Figure 6-7 shows an ultra high-performance fibre-reinforced concrete box with
transverse ribs and longitudinal stiffeners (only a few are shown for clarity). Rotational restraint at the
web-flange corners of the hollow ribbed box is highest near the transverse ribs and lowest half way
between the transverse ribs. To simplify the problem, it is assumed that no rotational restraint is provided
along these corners. Thus, the compression flange of the box can be assumed to be simply supported along
all edges, which corresponds to a minimum plate buckling coefficient kE of 4. Substituting material proper-
ties for ultra high-performance fibre-reinforced concrete: Etan=21100 MPa, fc!=120 MPa, and *=0.20, results
in a maximum wall slenderness ratio of 24.5. This means that as wall thicknesses are reduced down to 50
mm, the maximum width of plate between webs needed to attain the full compressive strength of concrete
is 1230 mm, which is too small for the deck and arch members. Thus, longitudinal stiffeners are needed to
increase the allowable width of these thin plates. The longitudinal stiffeners shown in Figure 6-7 are as-
sumed to provide additional lines of pinned supports, effectively dividing the compression flange into mul-
214
6. New Concepts for Arch Bridges
tiple plates. Maximum plate widths for other plate thicknesses are tabulated in Table 6-7, and are used in
Table 6-7. Maximum plate widths for thin ultra high performance fibre-reinforced concrete plates
As discussed in Section 4.4.1 and illustrated in Figure 4-15 on page 106, spandrel columns can be assumed
with zero end bending moments must be designed with minimum end eccentricities of load of 15 mm plus
3% of the structural depth of the column. The appropriate analytical model for an eccentrically-loaded
column is shown in Figure 6-8. The general analysis method for calculating ultimate loads Q* of columns,
which is described in Section 3.2.5, was used to develop load capacity design curves for various hollow
cross-sections using ultra high-performance fibre-reinforced concrete. The input parameters for the calcu-
lation procedure were: column cross-section, material stress-strain curve, column length, and eccentricity
of load. The output result was ultimate compressive load Q* of the column. For each cross-section, the cal-
culation was repeated for a wide range of column lengths from zero up to the length corresponding to a
slenderness ratio ( of 200. Design curves for five hollow column sections are shown in Figure 6-8. Because
wall slenderness ratios of these sections are all below 24.5, local buckling problems are avoided.
215
6. New Concepts for Arch Bridges
factored ultimate
load capacity !c·Q* in MN 50
!c = 0.75 E
40
Q
30
D
e
L 20
C
B
10 data points calculated
A by program QULT using
Q general analysis method
structural 0
model 0 20 40 60 80 100 column length L in m
eccentricity 100
position
of load Q centroid
of section
50 100
60
50 100
45 45
30 30
Generalized column load capacity curves can be obtained by normalizing ultimate load capacities Q* by
cross sectional area of the column A and by the compressive strength of concrete fc!. Column length L is
normalized by column radius of gyration r to obtain slenderness ratios. As shown in Figure 6-9, normaliz-
ing all five design curves from Figure 6-8 results in very similar generalized design curves. The results
show that ultra high-performance fibre-reinforced concrete columns with slenderness ratios of 60 have
load capacities that are about half the failure load of the section in pure axial compression. These general-
ized column design curve can be used to estimate the load capacities of any hollow, rectangular ultra high-
performance fibre reinforced concrete column with pinned supports, as long as wall slenderness ratios are
216
6. New Concepts for Arch Bridges
Q*
normalized load capacity 1
A·fc!
0.8
0.6
0.4
0.2
0
0 50 100 150 200 slenderness ratio "
This section reports the results of the parametric design study. Design quantities are tabulated in Section
6.2.1. Member capacity interaction diagrams of concept arches are also shown along with axial force-mo-
ment demands. Preliminary drawings of Concept 80, which is a partially deck-stiffened arch system with
main span of 400 m and arch rise of 50 m, is described in Section 6.2.3. The precast segmental construc-
Design concepts with span-to-rise ratios of 16:1 were found to be inherently inefficient because of the un-
avoidable large bending moment demands that arise due to the beam action caused by the shallowness of
the arch. No practical arch design solutions were found. Because of this, these concepts are absent from the
parametric design solutions that follow. The efficiency and practicality of very shallow arches will be furth-
The proportions of 72 arch design concepts using ultra high-performance fibre-reinforced concrete are giv-
en in Table 6-8. The first group of columns in the table specify the design parameters that each concept
was designed for. The second group of columns in the table specify the proportions of the deck girders.
Girder widths are not specified for self-stiffened arch concepts because they have double-tee sections
217
6. New Concepts for Arch Bridges
rather than box sections. VSL Multistrand Systems are used as the longitudinal deck prestressing steel.
Steel quantities are specified by the number of required Type 0.6” grade 270 ksi/1860 MPa strands. The
third group of columns in the table specify the proportions of the arch. Web and slab thicknesses are not
specified for concepts with solid arch ribs. For the deck-stiffened concepts, arch width is reported as the
combined width of the two arch ribs. The last group of columns in the table specify the proportions of
spandrel columns. Width values for columns refer to the combined transverse width of the two columns
(i.e. 2.0 m means two columns that are each 1.0 m wide). Cross-sectional areas and moments of inertia are
Table 6-9 tabulates geometric ratios and structural demands of the 72 design concepts. The first group of
columns tabulates the total volume of continuous, longitudinal concrete V according to Equation 5-13 on
page 188, and volume distributions among deck, arch, and spandrel columns. The second group of
columns tabulates quantities related to system stiffness, including: span-to-system-depth ratio (system
depth is the sum of deck and arch depths), system moment of inertia (sum of deck and arch moments of
inertia), system radius of gyration (system moment of inertia divided by cross sectional area of arch), and
column-to-system-moment of inertia ratio. The third group of columns tabulates slenderness and modi-
fied slenderness ratios (Equation 5-6 on page 179 and Equation 5-8 on page 184), equivalent slab thickness
by volume (Equation 5-12 on page 187), and normalized system moment of inertia (Equation 5-2 on page
176).
The fourth group of columns in Table 6-9 tabulates quantities related to the design dead load, or self-
weight of the structure. Extra weight factors are ratios between the total weight of all structural elements
(including ribs, diaphragms, stiffeners, and tendons) and the weight of continuous, longitudinal concrete.
A sample calculation of extra weight factor is shown in Calculation D3 on page 319. Based on these extra
weight factors, uniformly distributed dead loads and horizontal reactions due to dead load are calculated
for serviceability limit states (unfactored loads). Arch action values are calculated based on modified slen-
derness (f as given by the second factor in Equation 4-43 on page 115: 1/(1+45·(f-2).
218
6. New Concepts for Arch Bridges
8 400 100.0 4 2.22 180 500 50 3.0 1570 248 8.00 50 8 50 50 1.6 16700 1.5 3.0 60 100 1.12 368
9 50 6.3 8 0.28 180 1000 50 1.3 8 114 1.25 40 8 50 250 4.1 1020 0.5 2.0 4 50 0.28 11
10 100 12.5 8 0.56 180 1000 50 1.8 57 114 2.00 50 8 50 100 1.8 1490 0.5 2.0 8 50 0.28 11
11 150 18.8 8 0.83 180 500 50 1.6 108 114 3.00 50 8 50 50 1.1 1940 0.5 2.0 11 50 0.28 11
12 200 25.0 8 1.11 180 500 50 1.9 240 132 4.00 50 8 50 50 1.2 3610 0.5 2.0 15 100 0.52 17
13 250 31.3 8 1.39 180 500 50 2.1 443 132 5.00 50 8 50 100 2.1 10500 1.0 2.0 19 50 0.38 57
14 300 37.5 8 1.67 180 500 50 2.4 728 186 6.00 50 8 50 100 2.2 15600 1.0 2.0 23 50 0.38 57
15 350 43.8 8 1.94 180 500 50 2.7 1090 186 7.00 50 8 50 100 2.3 21700 1.0 2.0 26 50 0.38 57
16 400 50.0 8 2.22 180 500 50 3.0 1570 186 8.00 50 8 50 150 3.2 40800 1.0 2.0 30 50 0.38 57
17 50 4.2 12 0.28 180 1000 50 1.3 8 114 1.25 40 8 50 250 4.1 1020 0.5 2.0 3 50 0.28 11
18 100 8.3 12 0.56 180 1000 50 1.8 57 114 2.00 50 8 50 100 1.8 1490 0.5 2.0 5 50 0.28 11
19 150 12.5 12 0.83 180 500 50 1.6 108 114 3.00 50 8 50 100 1.9 3550 0.5 2.0 8 50 0.28 11
20 200 16.7 12 1.11 180 500 50 1.9 240 132 4.00 50 8 50 150 2.8 9320 0.5 2.0 10 50 0.28 11
21 250 20.8 12 1.39 180 500 50 2.1 443 132 5.00 50 8 50 200 3.7 19300 0.5 2.0 13 100 0.52 17
22 300 25.0 12 1.67 180 500 50 2.4 728 186 6.00 50 8 50 300 5.3 40300 1.0 2.0 15 50 0.38 57
23 350 29.2 12 1.94 180 500 50 2.7 1090 186 7.00 50 8 50 400 7.0 71800 1.0 2.0 18 50 0.38 57
24 400 33.3 12 2.22 180 500 50 3.0 1570 186 8.00 50 8 50 500 8.7 116000 1.0 2.0 20 50 0.38 57
33 50 12.5 4 1.00 50 8 50 50 1.3 249 160 0.25 200 4 1.0 5 0.5 2.0 8 50 0.28 11
34 100 25.0 4 2.00 50 8 50 50 1.4 1090 160 0.50 200 4 2.0 42 0.5 2.0 15 50 0.28 11
35 150 37.5 4 3.00 50 8 50 100 2.3 4270 160 0.83 180 4 200 200 1.9 171 1.0 2.0 23 50 0.38 57
36 200 50.0 4 4.00 50 8 50 100 2.4 7840 160 1.11 180 4 200 200 2.2 361 1.0 2.0 30 50 0.38 57
37 250 62.5 4 5.00 50 8 50 100 2.5 12600 160 1.39 180 4 200 200 2.4 635 1.0 2.0 38 50 0.38 57
38 300 75.0 4 6.00 50 8 50 150 3.4 25200 160 1.67 180 4 200 200 2.6 1000 1.5 3.0 45 100 1.12 368
39 350 87.5 4 7.00 50 8 50 150 3.5 34900 160 1.94 180 4 200 200 2.8 1470 1.5 3.0 53 100 1.12 368
deck-stiffened arch concepts
40 400 100.0 4 8.00 50 8 50 150 3.6 46400 240 2.22 180 4 200 200 3.1 2040 1.5 3.0 60 100 1.12 368
41 50 6.3 8 1.00 50 8 50 50 1.3 249 200 0.25 200 4 1.0 5 0.5 2.0 4 50 0.28 11
42 100 12.5 8 2.00 50 8 50 50 1.4 1090 200 0.50 200 4 2.0 42 0.5 2.0 8 50 0.28 11
43 150 18.8 8 3.00 50 8 50 100 2.3 4270 200 0.83 180 4 200 200 1.9 171 0.5 2.0 11 50 0.28 11
44 200 25.0 8 4.00 50 8 50 100 2.4 7840 200 1.11 180 4 200 200 2.2 361 0.5 2.0 15 100 0.52 17
45 250 31.3 8 5.00 50 8 50 100 2.5 12600 240 1.39 180 4 200 200 2.4 635 1.0 2.0 19 50 0.38 57
46 300 37.5 8 6.00 50 8 50 150 3.4 25200 320 1.67 180 4 200 200 2.6 1000 1.0 2.0 23 50 0.38 57
47 350 43.8 8 7.00 50 8 50 150 3.5 34900 320 1.94 180 4 200 200 2.8 1470 1.0 2.0 26 50 0.38 57
48 400 50.0 8 8.00 50 8 50 150 3.6 46400 400 2.22 180 4 200 200 3.1 2040 1.0 2.0 30 100 0.72 98
49 50 4.2 12 1.00 50 8 50 50 1.3 249 200 0.25 200 6 1.5 8 0.5 2.0 3 50 0.28 11
50 100 8.3 12 2.00 50 8 50 100 2.2 1800 320 0.50 200 6 3.0 63 0.5 2.0 5 50 0.28 11
51 150 12.5 12 3.00 50 8 50 150 3.1 5800 320 0.75 200 6 4.5 211 0.5 2.0 8 50 0.28 11
52 200 16.7 12 4.44 45 8 50 200 4.0 16700 440 1.00 200 6 6.0 500 0.5 2.0 10 100 0.52 17
53 250 20.8 12 6.25 40 8 50 200 4.2 34400 550 1.25 200 6 7.5 977 0.5 2.0 13 50 0.28 11
54 300 25.0 12 7.50 40 8 50 200 4.3 50700 770 1.50 200 6 9.0 1690 1.0 2.0 15 50 0.38 57
55 350 29.2 12 8.75 40 8 50 200 4.4 70200 770 1.75 200 6 10.5 2680 1.0 2.0 18 50 0.38 57
56 400 33.3 12 10.00 40 8 50 250 5.4 111000 990 2.00 200 6 12.0 4000 1.0 2.0 20 50 0.38 57
65 50 12.5 4 0.50 100 8 50 100 2.0 82 200 0.50 100 8 50 100 1.6 66 0.5 2.0 8 50 0.28 11
66 100 25.0 4 1.00 100 8 50 100 2.1 403 200 1.00 100 8 50 100 1.7 330 0.5 2.0 15 50 0.28 11
67 150 37.5 4 1.50 100 8 50 100 2.1 975 200 1.50 100 8 50 100 1.7 804 1.0 2.0 23 50 0.38 57
68 200 50.0 4 2.00 100 8 50 100 2.2 1800 200 2.00 100 8 50 100 1.8 1490 1.0 2.0 30 50 0.38 57
69 250 62.5 4 2.50 100 8 50 100 2.2 2900 200 2.50 100 8 50 100 1.8 2410 1.0 2.0 38 50 0.38 57
partially deck-stiffened arch concepts
70 300 75.0 4 3.00 100 8 50 100 2.3 4270 200 3.00 100 8 50 100 1.9 3550 1.0 2.0 45 100 0.72 98
71 350 87.5 4 3.50 100 8 50 100 2.3 5910 240 3.50 100 8 50 100 1.9 4920 1.5 3.0 53 100 1.12 368
72 400 100.0 4 4.00 100 8 50 100 2.4 7840 320 4.00 100 8 50 100 2.0 6540 1.5 3.0 60 100 1.12 368
73 50 6.3 8 0.50 100 8 50 100 2.0 82 200 0.50 100 8 50 100 1.6 66 0.5 2.0 4 50 0.28 11
74 100 12.5 8 1.00 100 8 50 100 2.1 403 200 1.00 100 8 50 100 1.7 330 0.5 2.0 8 50 0.28 11
75 150 18.8 8 1.50 100 8 50 100 2.1 975 200 1.50 100 8 50 100 1.7 804 0.5 2.0 11 50 0.28 11
76 200 25.0 8 2.00 100 8 50 100 2.2 1800 200 2.00 100 8 50 100 1.8 1490 0.5 2.0 15 100 0.52 17
77 250 31.3 8 2.50 100 8 50 150 3.0 3930 240 2.50 100 8 50 150 2.6 3410 1.0 2.0 19 50 0.38 57
78 300 37.5 8 3.00 100 8 50 200 3.9 7210 320 3.00 100 8 50 200 3.5 6430 1.0 2.0 23 50 0.38 57
79 350 43.8 8 3.89 90 8 50 150 3.2 10100 440 3.89 90 8 50 150 2.8 8780 1.0 2.0 26 50 0.38 57
80 400 50.0 8 4.44 90 8 50 200 4.0 16700 440 4.44 90 8 50 200 3.6 15000 1.0 2.0 30 100 0.72 98
81 50 4.2 12 0.63 80 8 50 100 2.0 140 200 0.63 80 8 50 100 1.6 112 0.5 2.0 3 50 0.28 11
82 100 8.3 12 1.25 80 8 50 150 2.9 862 200 1.25 80 8 50 150 2.5 738 0.5 2.0 5 50 0.28 11
83 150 12.5 12 1.88 80 8 50 200 3.7 2580 240 1.88 80 8 50 200 3.3 2280 0.5 2.0 8 50 0.28 11
84 200 16.7 12 2.50 80 8 50 300 5.4 6460 440 2.50 80 8 50 300 5.0 5900 0.5 2.0 10 50 0.28 11
85 250 20.8 12 3.57 70 8 50 500 8.7 20400 440 3.57 70 8 50 500 8.3 19200 1.0 2.0 13 100 0.72 98
86 300 25.0 12 4.29 70 8 50 500 8.7 30800 550 4.29 70 8 50 500 8.3 29100 1.0 2.0 15 50 0.38 57
87 350 29.2 12 5.00 70 8 50 500 8.8 43500 660 5.00 70 8 50 500 8.4 41200 1.0 2.0 18 50 0.38 57
88 400 33.3 12 5.71 70 8 50 500 8.9 66900 770 5.71 70 8 50 500 8.5 55400 1.0 2.0 20 50 0.38 57
bullet indicates that decks are double-tee bullet indicates that arch ribs are solid
sections
219
6. New Concepts for Arch Bridges
8 2380 50 31 19 50 16700 3.24 2.2 50 62 0.37 2.32 1.46 0.22 99 44 4 1.5 161.0 0 72 28
9 282 22 75 3 40 1020 0.50 1.1 36 25 0.35 0.91 1.42 0.20 93 10 3 1.5 21.3 54 43 3
10 381 48 49 4 50 1490 0.91 0.8 40 27 0.24 1.04 1.59 0.15 94 15 3 2.0 31.2 31 59 10
11 429 55 40 5 50 1940 1.33 0.6 41 28 0.18 1.13 1.81 0.13 95 19 3 2.9 48.4 24 57 19
12 673 55 37 8 50 3610 1.74 0.5 42 29 0.21 1.39 1.70 0.15 95 28 2 3.5 75.3 26 47 27
13 1120 48 48 4 50 10500 2.25 0.5 41 28 0.28 1.99 1.53 0.17 95 41 2 2.7 115.0 34 41 25
14 1460 50 47 4 50 15600 2.68 0.4 41 28 0.30 2.27 1.51 0.19 95 53 2 3.1 165.0 34 37 29
15 1840 51 45 4 50 21700 3.09 0.3 41 28 0.33 2.53 1.48 0.20 95 66 2 3.5 226.0 32 34 34
16 2590 46 51 3 50 40800 3.59 0.1 41 28 0.40 3.13 1.40 0.23 95 87 2 3.1 308.0 37 30 33
17 275 23 75 2 40 1020 0.50 1.1 36 17 0.34 0.91 1.43 0.20 86 13 3 3.1 28.0 60 37 3
18 372 49 49 3 50 1490 0.91 0.8 39 18 0.23 1.04 1.60 0.15 88 20 3 4.4 42.3 37 48 15
19 539 44 53 3 50 3550 1.37 0.3 39 18 0.22 1.39 1.64 0.15 88 30 3 5.1 74.2 38 42 20
20 955 39 59 2 50 9320 1.83 0.1 39 18 0.30 1.91 1.49 0.18 88 48 2 5.0 128.0 45 32 23
21 1510 35 62 3 50 19300 2.30 0.1 39 18 0.38 2.44 1.39 0.21 88 71 2 5.1 207.0 47 27 26
22 2390 30 68 2 50 40300 2.75 0.1 39 18 0.50 3.11 1.31 0.27 88 105 2 4.8 331.0 52 22 26
23 3490 27 72 1 50 71800 3.20 0.1 39 18 0.62 3.78 1.25 0.32 88 147 2 4.8 495.0 54 18 28
24 4780 25 74 1 50 116000 3.65 0.0 39 18 0.75 4.43 1.21 0.37 88 195 2 4.9 708.0 55 16 29
33 136 47 42 10 40 254 0.50 4.4 40 50 0.17 0.58 1.75 0.13 98 3 5 0.7 11 0 98 2
34 397 35 58 7 40 1130 0.75 1.0 53 67 0.25 0.95 1.52 0.18 99 9 5 0.7 23 0 96 4
35 735 47 46 8 39 4440 1.51 1.3 40 50 0.31 1.49 1.44 0.23 98 17 3 0.6 39 28 66 6
36 1050 45 47 7 39 8200 1.94 0.7 41 51 0.33 1.83 1.42 0.24 98 24 3 0.7 54 28 63 9
37 1400 44 49 7 39 13200 2.35 0.4 43 53 0.35 2.15 1.40 0.26 98 32 3 0.7 69 28 61 11
38 2250 45 40 15 39 26200 3.17 1.4 38 47 0.47 2.70 1.30 0.34 98 51 2 0.7 100 32 54 14
39 2750 44 41 15 39 36400 3.58 1.0 39 49 0.49 3.01 1.30 0.36 98 62 2 0.7 126 31 53 16
GHFNVWLIIHQHGDUFKFRQFHSWV
40 3290 43 43 14 39 48400 3.98 0.8 40 50 0.51 3.31 1.29 0.38 98 74 2 0.8 158 30 51 19
41 124 52 42 6 40 254 0.50 4.4 36 25 0.15 0.58 1.82 0.11 93 5 3 2.2 14 28 66 6
42 361 38 58 4 40 1130 0.75 1.0 48 33 0.23 0.95 1.57 0.17 96 16 3 2.4 31 30 58 12
43 667 51 46 3 39 4440 1.51 0.3 36 25 0.28 1.49 1.48 0.20 93 29 3 2.2 59 39 47 14
44 980 49 46 5 39 8200 1.94 0.2 37 26 0.31 1.83 1.44 0.23 94 42 3 2.5 86 39 43 18
45 1290 48 48 4 39 13200 2.35 0.4 39 27 0.32 2.15 1.43 0.24 94 56 2 2.7 122 37 39 24
46 1880 54 43 3 39 26200 3.17 0.2 34 24 0.39 2.70 1.37 0.29 93 80 2 2.7 186 41 34 25
47 2310 53 45 3 39 36400 3.58 0.2 36 24 0.41 3.01 1.36 0.30 93 99 2 2.9 243 38 32 30
48 2850 50 45 5 39 48400 3.98 0.2 37 25 0.45 3.31 1.34 0.33 93 122 2 3.3 321 36 29 35
49 146 44 52 3 40 257 0.41 4.4 43 20 0.18 0.58 1.70 0.13 90 9 3 4.7 18 34 56 10
50 533 41 57 2 40 1860 0.79 0.6 45 21 0.33 1.12 1.39 0.24 91 33 3 4.5 49 45 40 15
51 1160 40 59 1 40 6010 1.16 0.2 46 22 0.48 1.65 1.28 0.36 91 73 3 4.8 97 48 30 22
52 2060 39 59 2 37 17200 1.69 0.1 42 20 0.64 2.35 1.23 0.47 90 127 2 4.7 185 55 21 24
53 2980 35 64 1 33 35400 2.17 0.0 41 19 0.75 2.98 1.21 0.55 89 183 2 4.5 312 59 17 24
54 4080 32 67 1 33 52400 2.41 0.1 44 21 0.85 3.40 1.20 0.62 91 254 2 5.3 430 51 16 33
55 5340 29 70 1 33 72900 2.63 0.1 47 22 0.95 3.80 1.18 0.70 92 337 2 5.9 603 46 14 40
56 7080 30 69 1 33 115000 3.10 0.0 46 22 1.11 4.42 1.16 0.81 91 444 2 6.1 868 46 12 42
65 209 48 45 7 50 147 0.30 7.6 67 83 0.26 0.48 1.55 0.19 99 5 5 0.9 12 0 95 5
66 429 48 45 7 50 732 0.66 1.5 61 76 0.27 0.82 1.55 0.20 99 10 5 0.9 24 0 93 7
67 675 47 44 9 50 1780 1.01 3.2 59 74 0.28 1.10 1.53 0.21 99 15 5 0.9 37 0 90 10
68 922 47 44 8 50 3290 1.36 1.7 59 74 0.29 1.35 1.53 0.21 99 21 4 1.0 51 0 83 17
69 1180 47 45 8 50 5310 1.70 1.1 59 73 0.30 1.59 1.53 0.22 99 27 4 1.2 74 0 77 23
SDUWLDOO\GHFNVWLIIHQHGDUFKFRQFHSWV
70 1550 44 42 14 50 7820 2.04 1.3 59 74 0.32 1.80 1.49 0.24 99 35 4 1.5 105 0 71 29
71 1990 41 39 20 50 10800 2.37 3.4 59 74 0.36 2.01 1.46 0.26 99 45 4 1.9 147 0 63 37
72 2320 41 39 20 50 14400 2.70 2.6 60 74 0.36 2.21 1.47 0.27 99 53 4 2.3 195 0 58 42
73 193 52 44 4 50 147 0.30 7.6 61 42 0.24 0.48 1.58 0.18 97 9 3 3.0 12 17 70 13
74 397 52 44 4 50 732 0.66 1.5 55 38 0.25 0.82 1.57 0.18 97 18 3 3.1 28 19 61 20
75 611 52 44 4 50 1780 1.01 0.6 54 37 0.25 1.10 1.57 0.19 97 27 3 3.6 49 19 54 27
76 859 51 43 6 50 3290 1.36 0.5 54 37 0.27 1.35 1.55 0.20 97 38 2 4.5 78 18 42 40
77 1480 51 46 3 50 7340 1.67 0.8 54 37 0.37 1.77 1.41 0.27 97 66 2 4.4 125 20 35 45
78 2300 50 47 3 50 13600 1.98 0.4 55 38 0.48 2.17 1.33 0.35 97 102 2 4.7 196 21 28 51
79 2180 51 46 3 45 18900 2.62 0.3 49 33 0.39 2.42 1.44 0.29 96 96 2 4.9 252 22 29 49
80 3250 49 46 5 45 31700 2.97 0.3 49 34 0.51 2.88 1.34 0.37 96 144 2 5.2 370 22 24 54
81 190 54 44 2 40 252 0.39 4.5 45 21 0.24 0.57 1.64 0.17 91 12 3 5.0 17 31 56 13
82 553 52 46 2 40 1600 0.80 0.7 45 21 0.35 1.06 1.44 0.25 91 35 3 4.9 46 38 43 19
83 1090 52 47 1 40 4860 1.20 0.2 44 21 0.45 1.54 1.35 0.33 91 68 3 5.2 88 40 33 27
84 2110 51 48 1 40 12400 1.58 0.1 45 21 0.66 2.10 1.25 0.48 91 132 2 5.5 167 43 22 35
85 4330 50 49 1 35 39600 2.19 0.2 41 19 1.08 3.10 1.16 0.80 89 265 2 5.0 353 53 15 32
86 5200 50 49 1 35 59900 2.68 0.1 40 19 1.08 3.55 1.17 0.80 89 317 2 5.6 509 51 14 35
87 6120 50 49 1 35 84700 3.18 0.1 39 18 1.09 3.99 1.17 0.80 88 372 2 6.2 707 48 13 39
88 7050 50 49 1 35 122000 3.79 0.0 38 18 1.10 4.51 1.15 0.81 87 424 1 6.8 1060 46 9 45
220
6. New Concepts for Arch Bridges
The last group of columns tabulate maximum deflections and maximum flexible system bending moments
at ultimate limit states, as caused by the five load combinations shown in Figure 4-21 on page 127. These
maximum moments are then separated into the following components: (1) moments caused by creep,
shrinkage, and temperature, labelled as “CST system moment,” (2) moments caused by live load, and (3) ad-
ditional moments caused by second-order effects, labelled as “H-e system moment.” Comparisons among
all design concepts will be discussed in greater detail in Section 6.3, using the data presented in Tables 6-8
and 6-9.
The adequacy of given trial designs were assessed using N-M* interaction diagrams, as calculated by the
general method described in Section 3.2. Diagrams were calculated using factored material resistances.
Five load combinations at ultimate limit states were used for calculating maximum, second-order sectional
forces using the simplified analysis methods described in Section 4.5. Trial designs were deemed satisfact-
Interaction diagrams for the arch members among all design concepts are presented in Figures 6-10, 6-11,
and 6-12. All axial force N results have been divided by the cross-sectional area of the arch A to obtain axial
stress, and then normalized by the factored compressive strength of concrete ,c fc!:
N 1
n= ! Equation 6-7
A "c fc#
All ultimate moment M* results have been divided by section modulus of the arch 2I÷d to obtain effective
flexural stress at the extreme fibre, and then normalized by half the factored compressive strength of
concrete:
M* d 1
m* = ! ! Equation 6-8
I 2 "c fc#
Two sets of sectional force demands are plotted on each interaction diagram. Solid markers represent sec-
tional forces at ultimate limit state while hollow markers represent sectional forces at serviceability limit
states. Because the interaction diagrams were calculated based on factored material resistance, the section-
221
6. New Concepts for Arch Bridges
al forces at serviceability limit states were also factored down by the material resistance of concrete. This
normally associated with serviceability limit states. Instead of comparing with member capacity interac-
tion diagrams (solid line), unfactored sectional forces at serviceability limit states are compared with con-
tours of cracking, represented by dashed lines on the diagrams. The contour of cracking is defined as the
set of sectional forces at which initial cracking occurs at the flexural tension face of the section. Decks and
arches among all concepts were designed to remain uncracked under service loads so that deformations of
the structure remain small and in the linear range. This also serves to maintain a margin of safety between
Two other criteria were used to determine wether given trial solutions were satisfactory or not. The first
criterion is related to admissible axial forces imposed on the section. Axial force demands and axial forces
introduced by prestressing were kept below half the maximum load under uniform compression. This en-
sured that additional deflections and moments due to second order effects were not excessively large. It
also ensures some minimum level of ductility, since flexural failures would be initiated by the progressive
pull-out of steel fibres, rather than the crushing of concrete. The second criterion is related to admissible
deflections at ultimate limit states. Trial designs were dismissed if deflections at ultimate limit states were
greater than 5% of the initial rise of the arch. Like the first criterion, this ensured that second-order effects
were not excessively large. More importantly, this deflection criterion ensured the applicability of the sim-
plified calculation methods used to calculate sectional force demands. When crown deflections are larger
than 5%, the reduction of the rise of the arch imposes additional moments on the system, which are not ac-
222
sample
interaction diagram
superposition of
normalized 1
arch interaction diagrams
axial force:
from all self-stiffened concepts
N 1 1
n= · n
A !c·fc!
223
concepts L 50 m 100 m 150 m 200 m 250 m 300 m 350 m 400 m
L:f
1-8 4:1
9-16 8:1
M·d 1 0
normalized moment: m = ·
2I !c·fc! 0 0.5 1 1.5
m
224
concepts L 50 m 100 m 150 m 200 m 250 m 300 m 350 m 400 m
L:f
33-40 4:1
41-48 8:1
M·d 1 0
normalized moment: m = ·
2I !c·fc! 0 0.5 1 1.5
m
225
concepts L 50 m 100 m 150 m 200 m 250 m 300 m 350 m 400 m
L:f
65-72 4:1
73-80 8:1
Figure 6-13. Sections of selected design concepts and existing arch bridges
226
6. New Concepts for Arch Bridges
Figure 6-13 shows typical cross-sections of twelve of the proposed concepts. The last column of diagrams
shows typical cross-sections of existing concrete arch bridges with spans similar to those of the concepts in
the same row of diagrams. The proposed concepts shown are proportioned for span-to-rise ratios of 8:1.
Longitudinal stiffeners are spaced so that local buckling does not occur before the maximum compressive
stress of concrete is reached. Transverse and vertical ribs are proportioned to improve the ability of the
sections to transmit transverse and torsional loads. The cross sectional areas of continuous longitudinal
concrete have been lightly shaded to distinguish them from the transverse and vertical ribs. The shading
allows for visual comparisons of the quantity of concrete used among concepts and existing concrete arch
bridges. Diaphragms, which would normally correspond to sections taken through the centres of the span-
Manual design calculations for Concept 80 are included in Appendix D. These calculations demonstrate
the use of the simplified design methods for calculating sectional force demands in arch systems and cal-
culating member capacity using the reduced n-m interaction diagram. Calculations for two ultimate limit
state load combinations are considered. In both cases, the resulting second-order sectional forces were
shown to be resisted safely by the given arch and deck sections (see Calculation D12 on page 328). These
calculations can be used as a guideline for designers, and as a model example for implementing calcula-
tions on a spreadsheet.
In this parametric study, it has been assumed that axial forces and bending moments govern the design of
arch systems. Because arches carry loads primarily in axial compression, this assumption is typically valid.
For this reason, the effects of shearing forces in arch systems have not been rigorously considered. In order
to validate the use of very thin 50 mm webs without conventional stirrups, a rudimentary shear design
check is shown in Calculations D16 to D18, starting on page 332. Ultimate limit state load combination 1
from S6-06 is used, in which the lane load is applied over half the span along with a uniformly distributed
dead load. Axial force, bending moment, and shear force diagrams are calculated along the arch span. Us-
227
6. New Concepts for Arch Bridges
ing a linear-elastic approach, axial and shear stresses are calculated using properties of the arch section in
Calculation D16. Using Mohr’s circle of plane stress, all principal stresses in the arch member were found
to be less than the factored cracking stress of the material ,c·ft of 5.25 MPa. The most critical (i.e. the most
positive) principal stresses occur near the springing lines, where sectional forces are highest.
It may be the case that there are other design concepts and other load cases that have principal stresses that
exceed the cracking limit at service loads or the tensile strength limit at ultimate loads. The concepts that
would be most susceptible are: (1) concepts that are shallow, because of the beam action under permanent
loads, and (2) concepts that have low arch prestress (i.e. short spans), since precompression helps to reduce
tensile stresses.
For these cases, shear capacity can be increased by increasing web thickness or by prestressing the member
vertically. Increasing web thickness is undesirable since it detracts from the goal of minimizing concrete
consumption and minimizing weight. If webs are thickened, they can be thickened locally in areas of high
shear, rather than uniformly over the whole span. Using vertical prestressing would be the better option,
since it does not add much additional weight to the structure. The vertical compressive stresses introduced
by prestressing would cause the principal stresses in a member to become more compressive (i.e more neg-
ative in a Mohr’s circle diagram). Prestressing steel could be installed at each transverse rib. This would not
only help prevent shear cracks under service loads, but would also create a well-defined truss-like flow of
forces should the structure be cracked at ultimate limit states. Vertical prestressing steel would carry
tensile forces upward, while the webs would carry diagonal compressive forces along the web panels, simil-
ar to how forces are carried in conventionally reinforced concrete beams. This concludes a brief discussion
on the shear resistance and shear design of the proposed arch systems.
6.2.3 Concept 80
In order to present a more detailed design of one of the proposed concepts, preliminary drawings for
Concept 80 have been developed. Plan, elevation, and sections views of the bridge concept are shown in
Figure 6-14. The arch has a span of 400 m, a span-to-rise-ratio of 8:1, and is partially stiffened by the deck.
228
6. New Concepts for Arch Bridges
The ground profile is fictitious and has been adapted to suit the geometry of the arch. The geometry of the
As shown by Section A-A, the deck girder has a constant depth of 4.44 m and is composed of identical pre-
cast segments, each 3 m long. The arch is also made from 3 m long segments that are nearly identical to the
deck girder segments, except for the cantilevers. The arch rib has a constant depth of 4.44 m, except for the
outermost spans near the springing lines, where the arch is made deeper and thicker to accommodate
According to S6-06 §8.8.5.3, tensile stress across segmental joints without continuous reinforcing bars is
limited to zero. Most of the of the design concepts developed in this study satisfy this requirement. For the
15 out of 72 concepts that do not, alternative joint types or an increase in longitudinal post-tensioning will
be required.
As shown by Sections B-B and C-C, diaphragms are provided above and below each set of columns. These
diaphragm segments can be constructed using pre-cast or cast-in-place methods. The former allows for
high quality of concrete and higher geometry control, while the latter allows for greater flexibility in ad-
justing for geometrical imperfections and misalignments during construction. The proposed spandrel
columns are hollow and thin, which provides for a high degree of visual transparency. The short columns
near the arch crown will probably cause problems for seismic loading. A viable alternative would be to fuse
the arch and deck at the crown. This provides a direct load path for longitudinal seismic-induced inertial
forces to be transferred from the massive deck to the arch, and then from the arch to the foundations.
Along the transverse and vertical ribs of the deck and arch segments, holes for the main longitudinal post-
tensioning are provided. For the deck, twenty VSL Type 6-22 Multistrand tendons are positioned along the
perimeter of the segment cavity to provide concentric prestress. VSL Type E Anchors for these tendons are
shown in Section E-E in Figure 6-15. Thicker diaphragms are provided above the springing lines to accom-
modate for the disturbed regions near the anchors. Figure 6-15 also shows longitudinal sections of half the
229
6. New Concepts for Arch Bridges
230
6. New Concepts for Arch Bridges
231
6. New Concepts for Arch Bridges
The trade-off of using minimum materials in highly efficient systems is that the resulting geometrical
shapes are not easily built. Incorporating transverse ribs and longitudinal stiffeners into the segments
poses potential problems for the precasting process. The problem with the proposed ribs and stiffeners is
that they intersect along three dimensions. Every waffle-like recess requires five abutting surfaces to be
formed. The successful release of these forms is another source of complexity. These challenges require a
new and innovative precasting method. One possible segmental precasting method is presented below.
The proposed precasting method involves two separate casting procedures, one for the end ribs of the seg-
ment, and one for the body of the segment between the two ribs. The first procedure, consisting of four
stages, is illustrated in Figure 6-17. The first stage consists of preparing all the components of the form. A
custom-built steel bulkhead is used to form the face of the principal end rib. Block-outs for shear keys are
built into the bulkhead. Steel angles are fastened to the bulkhead to form the perimeter of the rib shape.
Block-outs for prestressing tendon holes are also fastened to the bulkhead. Two lengths of corrugated steel
duct, which will be used later for the transverse post-tensioning of the deck, are installed into the form.
Deformed steel wire is arranged among the areas of the ribs that will abut against the inside body of the
segment. The quantity of deformed steel wire is chosen such that its effective yield force is equal to or
Once all these components have been setup, the principal end rib is cast and cured (Stage 2). Once the
principal end rib has gained sufficient strength, it is carefully lifted, turned over, and set upon spacer
blocks (Stage 3). This exposes the underside of the rib, which has the recessed impression of the steel bulk-
head. A second set of steel angles is used to line the perimeter of the principal end rib. A second set of
post-tensioning ducts and deformed wires are also installed. The secondary abutting end rib is then cast
directly against the surface of the principal end rib (Stage 4). This technique, called match-casting, pro-
duces optimal bond quality between the contact surfaces of the two ribs. These perfectly match-catch sur-
faces will later be rejoined on site during the erection of the bridge. Once the match-cast end rib has
gained sufficient strength, the ribs are transported to the area dedicated to the second casting procedure.
232
6. New Concepts for Arch Bridges
233
6. New Concepts for Arch Bridges
The procedure for casting the interior of the segment consists of eight stages, and is illustrated in Figure 6-
16. This figure is intended as a schematic representation of the procedure, rather than a detailed drawing.
The outside base form for the 3 metre segment is represented by the gray, U-shaped polygon. This form
along with the inside core form would consist of an array of steel sections, plates and stiffeners. It would be
designed so that several components could be retracted away from the completed segment.
The first stage of the procedure is the assembly of the base form, cantilever form, and core form. In the
second stage, precast end ribs are positioned on either side of the core form with their shear keys facing
outward. The two precast ends are not a match-cast pair, since they will be on opposing sides of a segment.
The third stage consists of suspending the core form and clamping it to the end ribs using the holes inten-
ded for prestressing tendons in the final structure. Once the forms are sufficiently sealed, the interior body
of the segment is cast (Stage 4). No vibration is required, since the concrete is self-consolidating. The cold
construction joints between the end ribs and body are effectively bonded through the embedded deformed
wires. Once the segment has gained sufficient strength, the core form is unclamped and released, and re-
moved in parts, starting with the top and bottom slab forms, then the web forms, and last the corner forms
(Stages 5 through 7). The geometry of the segment should be adjusted to ease the removal of forms. The
rectangular stiffeners and right-angled corners shown in the figure should be modified as tapered stiffeners
and sloped corners to reduce the friction between the form and cast segment. The final stage consists of
lifting the completed segment out of the base form and then moving it to a storage area.
The proposed partially deck-stiffened systems, including Concept 80, have arch segments that are nearly
identical to their deck segments. This provides unique cost-savings in terms of the cost of the custom bulk-
head and forms. The repetitive precasting procedure, which is the same for the arch and deck, should res-
ult in increased economies of scale and improved efficiency in production, relative to the other arch sys-
234
6. New Concepts for Arch Bridges
235
6. New Concepts for Arch Bridges
Renderings of one assembled precast segment and of multiple match-cast segments for Concept 80 are
shown below. An average height human figure is shown to establish a sense of scale.
236
6. New Concepts for Arch Bridges
This section compares the design concepts with respect to the three design parameters: span length, arch
rise, and distribution of flexural rigidity among arch and deck. In order to illustrate the similarities and di-
fferences among concepts, various design quantities will be organized in the form of a comparison matrix
as shown in Table 6-10. The effects of span length can be evaluated by comparing values in the same row.
The effects of arch rise can be evaluated by comparing values in each column. The effects of deck-stiffening
Span-to-deck-depth ratios of self-stiffened concepts are shown in Table 6-11. All deck girders were sat-
isfactorily designed with span-to-depth ratios of 180:1. Given the constant number of interior spans of 9 as-
sumed for all concepts, this ratio corresponds to interior-span-to-depth ratios of 20:1, which is not uncom-
mon for simply supported girder-type structural systems. Six vsl type 0.6” multistrand tendons (three per
web) with 19 to 31 strands each were found to be satisfactory in resisting bending moments at ultimate lim-
it states. Instead of increasing the amount of strands per tendon, eight tendons were used in the 400 m
span concepts to keep the tendon anchorages small. All tendons are internally bonded within the webs of
the double-tee sections. Webs were made 500 mm or 1000 mm wide to allow space for tendons, an-
237
6. New Concepts for Arch Bridges
Table 6-11 also shows span-to-arch-depth ratios and slab thicknesses of the self-stiffened arch concepts.
Most arches were satisfactorily designed with span-to-depth L:d ratios of 50:1, except for the 50 m concepts
with span-to-rise ratios of 8:1 and 12:1. These concepts with short spans were found to have relatively low
dead load precompression in the arch. Thus deeper arch sections were needed to compensate for the low
flexural resistance of the arch. Results from these concepts also showed a sensitivity to cracking at service-
For the arch section, top and bottom slab thicknesses of 50 mm were satisfactory for most concepts with
span-to-rise ratios L:f of 4:1, while 100 mm thicknesses were satisfactory for most concepts with L:f=8:1.
Slab thicknesses up to 500 mm were required for concepts with L:f=12:1 to keep ultimate deflections below
5% of initial arch rise. These shallow arch concepts were not made deeper than L:d=50:1 to decrease deflec-
tions because their modified slenderness ratios were already below the efficiency threshold of 20.
Given that slab thickness requirements are low among concepts with span-to-rise ratios of 4:1 and 8:1, it is
likely possible that arches in these concepts could be made more slender by increasing their span-to-depth
ratios (i.e. to 60:1). This reduction in lever arm would require increases in compressive force, and thus in-
creases in slab thickness. Deflections may also increase due to the increased flexibility of the system, which
238
6. New Concepts for Arch Bridges
Span-to-deck-depth ratios and slab thicknesses of deck-stiffened arch concepts are shown in Table 6-12.
Most deck girders were satisfactorily designed with span-to-depth ratios of 50:1. Among these designs, 50
mm top and bottom slab thicknesses were used for concepts with 50-100 m spans, 100 mm slab thick-
nesses for 150-250 m spans, and 150 mm slab thicknesses for 300-400 m spans. Concepts with span-to-rise
ratios of 12:1 required greater slab thicknesses to compensate for larger moment demands. Of these, con-
cepts with spans of ≥200 m were made deeper to reduce deflections. For some of these concepts, it was not
possible to keep deflections below 5% of initial arch rise, which lessens the validity of the simplified calcu-
lation methods used for design. Performing more complex finite element analyses on these deflection-
sensitive concepts may result in more severe bending moment demands and deflections than those calcu-
Table 6-12 also shows span-to-arch-depth ratios of deck-stiffened arch concepts. Due to the stiffening
provided by the deck girder, slender solid arch ribs with span-to-depth ratios of 200:1 and total rib widths
of 4 m were found to be satisfactory for concepts with low span-to-rise ratios and spans of ≤100 m. For
concepts with spans >100 m, solid arch ribs were found to consume more concrete than necessary. Thus,
arch ribs were made hollow to lighten the dead load and made deeper to maintain the flexural rigidity re-
239
6. New Concepts for Arch Bridges
quired to resist buckling between spandrel columns. Because most of the flexible system moments get dis-
tributed to the deck, all arches in these deck-stiffened systems remain uncracked at ultimate limit states (as
shown in Figure 6-11 on page 224). For concepts with span-to-rise ratios of 12:1, the total width of arch ribs
had to be increased by 50% and were kept solid. Increasing the cross sectional area and axial rigidity of the
arch allowed the modified slenderness of the system to stay above the efficiency threshold of 20. If axial ri-
gidity were not increased, the arches would be too shallow and be subjected to large bending moments
caused by beam action. These arch shallowness requirements increase the material consumption of the
Span-to-deck-depth ratios and slab thicknesses of deck-stiffened arch concepts are shown in Table 6-13.
Most concepts with span-to-rise ratios of 4:1 and 8:1 were satisfactorily designed with decks and arches
with span-to-depth ratios of 100:1 and with slab thicknesses of 100 to 200 mm. For concepts with span-to-
rise ratios of 12:1, deeper sections and thicker slabs were needed to keep deflections below 5% of initial arch
rise.
240
6. New Concepts for Arch Bridges
Normalized member capacity n-m* interaction diagrams for all concept arches are collectively shown in
Figure 6-19. These diagrams were calculated using the general method, and then normalized according to
Equations 6-7 and 6-8. Normalized contours of cracking are also shown as dashed lines.
The shapes of the n-m interaction diagrams are very similar for all concepts. The diagrams for the deck-sti-
ffened arches tend to have higher normalized moments than arches of the other two systems. This is a res-
ult of the shape of section used: in the deck-stiffened systems, solid or hollow sections with relatively walls
were used, while in the other two systems, hollow sections with thin walls were used.
In Figure 6-19, reduced n-m design interaction diagrams, calculated using the simplified method in Section
3.5, are superimposed on to the member capacity n-m* interaction diagrams calculated using the general
method. The figure shows that in all cases, the proposed design interaction diagram is both conservative
and accurate relative to the interaction diagrams based on more refined calculations. It should be noted
that the value of normalized axial force at Euler buckling nE is given the default value of 0.8 in the figure.
This value is changed according to the slenderness of the column, effectively reducing the dark gray area
above n=0.5. Calibrating this value would result in reduced interaction diagrams that follow the n-m* dia-
The figure also shows that very few design solutions had normalized axial force demands that were greater
than 0.5. This is a result of the design process, in which arches with very high axial-force demands were
subjected to large secondary moments. This is reflected in Vianello’s deflection magnifier equation, in
which additional secondary deflections and moments increase exponentially as the load approaches the
buckling resistance of the member. Thus, very few arches with normalized axial force demands greater
than 0.5 were capable of carrying the corresponding large second-order moments.
241
6. New Concepts for Arch Bridges
superposition of
arch interaction diagrams
0.5
nt —ncr from all self-stiffened concepts
n 1
0
ncr 0.5 1 1.5 0.5
nt
m
mn=0
QE
nE = = normalized axial force at Euler buckling, but not > 0.8
A·fc!
f
ncr = cr = cracking stress divided by compressive strength
fc!
f!
nt! = t = tensile strength divided by compressive strength 0
fc!
0 0.5 1 1.5
mn=0 = normalized ultimate moment when N=0
m
superposition of superposition of
arch interaction diagrams arch interaction diagrams from all
from all deck-stiffened concepts partially deck-stiffened concepts
n 1 n 1
0.5 0.5
0 0
0 0.5 1 1.5 0 0.5 1 1.5
m m
242
6. New Concepts for Arch Bridges
The structural depths of spandrel columns among all concepts are shown in Table 6-14. In most cases,
500-1000 mm column depths with wall thicknesses of 50-100 mm were capable of carrying the axial forces
and moments caused by construction tolerances and misalignments of load. Deeper 1.5 m column sections
were needed for concepts with taller spandrel columns, including those with spans >300 m and span-to-
rise ratios of 4:1. Transverse loading such as wind loading and temporary construction loads may increase
Equivalent slab thicknesses based on volume of continuous, longitudinal concrete are shown in Table 6-15.
Minimum values among concepts with the same span and rise are highlighted. Among concepts with
spans of 100 m or less, deck-stiffened arches consume the least amount of concrete. For example, Concept
41 achieves an effective slab thickness of 0.15 m, which is 37% less than the thickness of the other two sys-
tems for a span of 50 m. This implies that deck-stiffened arches are the most materially efficient over short
spans. Given the accuracy of the analysis method and resolution of design iterations, all systems can be
considered to be more or less equally efficient for spans of 100 m. For spans of ≥150 m, the data shows that
self-stiffened arch systems consume the least amount of concrete in almost all cases.
243
6. New Concepts for Arch Bridges
Table 6-15. Equivalent slab thicknesses based on volume of longitudinal concrete among all concepts
Reducing the concrete consumption and overall weight of arch bridges is the primary objective of this
thesis. Thus, equivalent slab thicknesses of the design concepts are shown again in Table 6-15, this time ex-
pressed as percentages of the envelope of minimum equivalent slab thicknesses observed among arches in
the database from Chapter 5. Results show that reductions of concrete volume by 50% or more can be read-
ily be achieved using any one of the three arch systems considered, as long as span-to-rise ratios are kept in
the range of 4:1-8:1. In this range, arches are not likely to be shallow, or sensitive to axial deformations of
the arch. For shallow arch concepts with span-to-rise ratios of 12:1, most deck-stiffened and partially deck-
stiffened concepts were unable to achieve a reduction of concrete volume by 50%. Thus the cost of building
these concepts are not likely to be much different than arches built using conventional concrete techno-
logy. If designers do consider using shallow arches, equivalent slab thickness results show that self-sti-
ffened arches are likely to be the most economical choice among the three arch systems considered in this
study.
A graphical comparison of the consumption of concrete among arches made from ultra high-performance
fibre-reinforced concrete and arches made from conventional concrete is shown in Figure 6-20. This figure
244
6. New Concepts for Arch Bridges
clearly shows the significant material efficiencies offered by the proposed ultra high-performance fibre-re-
inforced concrete arch systems. In most cases, the proposed concepts represent reductions in concrete
volume much greater than 50% when compared to the heavier concrete arch bridges in the database.
0
0 100 200 300 400 500 span length in m
Figure 6-20. Comparison of equivalent slab thicknesses of new arch concepts and existing arches in database
Material efficiency can also be discussed in terms of quantity of main longitudinal prestressing steel in the
deck girder among concepts. Table 6-16 shows that in most cases, self-stiffened arch systems use the least
amount of prestressing. This is not unexpected since the decks in self-stiffened systems carry the least mo-
Table 6-16. Quantity of longitudinal prestressing strands in deck girder among all concepts
Table 6-17 shows the relative concrete volume of arch and deck among all concepts. One might expect that
the stiffening element in the system would naturally require more material than the flexible element it sti-
245
6. New Concepts for Arch Bridges
ffens. Results show, however, that the arch and deck typically have about the same concrete volume regard-
less of which element stiffens the system. This can be possibly be explained by the additional limitations
imposed on the flexible elements in the systems. For example, flexible deck elements in self-stiffened arch
systems are required to transmit loads in bending over interior spans, and thus require some nominal
amount of concrete to achieve this. In deck-stiffened arch systems, flexible arch elements are required to
Table 6-17. Relative volume of arch and deck among all concepts
(volume of arch) / (volume of arch + deck) among all concepts, in %
concept ID class of arch system L = 50 100 150 200 250 300 350 400 m
9 to 16 self-stiffened 8:1 77 51 42 40 50 48 47 53
41 to 48 deck-stiffened 8:1 45 60 47 49 50 45 46 47
73 to 80 partially deck-stiffened 8:1 46 46 46 46 47 48 48 48
17 to 24 self-stiffened 12:1 77 50 55 60 64 69 73 75
49 to 56 deck-stiffened 12:1 54 58 60 60 65 68 71 70
81 to 88 partially deck-stiffened 12:1 45 47 48 49 49 49 49 49 %
Table 6-18 shows the relative concrete volume of spandrel columns to the total concrete volume of the sys-
tem. Results show that the weight of columns are typically small in comparison to that of the deck and
arch. This, however, is not be true for concepts with very tall piers, as illustrated by the concepts with spans
of ≥300 m and span-to-rise ratios of 4:1. These volume ratios were calculated based on the sectional geo-
metry required for the tallest spandrel columns within each design concept. Thus these values represent an
upper limit, since the proportions of the shorter spandrel columns can be reduced.
246
6. New Concepts for Arch Bridges
Span-to-system depth ratios among all concepts are shown in Table 6-19. Most of the proposed concepts
have ratios of 39:1-50:1. Figure 6-21 compares span-to-system-depth ratios of the concepts with those from
the concrete arch database in Chapter 5. The figure shows that the proposed concepts have ratios that are
more or less in the same range as those in the database. Comparing the highest values in the figure, con-
cepts with span-to-system-depth ratios of about 50:1 are on par with the thinnest arch systems ever built.
9 to 16 self-stiffened 8:1 40 50 50 50 50 50 50 50
41 to 48 deck-stiffened 8:1 40 40 39 39 39 39 39 39
73 to 80 partially deck-stiffened 8:1 50 50 50 50 50 50 45 45
0:1
0 100 200 300 400 500 span length in m
Normalized system moments of inertia among all concepts (as calculated by Equation 5-2 on page 176) are
shown in Table 6-20. Results show that partially deck-stiffened systems tend to have the lowest system mo-
ments of inertia among concepts with span-to-rise ratios of 4:1 and 8:1. This is not unexpected since the
moment of inertia of a single full-depth member (i.e. the arch in self-stiffened systems) is greater than two
half-depth members (i.e. the deck and arch in partially deck-stiffened systems). Among concepts with
span-to-rise ratios of 12:1, system moments of inertia are generally lowest in self-stiffened systems. This is
because members in the other systems had to be deepened to reduce crown deflections.
247
6. New Concepts for Arch Bridges
System radii of gyration among all concepts are shown in Table 6-21. The last rows of the table give the
range of values that can be expected of these new concepts for arch bridges. Low values of system radii of
gyration are preferred since they have high axial rigidity, and are the least likely to have problems associ-
1 to 8 self-stiffened L:f = 4:1 0.42 0.91 1.33 1.74 2.13 2.51 2.88 3.24
33 to 40 deck-stiffened 4:1 0.50 0.75 1.51 1.94 2.35 3.17 3.58 3.98
65 to 72 partially deck-stiffened 4:1 0.30 0.66 1.01 1.36 1.70 2.04 2.37 2.70
9 to 16 self-stiffened 8:1 0.50 0.91 1.33 1.74 2.25 2.68 3.09 3.59
41 to 48 deck-stiffened 8:1 0.50 0.75 1.51 1.94 2.35 3.17 3.58 3.98
73 to 80 partially deck-stiffened 8:1 0.30 0.66 1.01 1.36 1.67 1.98 2.62 2.97
17 to 24 self-stiffened 12:1 0.50 0.91 1.37 1.83 2.30 2.75 3.20 3.65
49 to 56 deck-stiffened 12:1 0.41 0.79 1.16 1.69 2.17 2.41 2.63 3.10
81 to 88 partially deck-stiffened 12:1 0.39 0.80 1.20 1.58 2.19 2.68 3.18 3.79 m
min 0.30 0.66 1.01 1.36 1.67 1.98 2.37 2.70 PLQLPXPLQFROXPQ
max 0.50 0.91 1.51 1.94 2.35 3.17 3.58 3.98 PD[LPXPLQFROXPQ
248
6. New Concepts for Arch Bridges
= 3 12 × system inertia
deck width
existing concrete arches 4
arch design concepts
0
0 100 200 300 400 500 span length in m
= system inertia
arch area
3
existing concrete arches
arch design concepts
2
0
0 100 200 300 400 500 span length in m
Figure 6-22. Comparison of normalized system moment of inertia and system radius of gyration
Figure 6-22 displays data from Tables 6-20 and 6-21 along with data from the database of existing concrete
arch bridges. The first plot shows that the system moments of inertia of the proposed concepts are gen-
erally lower than those of bridges in the database. This difference can be partially attributed to the differ-
ences in elastic moduli between conventional concrete and ultra high-performance fibre-reinforced con-
crete. Given that the elastic modulus of the latter is about 1.5 times that of the former, the system flexural
rigidity of the proposed concepts are more or less the same as those of bridges in the database. The second
plot shows that the system radii of gyration of the proposed concepts are about the same compared to
those of bridges in the database for spans <200 m, and are generally higher for spans ≥200 m. This means
that the concepts with spans ≥200 m are more susceptible to shallow arch behaviour than bridges in the
249
6. New Concepts for Arch Bridges
System slenderness ratios among all concepts are shown in Table 6-22. The highest system slendernesses
are achieved by partially deck-stiffened concepts with span-to-rise ratios of 4:1. These concepts have slen-
derness ratios of about 60, which is actually not that high when compared to the column tests described in
Chapter 3 (the shortest of the four columns had a slenderness ratio ( of about 60). Figure 3-6 on page 52
shows ultimate member capacity N-M* interaction diagrams for various slenderness ratios. Based on this
figure, slenderness ratios of 60 do not reduce the capacity of the member by much relative its sectional ca-
pacity. This, however, does not mean that additional second-order deflections can be neglected.
Figure 6-23 displays system slenderness ratios of the proposed concepts and concrete arch bridges in the
database. This figure shows that many existing concrete arches have much higher system slenderness ratios
than the proposed concepts. Slender behaviour should not be confused with the slender appearance of
bridge components. Slender behaviour, which is quantified by (, does not change proportionally with visu-
40
20
0
0 100 200 300 400 500 span length in m
250
6. New Concepts for Arch Bridges
Modified slenderness ratios among all concepts are shown in Table 6-23. As discussed in Chapter 4, modi-
fied slenderness ratios give measures of arch shallowness. As long as modified slenderness ratios of sys-
tems are kept above the threshold value of 20, then the arch will have greater than 90% arch action. The
results in Table 6-23 show that all concepts are well above this threshold, except those with span-to-rise ra-
tios of 12:1.
Modified slenderness ratios of concepts are compared with those of existing concrete arch bridges in Fig-
ure 6-24. The figure shows that the proposed concepts are generally more susceptible to arch shallowness
problems than bridges in the database. This is not totally unexpected, since most bridges in the database
For concepts with span-to-rise ratios of 12:1, design calculations were governed by arch shallowness and
large crown deflections. The only way to reduce the effects of arch shallowness is to increase (f. This is
done by adding material such that it increases the cross sectional area of the arch faster than it increases
system moment of inertia. Adding material, however, also increases dead load which then also increases
251
6. New Concepts for Arch Bridges
sectional forces in the system. Using the simplified methods of analysis, design iterations for concepts with
span-to-rise ratios of 12:1 eventually led to seemingly satisfactory solutions. The simplified methods of ana-
lysis, however, assume that changes in the rise of the arch are small relative to the initial rise. When deflec-
tions are large, these methods may grossly underestimate the maximum sectional forces imposed on the
system. Thus, the proposed design concepts with span-to-rise ratios of 12:1 should be used with caution.
More refined structural analyses are needed to further validate these designs.
Table 6-24. Maximum deflections at ultimate limit states among all concepts
maximum deflections at ULS ÷ arch rise among all concepts, in %
concept ID class of arch system L = 50 100 150 200 250 300 350 400 m
1 to 8 self-stiffened L:f = 4:1 0.6 0.6 0.8 0.9 1.0 1.2 1.4 1.5
33 to 40 deck-stiffened 4:1 0.7 0.7 0.6 0.7 0.7 0.7 0.7 0.8
65 to 72 partially deck-stiffened 4:1 0.9 0.9 0.9 1.0 1.2 1.5 1.9 2.3
9 to 16 self-stiffened 8:1 1.5 2.0 2.9 3.5 2.7 3.1 3.5 3.1
41 to 48 deck-stiffened 8:1 2.2 2.4 2.2 2.5 2.7 2.7 2.9 3.3
73 to 80 partially deck-stiffened 8:1 3.0 3.1 3.6 4.5 4.4 4.7 4.9 5.2
17 to 24 self-stiffened 12:1 3.1 4.4 5.1 5.0 5.1 4.8 4.8 4.9
49 to 56 deck-stiffened 12:1 4.7 4.5 4.8 4.7 4.5 5.3 5.9 6.1 •values greater than or
81 to 88 partially deck-stiffened 12:1 5.0 4.9 5.2 5.5 5.0 5.6 6.2 6.8 % equal to 5.0% are in red type
Table 6-24 shows the greater of deflections at the quarter-point and deflections at crown of the proposed
concepts at ultimate limit states. These deflections are the sum of second-order deflections caused by live
load, creep, shrinkage, and temperature. Elastic deflections caused by dead load are excluded because they
can be accounted for with camber during construction. Most of the concepts with span-to-rise ratios of 4:1
and 8:1 have relatively small ultimate deflections of about 3% of the initial rise or less. Concepts with span-
to-rise ratios of 12:1 have relatively high maximum deflections of about 4% to 6%. Any further modification
Estimating dead loads during preliminary design can be difficult because the quantity of materials are not
accurately known. Underestimating dead load results in unconservative estimates of structural demand.
Overestimating dead load results in unconservative estimates of precompression in the arch. Thus it is be-
neficial to have a good estimate of dead load during the early stages of design. The total weight of all struc-
tural components, including the weight of the deck, columns, and arch, transverse ribs, stiffeners, dia-
252
6. New Concepts for Arch Bridges
phragms, and prestressing steel were calculated for each concept. Extra weight factors, which are
calculated as the total weight of all structural components divided by the weight of continuous, longitudin-
al concrete, are shown for each concept in Table 6-25. Extra weight factors of the concepts range from 1.15
to 1.82. Figure 6-25 plots extra weight factors versus span length. Results show that using values between
1.4 and 1.6 results in good first estimates of dead load. The lower end of the range is more appropriate for
longer spans, while the higher end of the range is more appropriate for shorter spans.
1 to 8 self-stiffened L:f = 4:1 1.52 1.56 1.74 1.68 1.62 1.54 1.48 1.46
33 to 40 deck-stiffened 4:1 1.75 1.52 1.44 1.42 1.40 1.30 1.30 1.29
65 to 72 partially deck-stiffened 4:1 1.55 1.55 1.53 1.53 1.53 1.49 1.46 1.47
9 to 16 self-stiffened 8:1 1.42 1.59 1.81 1.70 1.53 1.51 1.48 1.40
41 to 48 deck-stiffened 8:1 1.82 1.57 1.48 1.44 1.43 1.37 1.36 1.34
73 to 80 partially deck-stiffened 8:1 1.58 1.57 1.57 1.55 1.41 1.33 1.44 1.34
17 to 24 self-stiffened 12:1 1.43 1.60 1.64 1.49 1.39 1.31 1.25 1.21
49 to 56 deck-stiffened 12:1 1.70 1.39 1.28 1.23 1.21 1.20 1.18 1.16
81 to 88 partially deck-stiffened 12:1 1.64 1.44 1.35 1.25 1.16 1.17 1.17 1.15
min 1.42 1.39 1.28 1.23 1.16 1.17 1.17 1.15 PLQLPXPLQFROXPQ
max 1.82 1.60 1.81 1.70 1.62 1.54 1.48 1.47 PD[LPXPLQFROXPQ
1.5
1
0 100 200 300 400 500 span length in m
Figure 4-21 on page 127 shows the five load combinations used to calculate critical sectional forces at the
quarter-point and midpoint of each design concept. Live load cases 1, 2, and 3 cause maximum sectional
forces at the springing lines and at the crown. These sectional forces are combined with those caused by
dead load, creep, shrinkage, and temperature. Live load cases 4 and 5 cause maximum sectional forces at
the quarter-points. These sectional forces are combined with only those caused by dead load.
253
6. New Concepts for Arch Bridges
Table 6-26 shows which of the five live load cases and accompanying loads cause maximum sectional
forces in each design concept. Results show that self-stiffened and partially deck-stiffened concepts with
span-to-rise ratios of 4:1 were governed by sectional forces at the quarter-points caused by truck or lane
loads. The effects of restrained deformation were not critical for these concepts. Nearly all other concepts
were governed by live load cases 2 or 3. This shows that the effects of restrained deformations become in-
The second-order flexible system moments associated with the governing load cases at ultimate limit states
are shown in Table 6-27. These moments are taken as the greater of the quarter-point and midspan mo-
ments. In most cases, self-stiffened concepts have the least bending moment demands among the arch sys-
tems considered.
Table 6-27. Maximum flexible system moments at ultimate limit states among all concepts
maximum flexible system moments at ULS among all concepts, in MN·m
concept ID class of arch system L = 50 100 150 200 250 300 350 400 m
1 to 8 self-stiffened L:f = 4:1 11 23 36 49 69 94 126 161 •minimum values among
33 to 40 deck-stiffened 4:1 11 23 39 54 69 100 126 158 triplets are in red type
65 to 72 partially deck-stiffened 4:1 12 24 37 51 74 105 147 195
9 to 16 self-stiffened 8:1 21 31 48 75 115 165 226 308
41 to 48 deck-stiffened 8:1 14 31 59 86 122 186 243 321
73 to 80 partially deck-stiffened 8:1 12 28 49 78 125 196 252 370
17 to 24 self-stiffened 12:1 28 42 74 128 207 331 495 708
49 to 56 deck-stiffened 12:1 18 49 97 185 312 430 603 868
81 to 88 partially deck-stiffened 12:1 17 46 88 167 353 509 707 1060 MN·m
254
6. New Concepts for Arch Bridges
150
self-stiffened arch concepts
deck-stiffened arch concepts 100
partially deck-stiffened arch concepts
moments caused by load case 4 or 5 50
0
span in m 0 100 200 300 400 0 100 200 300 400 0 100 200 300 400
Figure 6-26 plots the governing bending moment results from Table 6-27. Moments are normalized by
span and span-to-rise ratio and are thus expressed in kN units of force. As shown by the first and second
graphs on the left, concepts with spans-to-rise ratios of 4:1 and 8:1 have about the same normalized mo-
ments over the entire range of spans considered in the study. This suggests that for the span lengths con-
sidered, doubling the span-to-rise ratio from 4:1 to 8:1 results in a doubling of moment demand. The
circled data points in the figure refer to governing moments that are caused by dead and live loads only.
These occur only in the concepts with span-to-rise ratios of 4:1. By comparing the centre and right-hand
graphs in the figure, results show that the effects of restrained deformation becomes increasingly severe as
span-to-rise ratio is increased from 8:1 to 12:1 for spans ≥200 m. As the arch system is made flatter and
longer, moments caused by restrained deformation grow exponentially. Thus, there are practical limits to
255
6. New Concepts for Arch Bridges
Table 6-28. Portions of governing moments caused by first-order live loads among all concepts
SRUWLRQRIPD[LPXPÁH[LEOHV\VWHPPRPHQWVFDXVHGE\ÀUVWRUGHUOLYHORDGVDPRQJDOOFRQFHSWVLQ
L = 50 100 150 200 250 300 350 400 m
9 to 16 VHOIVWLfIHQHG 8:1 43 59 57 47 41 37 34 30
41 to 48 GHFNVWLfIHQHG 8:1 66 58 47 43 39 34 32 29
73 to 80 SDUWLDOO\GHFNVWLfIHQHG 8:1 70 61 54 42 35 28 29 24
17 to 24 VHOIVWLfIHQHG 12:1 37 48 42 32 27 22 18 16
49 to 56 GHFNVWLfIHQHG 12:1 56 40 30 21 17 16 14 12
81 to 88 SDUWLDOO\GHFNVWLfIHQHG 12:1 56 43 33 22 15 14 13 9
Table 6-28 shows first-order live load moments expressed as percentages of maximum moments at ulti-
mate limit state among all concepts. This set of data measures the validity of using first-order analysis con-
sidering dead and live loads only. The data is presented graphically in Figure 6-27. The results show that
this simple analysis is only accurate for concepts with spans of ≤100 m and span-to-rise ratios of 4:1. For
concepts with spans of >100 m and span-to-rise ratios of 4:1, second-order analysis is needed to account
for the additional moments caused by deformations of the arch off the pressure line. The effects of re-
strained deformation need not be considered for these concepts. For all other concepts (those with span-
to-rise ratios greater than 4:1), second-order analyses including the effects of restrained deformations are
needed because first-order live loads contribute less than 70% of the total flexible system moment at ulti-
mate limit state. In these cases, bending moment demands would be grossly underestimated if the effects
L:f = 8:1
20
L:f = 12:1
0
0 100 200 300 400 500 span length in m
256
6. New Concepts for Arch Bridges
The most common geometrical quantities that were found to be satisfactory among design concepts are
summarized in Table 6-29. Concepts whose demands required proportions greater than those shown in
the table include: (1) some concepts with spans of 50 m due to the relatively low amount of natural
prestress in the arch, and (2) most concepts with span-to-rise ratios of 12:1 due to high sectional force de-
Table 6-29. Most common geometrical quantities used among design concepts
class of system element section type span-to- slab width of box ÷
depth ratio thickness width of deck
partially deck-stiffened arch deck hollow box 90:1, 100:1 100-200 mm 1/2
arch hollow box 90:1, 100:1 100-200 mm 1/2
Regarding efficiency in terms of the consumption of materials, the following observations were made:
• Among the arch systems considered, deck-stiffened systems consume the least amount of concrete
for spans of 50 m, while self-stiffened systems consumed the least amount of concrete for spans of
150-400 m. For spans of 100 m, all systems consumed approximately the same amount of concrete.
• For span-to-rise ratios of 4:1 and 8:1, all concepts consistently had less than 50% of the minimum
equivalent slab thicknesses of existing concrete arch bridges in the database. Concepts with span-to-
rise ratios of 12:1 were not able to achieve the same reduction in concrete volume.
• Concrete volume in the deck and in the arch tend to be roughly the same, regardless of which ele-
257
6. New Concepts for Arch Bridges
• In terms of visual slenderness, concepts tended to have span-to-system-depth ratios between 39:1 and
50:1, which is on par with the ratios observed among existing concrete arch bridges in the database.
• Concepts tended to have normalized system moments of inertia that were less than or about the
same as those observed among existing concrete arch bridges in the database.
• Concepts tended to have system radii of gyration that were greater than or about the same as those
• Concepts tended to have system slenderness ratios between 39 and 60, which is not as high as many
existing concrete arch bridges in the database. In this range of slenderness ratios, additional deflec-
tions and moments caused by second-order effects can be significant. For these systems, reduced
member capacity N-M* interaction diagrams are only slightly less than sectional capacity N-M inter-
action diagrams.
• In terms of arch shallowness, concepts with span-to-rise ratios of 4:1 and 8:1 tended to have modified
slenderness ratios much greater than 20. Permanent loads are thus carried primarily in compression
along the arch. Concepts with span-to-rise ratios of 12:1 tended to have modified slenderness ratios
around 20 or less. Their responses to load are thus affected by significant amounts of beam action.
Regarding loads, deflections and complexity of analysis, the following observations were made:
• Extra weight factors of 1.4 to 1.6 can be used to estimate the total weight of the structure based on the
weight of continuous, longitudinal concrete. Shorter spans of 50 m tended to have higher extra
weight factors while longer spans of 400 m tended to have lower factors.
• Most concepts were successfully designed to have crown deflections that were less than 5% of the ini-
tial rise of the arch at ultimate limit states. This suggested deflection limit ensures that results from
258
6. New Concepts for Arch Bridges
• For most concepts, the bending moments caused by restrained deformations and second-order
effects are significant, and are of similar magnitude to those caused by live load. As such, second-or-
6.4 Recommendations
In this section, some of the the most important questions regarding the potential opportunities offered by
the arch concepts proposed in this chapter are addressed. Recommendations are made in response to sev-
The parametric study provided examples of arch concepts that are designed for spans of up to 400 m.
These designs were shown to be much lighter than the longest bridges made using conventional concrete
arch technology.
Figure 6-26 on page 255 shows that moments caused by the effects of restrained deformation and second-
order effects grow exponentially with span length. Thus concepts cannot simply be scaled up from 400 m
to 1000 m and be expected to be satisfactory. This means that slab thicknesses much greater than 200 mm
are needed if span-to-depth ratios are kept more or less the same, resulting in heavy segments and greater
complexity of construction. Arches spanning 600 m are likely possible but would require higher equival-
ent slab thicknesses than those used in the parametric design study. Although these long-span concrete
arches are probably possible, and would use materials efficiently to carry loads in the final system, the cost
of temporary works and throwaway materials would outweigh any cost-savings related to the consumption
of concrete. For such long spans, concrete arches would not be economically competitive against cable-
stayed or suspension-type structural systems, since these efficient systems can be built with essentially zero
throwaway materials. These considerations impose economic barriers that limit the applicability of using
ultra high-performance fibre-reinforced concrete arches for spans greater than 400 m.
259
6. New Concepts for Arch Bridges
As the rise of an arch is decreased, its behaviour transitions from carrying permanent loads in pure axial
compression along the pressure line to carrying permanent loads in bending as in conventional beam-type
structures, as discussed in Section 4.4.3. The parametric design study shows examples of design concepts
with span-to-rise ratios of 12:1, which have modified slenderness ratios that are slightly less than the pro-
posed minimum threshold value of 20. Adding material to these designs tended to increase structural de-
mands faster than they increased member capacities. Thus the maximum span-to-rise ratio that should be
used for these proposed systems is less than 12:1, but higher than 8:1.
Using conventionally reinforced concrete, the Infant Henrique Bridge was able to achieve a span-to-rise
ratio of 11.2:1 (Fonseca and Mato 2005). In order to achieve this, it required a very stiff deck girder and a
very wide voided slab for the arch. A similar design could be made using ultra high-performance fibre-re-
inforced concrete but the savings in material would be minimal, since the axial rigidity of the arch would
Slenderness can be interpreted in three different ways: (1) visual slenderness as measured by span-to-depth
ratios, (2) sensitivity to second-order effects as measured by slenderness ratio, and (3) thinness of elements
Regarding visual slenderness, the highest span-to-system depth ratio achieved among concepts was 50:1.
Greater visual slenderness can be achieved by reducing the depth of deck and arch, and increasing their
slab thicknesses. Span-to-system-depth ratios only slightly higher than 50:1 are possible because deflec-
tions and second-order effects become increasingly severe as the flexural rigidity of the system is reduced.
If the arch alone is considered, the highest span-to-arch-depth ratio achieved among concepts was 200:1,
The highest system slenderness ratio achieved among concepts was 60, which is not high compared to
those in existing concrete arches. This quantity is deceiving, since higher slenderness ratios in these arches
260
6. New Concepts for Arch Bridges
are a result of using sections that are solid, or at least more solid than the sections proposed for the con-
cepts. For example, if a solid member and a hollow member were given the same effective length, depth
and width, it is the solid member that would have the higher slenderness ratio. In this comparison, differ-
ences in slenderness ratio have no impact on the external visual slenderness of the member.
Minimum thicknesses of concrete components are governed by: (1) geometrical requirements for obtain-
ing random fibre distributions, (2) concrete cover requirements for reinforcing steel, and (3) by local plate
buckling problems. Minimum thicknesses for ensuring a random distribution of fibres is usually taken as
three to five times the fibre length, which is 30-50 mm for the mix design considered in this thesis. Because
no conventional reinforcing steel is used in the concepts, only cover requirements for internal prestressing
steel need to be considered. The proposed segment end ribs in this study are specified to be 100 mm thick,
which allows for just enough cover for transverse prestressing reinforcement. Reinforcement-free mini-an-
chorages, such as those used in the Sun-Yu Bridge in Korea (Huh and Byun 2005), allow for post-ten-
Concepts for thin ribbed slabs using ultra high-performance fibre-reinforced concrete were investigated by
Spasojević (2008). Using various flange and web thicknesses of 40-60 mm combined with internal, post-
tensioned monostrands, Spasojević found that these light-weight ribbed slabs were capable of transmitting
wheel loads over spans of 12 m between supporting girders. Her findings help validate the use of thin, ultra
high-performance fibre-reinforced concrete slabs for the decks of the concepts proposed is this thesis.
Problems associated with local buckling also limit how thin components can be made, as discussed in Sec-
tion 6.1.5. Based on a rudimentary stability analysis of ultra high-performance fibre-reinforced concrete
plates, it was shown that plate buckling could be avoided if wall slenderness ratios less than 24.5 were used.
Wall slenderness ratios are calculated as the width of plate between line supports divided by plate thick-
ness. According to this limit, 50 mm thick plates with maximum plate widths of 1230 mm are capable of
reaching the peak compressive stress of the material without buckling. The maximum allowable plate
width can be extended though the use of longitudinal stiffeners, as is proposed for the design concepts.
261
6. New Concepts for Arch Bridges
Which arch system is most efficient? Should the deck be used to stiffen the arch?
Based on the results of the parametric design study, self-stiffened arch concepts, in most cases, were found
to consume the least amount of concrete among the arch systems considered (see Table 6-15 on page 244).
This result agrees with the conclusion made at the end of Section 4.7: that it is most efficient to maximize
the use of precompression in the arch for carrying flexible system moments. This also serves to isolate deck
girders from the global system, which minimizes their bending moment demands.
For span lengths of 50 m, self-stiffened arches were not the most efficient because the natural prestress of
the arch caused by dead load was low in comparison to the imposed flexural stresses. In these cases, deck-
stiffened arch systems consumed the least amount of concrete. The resistances of deck girders in these sys-
In spite of these findings, designers should not automatically dismiss partially deck-stiffened systems, since
all concepts in the parametric design study were able to reduce the consumption of concrete by at least
50%, except for those with span-to-rise ratios of 12:1. Because the proposed concepts are all more efficient
than the existing concrete bridges in the database, a greater set of criteria should be used to decide which
arch system should be used, such as ease of construction, consumption of prestressing steel, and visual im-
pact on the surrounding environment. For example, one benefit of the proposed partially deck-stiffened
concepts is that they have nearly identical arch and deck segments. This repetition of geometry may lead to
significant economies of scale, since the same set of forms could be modified to precast both the deck and
arch. As a second example, self-stiffened concepts have the highest arch flexural rigidity among the sys-
tems, which means that they have the highest resistances against buckling during the stages of construc-
tion when the arch is unsupported by temporary works. These examples emphasize the importance of hav-
ing a design-oriented mindset that is able to interpret and adapt to the specific requirements of given
designs. Thus, the proposed concepts should be treated as malleable starting points for design rather than
262
6. New Concepts for Arch Bridges
Would the addition of rebar improve the efficiency or performance of the proposed systems?
Supplementing ultra high-performance fibre-reinforced concrete with rebar results in moments of resist-
ance that are more or less the same as those attained in conventional reinforced concrete members. This is
because moments of resistance are typically governed by the yield force of steel, rather than by the com-
The tensile strength provided by the combination of the fibre-reinforced concrete and steel reinforcing
bars is not equal to the sum of tensile strengths provided by each individual component. This is because
the fibre-reinforced concrete reaches its maximum tensile stress at a strain lower than the yield strain of
steel. Thus, in terms of strength requirements, it is best to design members such that tensile strength is
provided: (1) by the fibre-reinforced concrete only, (2) by external unbonded prestressing tendons, which
can be used to delay initial concrete cracking, or (3) by mild steel reinforcing bars only. The first two op-
tions allow for minimum wall thicknesses to be used since conventional cover requirements are not applic-
able. The last option can be used if required wall thicknesses already exceed minimum cover requirements
The addition of mild steel reinforcing would be useful in members that have high ductility demands. Al-
though ultra high-performance fibre-reinforced concrete is more ductile than ordinary concrete, it is less
ductile than reinforced concrete. Once localized cracks form in ultra high-performance fibre-reinforced
concrete members, curvatures at the cracked section can only be increased if loads are reduced. Because of
the long yield plateau of rebar, reinforced concrete members can withstand large rotations before their mo-
ments of resistance are reduced. It would thus be prudent to supplement ultra high-performance fibre-re-
inforced concrete members with rebar at critical locations in the structure. For the proposed design con-
cepts, mild steel reinforcing or additional prestressing steel should be provided at the sections of the deck
263
6. New Concepts for Arch Bridges
Will these concepts with their very thin concrete components be sensitive to impact loads?
Regarding structural robustness and sensitivity to impact loads, analytical and experimental studies have
done by Habel and Gauvreau (2008) on 50 mm ultra high-performance fibre-reinforced concrete plates.
Specimens were subjected to impact loads using drop weights. Results showed that the strength of the ma-
terial increased with increasing strain rate, for strain rates of up to 2 s-1. At this rate, the material continued
to exhibit strain-hardening characteristics and multiple cracking. The bending resistance increased by
more than 25% relative to static behavior. Given these results, it seems that ultra high-performance fibre-
reinforced concrete has the capacity to withstand accidental or exceptional loads. Future research, how-
Based on the results of the parametric design study, ultra high-performance fibre-reinforced concrete is an
excellent material to use in arch bridges. This material allows for significant reductions in the consumption
of concrete and in the total weight of the structure, relative to arches made using conventional concrete
technology. These reductions should result in considerable savings related to the costs of raw materials,
temporary works, and hours of skilled labour. These reductions will also reduce the complexity and dura-
tion of construction, resulting in further cost-savings. The enhanced properties of ultra high-performance
fibre-reinforced concrete were found to positively affect the behaviour and efficiency of arch systems in the
following ways:
• The high compressive strength of the material allowed for high levels of precompression in the arch,
which in turn increased arch flexural capacity. The same applied to the high levels of allowable
prestress in the deck. These increases in compression and bending capacities enabled decks and
• The high tensile strength of the material allowed for the elimination of most mild reinforcing in the
structure. Because the material has a higher cracking strength than conventional concrete, the per-
264
6. New Concepts for Arch Bridges
• Creep and shrinkage strains are less than in conventional concrete. This reduces long term-deflec-
• The short fibre length of the embedded steel fibres allowed for very thin components to be used.
265
Chapter 7. Conclusions and Summary
In this chapter, the principal findings of this thesis related to the the design and validation of new systems
for arch bridges using ultra high-performance fibre-reinforced concrete are presented. A concise list of
conclusions is given in Section 7.1. Summaries of material behaviour, sectional behaviour, member beha-
viour, and arch system behaviour are given in Sections 7.2 to 7.5. This is followed by summaries of the
comparative study of existing concrete bridges and the parametric design study in the last two sections.
Sections 7.2 to 7.7 are written assuming that the reader has not thoroughly read the main body of the thes-
is. Hence, the most important equations and figures are reprinted and briefly explained. Potential research
topics for future work are also identified and discussed in these sections.
7.1 Conclusions
The primary conclusions drawn from the parametric design study are:
• It is possible to reduce the weight and consumption of concrete in arch bridges by at least 50% by us-
ing ultra high-performance fibre-reinforced concrete without conventional steel reinforcing bars.
This reduction applies to spans of 50-400 m and span-to-rise ratios of 4:1 and 8:1.
• Lightness is achieved primarily by the thinning of structural components (i.e. webs, flanges, and
slabs) and the efficient use of precompression in the arch, and rather than by the decrease of bending
stiffness.
266
7. Conclusions and Summary
• Deck-stiffened arch systems consume the least amount of concrete for spans of 50 m, while self-sti-
ffened arch systems consume the least amount of concrete for spans of 150-400 m. For spans of 100
• The use of lightweight segments and thin components requires that stiffeners and ribs be used to pre-
vent local buckling. These additional components increase the complexity of the segmental precast-
ing process.
The primary conclusions drawn from the comparative study of existing arch bridges are:
• The typical range for span-to-rise ratios in concrete arches is 2.3 to 8.0.
• The envelope of minimum equivalent slab thicknesses hV observed among existing arch bridges in
# L
% + 0.569 m 0 ! L ! 193m
% 1276
hV = $
% L2 L
" + 2.21m L > 193m
% 22800 m 61.7
&
The primary conclusions drawn from the study of shallow arches are:
• The transition between fixed arch behaviour and fixed beam behaviour was found to be governed by
the modified slenderness ratio, which is proportional to arch rise and inversely proportional to sys-
• To ensure that given arch designs behave efficiently, a minimum modified slenderness ratio of 20 is
proposed. This threshold ensures that more than 90% of uniform loads are carried in axial compres-
• Based on an empirical relationship between system radius of gyration and span length, the following
267
7. Conclusions and Summary
L ! L $
< 16.1 #
f " L + 48.4 m &%
• There remains an opportunity to design efficient shallow arches that have span-to-rise ratios greater
than 11.2, which is the highest ratio ever used in a concrete arch bridge.
The primary conclusions drawn from the study of slender members under compression and bending:
can reach ultimate load before reaching sectional failure. Thus, a reduced member capacity interac-
tion diagram must be used when designing for ultimate limit states.
• Ultimate loads of 4 slender column tests were overestimated by general method predictions by up to
14.5%. Considering the ±2 mm error in measuring initial eccentricities of load (representing ±16.2%
to ±37.7% of the eccentricity), the general method predictions were fairly good.
• Member capacity interaction diagrams calculated using the simplified design method were shown to
be accurate and conservative relative to those calculated using the more rigorous general analysis
method.
268
7. Conclusions and Summary
Sets of ultra high-performance fibre-reinforced concrete cylinders and prisms made from local materials
were tested to obtain stress-strain models suitable for use in design calculations. The concrete specimens
exhibited, on average, a 28-day compressive strength of 106 MPa, a tensile cracking strength of 7 MPa, and
a tensile strength of 14.6 MPa. These high strengths are achieved by the use of water reducers, a low water-
to-cement ratio, no coarse aggregate, and 5.5% steel fibre content by volume. Based on test results, the mul-
tilinear stress-strain curves shown below are proposed for use in design calculations. The proposed materi-
al model is assumed to be valid for compressive strengths between 100 MPa and 140 MPa, and is intended
to be used specifically for the mix design developed at the University of Toronto.
The stress-strain response in compression can be modified according to three parameters: compressive
strength fc!, tangent modulus of elasticity in compression Ec, and secant modulus at peak stress Ec!. The
stress-strain response in tension is based on four parameters: strain and stress at cracking, %cr and fcr, and
strain and stress at peak tensile stress, %t! and ft!. Post-peak tensile stresses are neglected.
50
0
0 !c′ Compressive strain cylinder test data
material model
0
0 4·10-3 8·10-3
Compressive strain
269
7. Conclusions and Summary
Obsolete
Et = 46.7 GPa
20 prism test data only !cr = 0.15·10-3 fcr! and !cr! based on
valid before cracking averages from prism data
fcr = 7.0 MPa
fcr
!t! = 1.5·10-3 !t! estimated using results
ft! = 14.6 MPa from direct tension tests by
10
Habel and Gauvreau (2008)
Tensile strain
A compressive strength of 120 MPa is used for the parametric design study, representing a nominal
strength that can readily be achieved by many different ultra high-performance fibre-reinforced concrete
mixes, without the use of specialized heat or steam treatment. Although the 28-day cylinder tests exhibited
only 106 MPa at 28 days, this modest increase in assumed strength can be attained by further optimizing
the selection and mix proportions of raw materials and by improving the equipment and procedure used
Tests done by others on similar mixes of ultra high-performance fibre-reinforced concrete resulted in av-
erage long-term shrinkage strains of 550·10-6 and long-term creep strains of 0.8. These quantities, which
are used in the design study, result in long-term strains that are much lower than in conventional concrete.
One possible objective for future research is to improve the ductility response of the material. While ultra
high-performance fibre-reinforced concrete is more ductile than plain concrete, both in compression and
tension, it is less ductile than concrete that is reinforced with mild or prestressing steel. This has implica-
tions on the resilience of the structure and its ability to undergo large rotations and deformations. Rather
than resorting to minimum rebar requirements, it might be adequate to provide reinforcement only at crit-
ical sections of the structure, designed specifically to improve the ductility of the member. These bars
would only have to match the tensile force provided by the ultra-high fibre-reinforced concrete.
270
7. Conclusions and Summary
Using the proposed stress-strain curve, the axial force-moment-curvature response of a 100 mm by 100
mm square cross-sections was investigated in Section 3.2.2. Contours of equal curvature for this cross-sec-
tion are plotted below in axial force-moment space. These contour plots provide: (1) a visual representation
of the nonlinear behaviour of the section, and (2) a means of calculating moment-curvature diagrams
without using iterative calculation. The most important observations made from this analysis are: (1) the
section behaves linear-elastically up until initial cracking, and (2) limited amounts of moment-curvature
ductility can be expected if axial forces are less than the axial force at the balanced condition.
–600 100 mm
! = 10 20 30 40 50
100 mm
! = 30 rad/km ... ...
–400
Partial reprint of Figure 3-3. Contours of equal curvature and corresponding moment-curvature diagrams
271
7. Conclusions and Summary
Moment-Curvature
in MN·m and rad/km N=-55.2 MN
Axial force- Axial force-
Moment interaction Secant stiffness M
in MN and MN·m in MN and EI0
150
Mu
N N
Mcr
post-peak response
!200 depends on the control
of localized cracking
EIu
!150
0
"cr "u "
!100 EIcr
EI10 EIcr
Mcr
Mu EIu
!50
0
0 0 0.5 1 1.5 "
0 50 100 150 M 0 10
EI EI!
!EI0
Secant stiffness-Curvature
(a) (b) in EI0 and rad/km N=-55.2 MN
(c)
Reprint of Figure 3-23. Sectional analysis of an ultra high-performance fibre-reinforced concrete box
The sectional response of a hollow, ultra high-performance fibre-reinforced concrete box section was
investigated in Section 3.5. Based on the results of the analysis (shown in the figure above), expressions for
secant stiffness at ultimate EIu are proposed as follows, where N is axial force, N! is maximum axial force in
+ # N &
- 1
EI 0 % +1 0 ) N ) 12 N "
-
2
$ N " ('
EI u ! ,
- # N&
1
EI 0 % 2 * ( 1
N" ) N ) N"
- 2
$ N"'
2
.
These conservative secant stiffness values are used in design calculations to account for the softening of
systems caused by material nonlinearity. These expressions result in effective secant stiffness values
between 0.5EI0 and 0.75EI0, which are higher those exhibited by conventionally reinforced concrete
sections.
272
7. Conclusions and Summary
The response of slender concrete members under combined compression and bending was investigated in
Chapter 3. The analytical model considered is shown in the figure below. In slender columns, additional
secondary moments are caused by deflections of the column away from the line of action of force. In very
slender columns, these second-order effects can cause the ultimate load of columns Q* to be reached be-
fore sectional failure is reached. Hence, the member capacity of very slender columns is always less than
M
N 250
16mm
Q 32mm
0
0 4 8 12 16
Bending moment M
(a)
In order to describe this behaviour analytically, a general analysis method for calculating reduced member
capacity N-M* interaction diagrams is presented in Section 3.2.4. The input parameters are pin-to-pin
length, sectional geometry, and stress-strain curve. If initial eccentricity of load is also given, the general
method can also be used to calculate the load-deflection Q-w response of the column, as presented in Sec-
tion 3.2.5. A sample reduced N-M* interaction diagram is shown in the figure above as a dotted curve. The
dashed curve represents sectional failure. The solid curves represent the various possible sectional re-
sponses of the mid-length section, each corresponding to different initial eccentricities of load. The apexes
of the solid curves all correspond to different ultimate limit states of the column, each with different com-
273
7. Conclusions and Summary
The general method accounts for material nonlinearity by using nonlinear moment-curvature diagrams.
Geometric nonlinearity is accounted for by column deflection curves, which map centreline deflections of
a column relative to a fixed line of action of load. Column deflections curves are calculated by first assum-
ing a mid-length moment, and then solving for the end eccentricity of the column, using a numerical in-
tegration procedure. A family of column deflection curves for a given load Q is generated by changing the
assumed mid-length moment. Among these curves, the one at ultimate limit state Q=Q* is the one with
the greatest end eccentricity. Repeating this procedure for many different values of Q produces the reduced
N-M* interaction diagram. All these calculations are implemented in the computer program QULT, which
0.5
nt —ncr
0
ncr 0.5 1 1.5
nt
m
mn=0
QE
nE = = normalized axial force at Euler buckling, but not > 0.8
A·fc!
f
ncr = cr = cracking stress divided by compressive strength
fc!
f!
nt! = t = tensile strength divided by compressive strength
fc!
mn=0 = normalized ultimate moment when N=0
To reduce the analytical effort required by the general analysis method, a simplified design method is pro-
posed. This method produces a reduced member capacity diagram N-M* that approximates those given by
the general method. The envelope is defined in terms of normalized axial force and normalized moment,
274
7. Conclusions and Summary
as given in the figure above. This envelope was shown to be conservative and accurate relative to the more
precise general method calculations used for the arch concepts in the parametric design study (see Figure
Four eccentrically loaded columns made from ultra high-performance fibre-reinforced concrete were tes-
ted to failure at the University of Toronto. These columns had nominal slenderness ratios of 60, 80, 100,
and 120. The longest column was 3466 mm long with a square cross-section of 100 mm by 100 mm. The
following measurements were recorded: initial eccentricity of the load, deflection of the column at mid-
span, strains along the flexural compression face of the column, displacements of the centreline of the
column, and the applied load. From this data, load-deflection and axial force-moment responses from the
experiments were compared to responses predicted by the general method (see Figure 3-17 on page 67).
Observed and predicted ultimate loads were in good agreement, with percent differences of +6.1%, +0.8%,
+14.5%, and +10.1%, which are acceptable given the assumed margins of error. Hence, the general analysis
method is validated as a tool for modelling the behaviour of eccentrically loaded columns.
For future work, these column tests can be repeated with greater quantity to increase the statistical signi-
ficance of the results. Load tests of hollow members would better represent the type of highly efficient
members that are proposed in the design concepts, and ensure the applicability of the general analysis
A large quantity of technical information on the statical analysis of arch bridges was presented in Chapter
4. The most important topics and insights have already been summarized in Section 4.6 on page 140. The
The form of arches is chosen such that dead loads are carried in pure compression along the centroidal
axis of the arch. Nonuniform loads and restrained deformations are carried by both the deck and arch in
275
7. Conclusions and Summary
bending. First-order bending moments diagrams, bending moment influence lines and bending moment
envelopes are shown in Figures 4-6 through 4-10, starting on page 93.
The following three assumptions were used to define a simplified statical system for arches: (1) spandrel
columns are pin-connected to the arch and deck, (2) the flexural rigidity of the arch projected on to the
horizontal axis is constant over the entire span and is equal to the flexural rigidity of the arch at the crown,
and (3) vertical deflections of the deck and arch are equal along the length of the span. The validity and
consequences of these assumptions are illustrated and discussed in detail in Section 4.4.1. The overall beha-
viour that results from these assumptions is that bending moments imposed on the system are shared by
the arch rib and deck girder in proportion to their respective flexural rigidities.
Using the simplified statical system, the force method was used to solve closed form solutions of the axial
forces, bending moments, and deflections at critical points in fixed arch system. First this was done for
non-shallow arches, for which axial deformations are neglected in compatibility equations. Solutions for
this analysis are shown in Table 4-3 on page 109. The force method was then resolved for shallow arches,
for which axial deformations are included in compatibility equations. These shallow arch solutions are
The transition between fixed arch behaviour and fixed beam behaviour was identified by comparing sec-
tional forces in shallow arches and fixed beams. The governing non-dimensional ratio, called the arch-
beam parameter " is defined as follows, where f is arch rise, EIsys is system flexural rigidity and EAcrown is
45 EI sys
!= "
4 f 2 EAcrown
A related shallowness parameter called the modified slenderness ratio, is defined as follows, where rsys is
276
7. Conclusions and Summary
EAcrown 2 f
!f = 2 f =
EI sys rsys
The effect of these parameters are illustrated in the figure below. As (f is decreased from infinite to zero
(i.e. " is increased from 0 to 1), sectional forces are changed as follows: (1) bending moments of the system
caused by uniform loads are increased from zero to those in a fixed-fixed beam, and (2) horizontal arch re-
actions caused by uniform loads are decreased from those in a non-shallow arch to zero. Based on results,
a minimum modified slenderness ratio value of 20 is proposed. This threshold ensures that greater than
90% of permanent loads are carried in pure compression along the pressure line.
0
0 20 40 60
Modified slenderness ratio !f
Reprint of Figure 4-17. Transition between fixed arch and fixed beam
The modified slenderness ratio (f has also been used by others to determine whether antisymmetric bifur-
cation buckling or symmetric snap-through buckling modes occur in shallow arches. The buckling resist-
ance of shallow fixed arches are less than the buckling resistance of non-shallow fixed arches, and are con-
siderably less when (f < 19.8, for which symmetric buckling modes are critical.
277
7. Conclusions and Summary
When fixed arches buckle anti-symmetrically, their mode shape is similar to the second buckling mode
shape of straight fixed-fixed beams. Thus an appropriate model for non-shallow arch buckling is that of an
idealized straight fixed-fixed beam that is laterally supported at midspan. The flexural buckling resistance
HE of non-shallow arches can be calculated as follows, where EIsys is the sum of flexural rigidities of the
EI sys
HE = ! 2
( kS )2
This quantity along with modified slenderness ratio (f can be used to calculate the buckling resistance of
$ "2 '
H E, f = & 0.6 ± 0.4 1 ! 30.7 2 ) * H E # f > 19.8
% #f (
With these buckling resistances, second-order deflections w and moments of arches can be calculated us-
ing Vianello’s method of successive approximations. First-order deflections w0 calculated using the force
method are multiplied by a deflection magnifier, which depends on the ratio of horizontal reaction H to
1
w(x) = w0 (x)
H
1!
HE
To account for material nonlinearity, the flexural rigidity of the system EIsys is taken as the secant stiffness
of the system κ·EIsys. At serviceability limits states, secant stiffness factor κ can be taken as unity, represent-
ing the uncracked, gross flexural rigidities EI0 of the deck and arch. At ultimate limit states, κ can conser-
vatively be taken between 0.5 and 0.75, representing the expected material softening of the system. Expres-
278
7. Conclusions and Summary
The eccentricity e of the deformed arch member and the line of action of load is calculated by multiplying
second-order deflections by the eccentricity ratio &, which is 0.733 or 0.5 depending on the type of loading
q q
H H H H
Z = second-order deflection
Z Z at critical section
H
H H H H undeformed line of action
H H H deformed line of action
e=ƦÃZ e = second-order eccentricity
e=ƦÃZ
e=0.733ÃZ of load at critical section
e=0.5ÃZ
Applying these calculation methods can get complicated when considering factored load combinations of
dead load, live load, creep, shrinkage, and temperature changes. Long-term deflections have to be calcu-
lated separately and then expressed as an equivalent first-order short-term deflection before they are added
to actual first-order deflections. A detailed list of calculations for total sectional force demands are sum-
marized in Table 4-4 on page 128. Sample calculations of maximum sectional forces at the crown and
Many of the analytical methods for calculating forces in arches have been verified experimentally in the
past on full-scale arch bridges and on small-scale linear-elastic models. Load tests on fixed, shallow arches
made from ultra high-performance fibre-reinforced concrete would further substantiate the validity of the
simplified methods presented in this thesis. Other important analytical studies that can be considered for
future work include: (1) the dynamic response of arch systems to seismic and wind loading, and (2) closed
form equations for shear forces in shallow arches. Inertial forces caused by earthquakes are usually mitig-
ated in structures by providing sufficient ductility or damping at critical locations in the system. The added
complication in arch bridges is that deformations of the arch off the pressure line will cause potentially
severe second-order moments. With regard to the second topic, shear calculations can be derived from the
279
7. Conclusions and Summary
force method as done for axial force and moment. Although shear requirements are typically non-critical
in conventionally reinforce concrete arches, they might be critical as more efficient box sections are used.
Articles and drawings of fifty-eight existing concrete arch bridges built over the last century were collected
and processed into a comprehensive database. Because a majority of the reference bridges were collected
from authoritative engineering journals, the database is considered to be representative of the current
state-state-of-the-art of concrete arches bridges. The database serves as a reference guide for bridge design-
ers, as well as a tool for evaluating the material efficiency and feasibility of new design concepts. The com-
The following sets of data were collected for each bridge: name, location, method of construction, date of
completion, span, rise, number of spandrel columns, arch fixity, and sectional properties of arch, deck, and
spandrel columns. From these quantities, various geometric ratios and systems properties relevant to the
slenderness, shallowness, and degree of deck-stiffening of arch systems are calculated and compared. A
summary of all the trends observed from the database is given in Section 5.4 on page 194. Some of the
Advances in arch construction have allowed for concrete arch bridges to span up to 420 m, as in the
Wanxian Yangtze Bridge. Modern cantilever methods of construction have reduced the amount of skilled
labour and throwaway materials through the increased mechanization of work. The construction methods
for most of the longest spanning concrete arch bridges can be categorized as follows: (1) cantilevering by
using an effective truss, where ties at the level of the deck and along the diagonals form the tension mem-
bers, and the arch and spandrel columns form the compression members, (2) cantilevering using stays em-
anating from temporary pylons above the springing lines, (3) cantilevering a steel pilot truss that is used
temporarily to support formwork and used permanently to reinforce the arch, and (4) cantilevering by ro-
tation, where the two halves of the arch are constructed near the springing lines and then rotated vertically
280
7. Conclusions and Summary
or horizontally to their final positions. All these construction methods would also be suitable for building
Of the bridges in the database, 46 out of 55 bridges (84%) have span-to-system-depth ratios in the range of
24 to 48. Span-to-system-depth ratios are better measures of the overall proportions of arch bridges than
are span-to-arch-depth ratios because they account for the effects of deck-stiffening. This ratio largely
effects the visual slenderness and transparency of arch bridges, as well as the internal lever arm of the arch
and deck girder. New design concepts in ultra high-performance fibre-reinforced concrete should aim to
have span-to-system-depth ratios toward the higher end of the range to maintain high transparency and a
slender profile.
Equivalent slab thicknesses by concrete volume hV were calculated for each bridge in the database. An
envelope of minimum slab thicknesses achieved by existing concrete technology is given as a function of
span length L:
# L
% + 0.569 m 0 ! L ! 193m
% 1276
hV = $
% L2 L
" + 2.21m L > 193m
% 22800 m 61.7
&
This minimum equivalent slab thickness envelope serves as a benchmark for the design study. If satisfact-
ory design concepts can achieve slab thicknesses significantly less than this envelope, then significant con-
A simple statical analysis was used to estimate the effects of live load in fixed arch bridges in the database.
Comparisons of live load eccentricities, or displacements of resultant axial forces away from the centroid
of the arch, showed that the effects of live load diminish with increasing span length. This decreasing sens-
itivity to live load is analogous to the trivial effects of live loads in long-span suspension bridges. The
effects of restrained deformations, however, increase with increasing span length, and thus tend to govern
281
7. Conclusions and Summary
Because the observed trends had relatively little scatter considering the empirical nature of the study, in-
creasing the sample size of the database would not likely result in many new insights into the typical pro-
portions of concrete arch bridges. Thus, the perpetual increase of the database is not a worthwhile or prac-
tical goal for further research, unless it can be implemented in a more dynamic and interactive form, such
as through a website. Additions from bridges from certain under-represented countries such as Japan and
Seventy-two concepts for arches using ultra high-performance fibre-reinforced concrete were designed.
Concepts differ by the three main parameters of the design study: span length (50 m to 400 m), span-to-
rise ratio (4:1 to 12:1), and degree of deck-stiffening (full, partial, or none). Concepts are designed primarily
for longitudinal axial force and bending demands caused by dead load, live load, creep, shrinkage, and
uniform changes in temperature, calculated according to the simplified methods described in Chapter 4.
The adequacy of given trial design were evaluated by comparing sectional force demands with member ca-
pacity N-M* interaction diagrams of the arch rib and deck girder, calculated using the general method de-
scribed in Section 3.2. If given designs were found to be unsatisfactory, slab thicknesses and section depths
were systematically increased until capacities exceeded demands. Sample calculations for Concept 80 are
The most common geometrical properties that were found to be satisfactory among concepts are summar-
ized in Table 6-29. Concepts whose demands required proportions greater than those shown in the table
include: (1) some concepts with spans of 50 m due to the low amount of natural prestress in the arch, and
(2) most concepts with span-to-rise ratios of 12:1 due to the high sectional force demands and high sensit-
282
7. Conclusions and Summary
Reprint of Table 6-29. Most common geometrical quantities used among design concepts
partially deck-stiffened arch deck hollow box 90:1, 100:1 100-200 mm 1/2
arch hollow box 90:1, 100:1 100-200 mm 1/2
Wall thicknesses as small as 50 mm are specified. Such thinness is made possible by: (1) the elimination of
conventional concrete cover requirements to depths of reinforcing, (2) the use of short 10 mm steel fibres,
and (3) the use of longitudinal stiffeners to prevent local stability problems. Based on a simple analytical
study on plate buckling, a maximum wall slenderness ratio of 24.5 was determined for ultra high-perform-
ance fibre-reinforced concrete. This upper limit ensures that thin concrete slabs are able to attain full com-
pressive strength without prematurely buckling. This corresponds to a maximum stiffener spacing of 1225
The need for both transverse ribs and longitudinal stiffeners poses a challenge for forming and casting the
hollow box members. A new segmental precasting method is proposed in Section 6.2.3. This involves cast-
ing the 100 mm end ribs separately from the main body of the segment. End ribs are formed with shear
keys and holes for prestressing against a precision bulkhead (see Figure below). Prior to casting, deformed
steel wire is installed along the edges that will eventually abut against the interior body of the segment.
Once the principal end rib is cast and cured, it is reversed and is used to form a similar match-cast end rib.
283
7. Conclusions and Summary
Partial reprint of Figure 6-16. Proposed match-casting method for abutting end ribs
The main body of the segment is cast using an exterior U-shaped base form and an interior core form.
After the end ribs are positioned 3 metres apart in the base form, with shear keys facing outward, the core
form is clamped to the two end ribs. The interior body of the segment including longitudinal stiffeners are
then cast. Once cured, the interior core form is released in parts and lifted out through the cavity of the
segment. The completed segment can then be lifted off the base form and transported to storage.
The proportions of the parametric design concepts were compared in terms of material efficiency, distribu-
tion of weight, slenderness, shallowness, and structural demands. A summary of all conclusions is given in
Section 6.3.10 on page 257. Some of the most important conclusions are discussed below.
284
7. Conclusions and Summary
Reprint of Table 6-15. Equivalent slab thicknesses based on volume of longitudinal concrete among all concepts.
Concepts with span-to-rise ratios of 4:1 and 8:1 consistently had less than 50% of the minimum equivalent
slab thicknesses of existing concrete arch bridges in the database (see Table above). This suggests that the
costs associated with constructing concrete arch bridge has the potential to be significantly reduced. Con-
cepts with span-to-rise ratios of 12:1 were not able to achieve the same reduction in concrete volume.
Hence, there are cost premiums associated with using arches with span-to-rise ratios in excess of 8:1.
Between self-stiffened, deck-stiffened, and partially deck-stiffened systems, deck-stiffened systems con-
sumed the least amount of concrete for spans of 50 m, while self-stiffened systems consumed the least
amount of concrete for spans between 150 m to 400 m. For spans of 100 m, all systems consumed approx-
imately the same amount of concrete. These results agree with the conclusions made in the modified arch
stress analysis presented in Section 4.7. It was concluded that self-stiffened arch systems are most efficient
among systems because they: (1) maximize the use of increased flexural capacity provided by the natural
precompression in the arch, and (2) isolate deck girders from the the global system, which minimizes their
bending moment demands. For the shorter 50 m span, deck-stiffened arch systems were found to be the
285
7. Conclusions and Summary
most efficient because the natural prestress of the arch caused by dead load was low in comparison to the
Most concepts tended to have span-to-system-depth ratios between 39:1 and 50:1, which is on par with the
ratios found among existing concrete arch bridges in the database. Thus the degree of visual slenderness
and transparency that can be achieved by the arch concepts are about the same as in conventional concrete
arches. The savings of concrete are thus related to the thinning of webs and slabs rather than the reduction
of overall depth.
In terms of sensitivity to second-order effects, most concepts had system slenderness ratios between 39 and
60. In this range, additional deflections and moments caused by second-order effects can be significant.
These systems are, however, are not slender enough that ultimate loads Q* are reached before sectional
failure is reached. This means that capacities of the member (N-M* interaction diagram) are only slightly
In terms of arch shallowness, concepts with span-to-rise ratios of 4:1 and 8:1 tended to have modified slen-
derness ratios much greater than the efficiency threshold value of 20. Their responses to load are thus
caused primarily by arch action. Concepts with span-to-rise ratios of 12:1 tended to have modified slender-
ness ratios around 20 or less. These concepts are thus affected by significant amounts of bending moment
under uniform load and are sensitive to axial deformations and crown deflections.
The scope of this thesis has been limited to the design of concrete arch bridges for axial forces and bending
moments caused by dead load, live load, creep, shrinkage, and temperature. For conventional concrete
arches, these loads and sectional forces tend to govern over all other design considerations. Although this
is likely true for the concepts proposed in this thesis, there remains a number of design checks and details
that have to be investigated. Design checks for wind, seismic, and construction loads need to be done, as
well as for unbalanced lane loading. The transverse design of the structural system, including the trans-
286
7. Conclusions and Summary
The use of thin 50 mm webs has been validated with a rudimentary shear design check for Concept 80.
Since this concept has one of the longer spans and higher span-to-rise ratios, it is anticipated that the thin
webs specified in most of the parametric design concepts are adequate. The use of thin, ribbed slabs in the
proposed concepts requires further validation with respect to the transmission of local wheel loads to the
primary structural elements of the system. Analytical studies by Spasojević (2008) have shown that thin
post-tensioned ribbed ultra high-performance fibre-reinforced concrete slabs with flange and web thick-
nesses of 40-60 mm were capable of transmitting truck wheel loads over spans of 12 m between supporting
girders. For the concepts proposed in this thesis, similar thin slabs would only have to span 3 m or less
between end ribs of the segment. Much deeper rib depths of 400 mm to 500 mm are specified, which
serves to increase capacity relative to Spasojević’s ribbed deck designs. Based on these reference comparis-
ons, it is anticipated that the proposed thin slabs are satisfactory, or at least satisfactory with very minor
Other topics for the future development of these systems could include: (1) details and truss models related
to the spread of forces at the anchorage zones of the main prestressing tendons need to be developed, (2)
the transfer of shear forces across the match-cast end ribs of abutting segments, and (3) the transfer of
shear and tension across the spliced joint between the end ribs and interior body of the segment.
Based on the findings in this thesis, ultra high-performance fibre-reinforced concrete is considered to be
an excellent material for arch bridges. Its enhanced properties of increased compressive and tensile
strength, reduced creep and shrinkage strains, and short fibre lengths, all serve to increase member capa-
cities and reduce the dead load demand of the structure. The resulting systems for concrete bridges are
lighter, more efficient, and more economical to build than arches made using convectional concrete
technology.
287
References
Association française du génie civil (AFGC). 2002. Ultra High Performance Fibre-Reinforced Concretes: Interim
Recommendations. France.
Aitcin, P. C., M. Lachemi, R. Adeline, and P. Richard. 1998. The Sherbrooke reactive powder concrete footbridge.
Austin, W. J. 1971. In-plane bending and buckling of arches. ASCE Journal of the Structural Division. 97(5): 1575-1592.
Bazant, Z. P. 1972. Prediction of concrete creep effects using age-adjusted effective modulus method. ACI Materials
Benaim, R. 2007. The Design of Prestressed Concrete Bridges : Concepts and Principles. London: Taylor & Francis.
Bill, M., and R. Maillart. 1955. Robert Maillart. Translated by W. P. M. K. Clay. Zürich, Switzerland: Girsberger.
Billington, D. P. 1973. Deck-stiffened arch bridges of Robert Maillart. Journal of the Structural Division. 1527-1539.
Billington, D. P. 1979. Robert Maillart's Bridges: The Art of Engineering. Princeton: Princeton University Press.
Billington, D. P.. 2003. The Art of Structural Design: A Swiss Legacy. Princeton, New Jersey: Princeton University Art
Museum.
Bradford, M. A., T. Wang, Y. L. Pi, and R. I. Gilbert. 2007. In-plane stability of parabolic arches with horizontal spring
Cai, J. G., J. Feng, Y. Chen, and L. F. Huang. 2009. In-plane elastic stability of fixed parabolic shallow arches. Science
Cai, J., and J. Feng. 2009. Erratum to “In-plane elastic stability of fixed parabolic shallow arches.” Science in China
Candrlic, V., J. Bleiziffer, and A. Mandic. 2001. Bakar Bridge designed in reactive powder concrete. Arch'01: troisième
conférence internationale sur les ponts en arc (Third international arch bridges conference). Paris, France.
Culmann, C., and W. Ritter. 1906. Der Bogen (Arches). ETH Zurich.
288
Chen, B. 2007. An overview of concrete and cfst arch bridges in China. Arch'07: 5th International Conference on
Cross, H. 1932. Analysis of continuous frames by distributing fixed-end moments. Transactions of the American
Denarié, E., E. Brühwiler, A. Znidaric, Y. Houst, and R. Rohleder. 2005. Full scale application of Uhpfrc for the
rehabilitation of bridges: From the lab to the field. Sustainable and Advanced Materials for Road Infrastructure,
Dickie, J. F., and P. Broughton. 1971. Stability criteria for shallow arches. ASCE Journal of the Engineering Mechanics
El-Arousy, A. 1942. Studien über das elastische Verhalten von Brückengewöolben einschliesslich des
Zusammenwirkens mit dem Aufbau. Dissertation. Institut für Baustatik, ETH Zurich.
Ewert, S. 1999. Betonbogenbrücken: Mehr als 200 m Spannweite. Beton- und Stahlbetonbau. 94(9): 377-388.
Favre, R., and J. de Castro San Román. 2001. The arch: enduring and endearing. Structural Concrete. 2(4): 187-200.
Fonseca, A., and F. Mato. 2005. Infant Henrique Bridge over the River Douro, Porto. Structural Engineering
Freytag, B., M. Reichel, L. Sparowitz, G. Santner, and G. Heinzle. 2009. Groβversuch WILD-Brücke: versuchsgestützte
Gaber, E. 1934. Über die Knicksicherheit vollwandiger Bogen (On the bucking strength of solid arches). Bautechnik.
646-651.
Galambos, T. V. 1998. Guide to Stability Design Criteria for Metal Structures. John Wiley.
Gjelsvik, A., and S. R. Bodner. 1962. The energy criterion and snap buckling of arches. ASCE Journal of the
289
Graybeal, B. A. 2006. Material property characterization of Uhpc. No. Fhwa-Hrt-06-103. US Department of
Graybeal, B. A., and J. L. Hartmann. 2005. Construction of an optimized Uhpc vehicle bridge. ACI Special
Gruber, L., and C. Menn. 1978. Berechnung und Bemessung schlanker Stahlbetonstützen (Analysis and design of
slender reinforced concrete columns). Report No. 84. Institut für Baustatik und Konstruktion, ETH Zürich,
Zürich, Switzerland.
Habel, K. 2004. Structural behaviour of elements combining ultra-high performance fibre reinforced concretes and
reinforced concrete. Doctoral dissertation. Génie Civil, École Polytechnique Fédérale de Lausanne, Switzerland.
Habel, K., J. P. Charron, S. Braike, R. D. Hooton, P. Gauvreau, and B. Massicotte. 2008. Ultra-high performance fibre
reinforced concrete mix design in central Canada. Canadian Journal of Civil Engineering. 35(2): 217-224.
Habel, K., and P. Gauvreau. 2008. Response of ultra-high performance fiber reinforced concrete (UHPFRC) to impact
Huh, S. B., and Y. J. Byun. 2005. Sun-Yu Pedestrian Arch Bridge, Seoul, Korea. Structural Engineering International.
15(1): 32-35.
Hünleim, W., and P. Ruse. 1985. Ein neues Verfahren für den Bau von Bogenbrücken dargestelit am Bau der
Johnston, B. G. 1976. Guide to Stability Design Criteria for Metal Structures. John Wiley & Sons.
Kennedy, J. S., and A. S. Aggarwal. 1971. Effect of weight on large deflections of arches. ASCE Journal of the
Ketchum, M. S. 1934. Thin-section concrete arches as built in Switzerland. Engineering News Record. Jan. 11: 44-45.
Kollbrunner, C. F. 1936. Versuche über die Knicksicherheit und die Grundschwingungszahl vollwandiger Bogen
(Experiments on the buckling strength and fundamental frequency of solid arches). Bautechnik. March:
186-188.
290
Kremer, I. 2008. Hydraulic crown jacking procedure shapes the concrete arch of the Third Millenium Bridge.
Laffranchi, M., and P. Marti. 1997. Robert Maillart's curved concrete arch bridges. Journal of Structural Engineering.
123(10): 1280-1286
Lin, Y., Z. Zhang, B. Ma, and L. Zhou. 2004. Lupu Arch Bridge, Shanghai. Structural Engineering International. 14(1):
24-26.
Ma, J., and M. Orgass. 2004. Comparative investigations on ultra-high performance concrete with and without coarse
Maillart, R. 1934. Gekrümmte Eisenbeton-Bogenbrücken (Curved reinforced concrete arch bridges). Schweizerische
Melan, J., and D. B. Steinman. 1913. Theory of Arches and Suspension Bridges. Translated by D. B. Steinman. Chicago:
Menn, C. 1990. Prestressed Concrete Bridges. Translated and edited by P. Gauvreau. Basel: Birkhäuser Verlag.
Nathan, N. D. 1985. Rational analysis and design of prestressed concrete beam columns and wall panels. Prestressed
Ordóñez, F., and J. Antonio. 1979. Eugène Freyssinet. Barcelona: Coop. Industrial Trabajo Association.
Ostrander, J. R., and D. C. Oliver. 1987. Construction of the Broadway Bridge at Saskatoon, Saskatchewan, in 1932.
Radić, J., Z. Šavor, and I. Gukov. 2005. New contribution to concrete arch bridge construction. Transportation
Rambøll, B. J. 1944. Om buers stabilitet (About arch stability). Bygningsstatiske Meddelelser (2).
Ritter, W. 1877. Versteifungsfachwerke bei Bogen- und Hängebrücken (Stiffening trusses in arch and suspension
291
Šavor, Z., and J. Bleiziffer. 2008. Long span concrete arch bridges of Europe. Chinese-Croatian Joint Colloguium Long
Schreyer, H. L., and E. F. Masur. 1966. Buckling of shallow arches. ASCE Journal of the Engineering Mechanics
Seegers, K. H. 1963. Beton-Bogenbrücke über den Parametta-Fluβ in Sydney. der Bauingenieur 38(12): 477-478.
Spasojević, A. 2008. Structural implications of ultra-high performance fibre-reinforced concrete in bridge design.
Spiegel, M. R., and J. Liu. 1999. Mathematical Handbook of Formulas and Tables, Second Edition. Toronto: Schaum's
Outline Series.
Šram, S. 1982. Ausführung der Brücken vom jugoslawischen Festland zu den Adriainseln St. Marko und Krk. Beton-
Stellmann, W. L. O. 1966. Brücke über den Rio Paraná in Foz Do Iguaçú, Brasilien. Beton- und Stahlbetonbau. 61(6):
145-149.
Stussi, F. 1935. Aktuelle baustatische Probleme der Konstruktionspraxis (Current problems of construction design
Swartz, S. E., and V. H. Rosebraugh. 1974. Buckling of reinforced concrete plates. ASCE Journal of the Structural
Tanner, P., and J. L. Bellod. 2005. Widening of the Elche De La Sierra Arch Bridge, Spain. Structural Engineering
Taylor, A. W., R. B. Rowell, and J. E. Breen. 1995. Behaviour of thin-walled box piers. ACI Structural Journal. 92(3):
319-333.
Timoshenko, S. 1936. Theory of Elastic Stability. New York: McGraw-Hill Book Company, Inc.
Trost, H. 1967. Auswirkungen des Superpositionsprinzips auf Kriech- und Relaxationsprobleme bei Beton und
Spannbeton (Implications of the superposition principle in creep and relaxation problems for concrete and
292
US Department of Transportation Federal Highway Administration (US DOT FHWA). 2003. Az-Nv Hpp 93(3): Hoover
Vianello. 1898. Graphische Untersuchung der Knickfestigkeit gerader Stäbe. Zeitschrift des Vereins Deutscher
Ingenieure. 1436-1443.
Vogel, T, and P. Marti. 1997. Christian Menn, Brückenbauer. Basel, Switzerland: Birkhäuser Verlag.
Walraven, J. 2008. On the way to design recommendations for Uhpfrc. Second International Symposium on Ultra
Wang, T., M. A. Bradford, and R. I. Gilbert. 2006. Creep buckling of shallow parabolic concrete arches. Journal of
Yan, G., and Z. H. Yang. 1997. Wanxian Yangtze Bridge, China. Structural Engineering International. 7(3): 164-166.
293
Appendices
294
Appendix A: Integrated Database of Concrete Arch Bridges
This appendix contains the original integrated database of existing concrete arch bridges from which most
figures in Chapter 5 have been adapted from. The complete database is presented on three 11” by 17” sheets.
The first sheet shows the elevation and section views of each bridge, a world map with the location of each
bridge, and a list of references. Each bridge is assigned a bridge id, in order of span length, with id 1 being
the longest spanning bridge in the database. The second sheet tabulates the section properties, geometrical
ratios, structural type, country, type of construction, and year of completion of each bridge. This quantitat-
ive data is presented in graphical form in the second and third sheets. Most plots are accompanied by an-
notations that propose trends that can be observed among bridges in the database.
This database can be used as references and design tools for designers. It allows for the evaluation of new
bridge concepts in terms of structural efficiency and feasibility, relative to a large set of concrete arch
bridges. The integrated form of this database allows for further study of the relationships between data
sets, from which the reader can draw personal insights that are specific to the span range or arch types that
he or she may be interested in. For further reference, readers can consult the primary sources listed on the
first sheet, which contain further information on each bridge in the database.
295
Location of arch bridges
Bridge drawings bridge ID
A
Elevations at 1:10000 scale name
Sections at 1:1000 scale year of completion
ÇÇ Ç
Ç Ç
Ç Ç Ç
Ç ÇÇ
A
Ç Ç ÇÇÇ ÇÇÇ Ç
ÇÇÇ Ç ÇÇ ÇÇ ÇÇÇ
typical section
Ç Ç Ç Ç
ÇÇ
column section
Ç B
Ç
Ç ÇÇ Ç
Ç
Ç
Ç B
ÇÇ
Ç Ç Ç
Ç
Taiwan Europe
References
Organized by bridge ID
1. Dajun, D., and L. Weiqing. 1999. Neue Entwicklungen Bei Hochhäusern Und Gro
en Brücken Aus Beton in China. Beton- und Stahlbetonbau 94(4): 178-85.
2/11.Radi
, J., J. Bleiziffer, and D. Tkali
. 2005. Maintaining Safety and Serviceability of Concrete Bridges in Croatia. Bridge Structures 1(3): 327-44.
3. US Dept. of Transportation FHWA. 2003. AZ-NV HPP 93(3) Hoover Dam Bypass Colorado River Bridge: 60% Submittal. April 4.
4. Baxter, J. W., A. F. Gee, and H. B. James. 1965. Gladesville Bridge. Proceedings of the Institute of Civil Engineers.
5. Stellmann, W. L. O. 1966. Brücke Über Den Rio Paraná in Foz Do Iguaçú, Brasilien. Beton- und Stahlbetonbau 61(6): 145-49.
6. Fonseca, A., and F. Mato. 2005. Infant Henrique Bridge Over the River Douro, Porto. Struct. Eng. Int. 15(2): 85-87.
7. Liebenberg, A.C., and M.G. Latimer. 2001. Bloukrans Bridge. Arch'01: troisième conférence internationale sur les ponts en arc, Sept. 19-21, Paris.
8. @7BH3::
3X31B7=<!3/AC@3;3<BA=<B63'/<2Q@7253'7FB6=<5@3AA=4B63'
9. Kamimura, K., M. Kouno, K. Okada, and T. Tsuka. 2001. Design and Construction of the Tensho Concrete Arch Bridge. Arch'01: Third Int'l Conf. on Arch Bridges, Paris.
10. Radi
, J., Z. avor, J. Bleiziffer, and I. Kalafati
. 2008. ibenik Bridge - Design, Construction and Assessment of Present Condition. Chinese-Croatian Joint Colloquium:
Long Arch Bridges, July 10-14, Brijuni Islands, China.
12. D=<+Q:43:&73(/:0@S193J03@73+7:233@/Beton- und Stahlbetonbau 94: 546-51.
13. State of Nevada Department of Transportation. 2000. B-1948-N/S Galena Creek Bridge. April 19.
14. avor, Z., N. Mujkanovi
, and G. Hrelja. 2008. Design and Construction of Krka River Arch Bridge. Chinese-Croatian Joint Colloquium Long Arch Bridges, Brijuni Islands.
15. Radi
, J., Z. avor, V. Prpi
, M. Friedl, and . deric. 2008. Design and Construction of the Maslenica Highway Bridge. Chinese-Croatian Joint Colloquium, Brijuni Islands.
16. avor, Z., N. Mujkanovi
, G. Hrelja, and J. Bleiziffer. 2008. Reconstruction of the Pag Bridge. Chinese-Croatian Joint Colloquium: Long Arch Bridges, Brijuni Islands.
17. Wang, J. J., and F. T. K. Au. 2001. Modong Hongshui River Bridge, China. Structural Engineering International 11(2): 101-03.
18. Aigner, F. 1968. Stahlbeton Bogenbrücken Auf Der Österreichischen Brenner Autobahn. Der Bauingenieur.
19. Igarashi. 1976. Die Hokowazu-Brücke in Japan, Ein Stahlbetonbogen Im Freien Vorbau. Der Bauingenieur (8).
20. Anonymous. 1978. Communications Techniques Belges 8. Annales des Travaux Publics de Belgique, No. 1-2.
21. Cheng, K. M. 1994. The Pitan Bridge, Taiwan. Structural Engineering International 4(4): 231-34.
22. Standfu
, F. 1990. Nationalbericht: Brücken in Der Bundesrepublik Deutschland. Beton- und Stahlbetonbau (5): 106-13.
23. +Q
<3@306/@2B&'16</03:/<2+Q@<3@73(/:0@S193&=BBE37:"319/@0C@5;-C533@Beton- und Stahlbetonbau (10-11).
24. Bouchet. 1964. L'Echangeur La Araña Du Croisement Des Autoroutes Nord-Sud Et Est-Ouest À Caracas (Venezuela). La Technique des Travaux (Belgique).
25. Hünleim, W., and P. Ruse. 1985. Ein Neues Verfahren Für Den Bau Von Bogenbrücken Dargestelit Am Bau Der Argentobelbrücke Würgau. Der Bauingenieur (12).
26. deric, ., A. Runji
, and G. Hrelja. 2008. Design and Construction of Cetina Arch Bridge. Chinese-Croatian Joint Colloquium on Long Arch Bridges, Brijuni Islands.
27. Koppel, A. J., and R. Walser. 1991. Hundwilertobelbrucke: Ein Bemerkenswerter Neubau. Schweizer Ingenieur und Architekt 109(11): 250-56.
28. Taiwan Area National Expressway Engineering Bureau, MOTC. 1999. (In Mandarin) The Second Freeway: An Anthology of Particular Bridges.
29. BC Ministry of Transportation and Highways. 1995. R6-V433-3051 Big Qualicum River Upstream Bridge No. 3051 (Drawings).
30. Goodyear, D., N. Smit, J. Heacock, and S. Starkey. 2001. Respect for Tradition: The New Crooked River Gorge Bridge. Arch'01: Third Int'l Conf. on Arch Bridges, Paris.
31. Podolny, W., and J. M. Muller. 1982. Construction and Design of Prestressed Concrete Segmental Bridges. New York: Wiley.
32. Inaudi, D., A. Rufenacht, B. Arx, H.P. Noher, and S. 2002. Monitoring of a Concrete Arch Bridge During Construction. Proc. SPIE International Society for Optical Eng.
33. Vogel, T, and P. Marti. 1997. Christian Menn, Brückenbauer. Basel, Switzerland: Birkhäuser Verlag.
34. State of California Department of Public Works Division of Highways. 1932. Bridge Across Bixby Creek (as Built Plans) Doc. No. 50001001.
35. Vogel, T, and P. Marti. 1997. Christian Menn, Brückenbauer. Basel, Switzerland: Birkhäuser Verlag.
36. Eilzer, W., W. Schmidtmann, and R. Jung. 2005. Die Wirrbachtalbrücke. Beton- und Stahlbetonbau 100(3): 236-40.
37/55/57/58. Bill, M., and R. Maillart. 1955. Robert Maillart. Translated by W. P. M. K. Clay. Zürich, Switzerland: Girsberger.
38/50/56. Sinotech Engineering Consultants, Ltd. Personal correspondence, August 2007.
39. !Q@A16
73;S<23@B=03:@S19337(3C43<;/<B=<>>3<H37:Schweizerische Bauzeitung 53 (7): 81.
40. Schwaab, E., and A. Gattner. 1953. Stra
enbrücke Über Das Tiefe Tal Bei Rosshaupten, Allgau. Beton- und Stahlbetonbau (11).
41. Tonello, J., and P. Dal-Palu. 2001. Le Pont Du Triple Saut (the Triple Jump Bridge). Arch'01: troisième conférence internationale sur les ponts en arc, Sept. 19-21, Paris.
42. 'B/B3=4/:74=@<7/3>/@B;3<B=4$C0:71+=@9A7D7A7=<=4756E/GA@72531@=AA&CAA7/<C:16/@B6?C/93&3B@=WB$@=831B"=
43. Petrangeli, M.P. 2007. Prefabrication of Medium Span Arch Bridges. Arch'07: 5th Intl Conference on Arch Bridges, Sept. 12-14, at Madeira, Portugal.
44. Zimmermann, W., and N. im Gailtal. 1999. Der Bau Der Stampfgrabenbrücke. Beton- und Stahlbetonbau 99(4): 304-10.
45. State of California Department of Public Works Division of Highways. 1932. Bridge Across Rocky Creek (as Built Plans) Doc. No. 50001081.
46. Baxter, D. J., and T. A. Balan. 2008. Design of the Fulton Road Bridge Precast Segmental Concrete Arches. Journal of Bridge Engineering: 476-82.
47. Blinkov, L. S., E. Cosolo, and S. N. Valiev. 2001. Rehabilitation of a Bridge Over the Matsesta River, Russian Fed. Structural Engineering International 11(3): 181-83.
48. Ostrander, J. R., and D. C. Oliver. 1987. Construction of the Broadway Bridge At Saskatoon, Saskatchewan, in 1932. Canadian Journal of Civil Engineering 14: 429-38.
49. State of California Department of Public Works Division of Highways. 1937. Bridge Across Big Creek (as Built Plans) Doc. No, 50001093.
51. *73B(C3"36<('16:73;/<<&37<B83A/<2(/CA163@
<E3<2C<5*=<=16:37ABC<5A03B=<3<373@=53<0@S193+Q:9/C;-C533@/0
Beton- und Stahlbetonbau 100(11): 931-38.
52. Tanner, P., and J. L. Bellod. 2005. Widening of the Elche De La Sierra Arch Bridge, Spain. Structural Engineering International (3): 148-50.
53. Tandon, M. 1995. Arch Bridge At Dodan Nallah. Proceedings of the First International Conference on Arch Bridges, Bolton, UK.
bridge ID 51 41 31 21 11 1
volume = deck area span eccentricity of live load : half arch depth
+ arch area arch length horizontal reaction / (unit weight deck width)
+ column area total spandrel column length max arch depth in m 8
Variable definitions UVYTHSPaLKZ`Z[LTTVTLU[VMPULY[PH
modified system moment of inertia average arch depth in m
interior deck width system inertia = deck inertia + arch inertia equiv. slab thickness by volume min arch depth in m
span spandrel
0 if deck is discontinuous column inertia : system inertia
column 7 3
deck depth arch inertia : system inertia
column PRGLÀHGVOHQGHUQHVVUDWLR 2 rise 3
rise section system radius of gyration mod. slenderness ratio
arc length 6
slenderness ratio
slenderness ratio = effective arc length factor arc length span : arch + deck depth
span system radius of gyration span : arch depth
arch depth 5
system radius of gyration = system inertia system radius of gyr. 3
Bridge geometry and ratios arch area span2 : rise 2
means value not DUFKÀ[LW\ span : rise 4 3 2 3 3 3
applicable for comparison interior spans
? means value unknown construction method deck section arch section column section
3
span rise type width depth area inertia cont. type depth var. area inertia area inertia 3
2
2 2 2
ID bridge year country m m # m m m2 10-3 m4 m m m2 10-3 m4 m2 10-3 m4 m m % % m m m2 % 2
4 2 3
2
1 Wanxian-Yangtze 1997 China 425.0 85.0 14 T 21.0 ? ? ? -- 3 7.00 27.9 220000 5.4 4400 5.0 2130 2.80 61 ? 58 61 100 2 ? 5.01 ? ?
2 Krk I 1980 Croatia 390.0 60.0 14 T 10.8 2.00 4.8 1800 -- 3 6.50 21.5 130000 1.7 300 6.5 2540 2.46 60 46 59 49 100 <1 2.61 5.25 828 6 2
2
3 Colorado 2010 United States 323.0 84.4 9 26.9 2.50 8200 — 4.27 14.8 40000 6.9 2900 3.8 1240 1.80 76 48 73 94 83 6 2.78
4 Gladesville 1964 Austrailia 309.0 39.0 s 10 T 26.4 1.80 9.8 2900 — 5.60 ± 1.34 25.1 100000 3.0 63 7.9 2450 2.02 55 42 56 39 97 <1 1.37 3.60 421 8 T
T v
5 Rio Parana 1965 Brazil 290.0 53.0 s 13 T 13.5 2.20 10.5 2100 — 3 4.00 ± 0.80 17.7 44000 0.6 92 5.5 1590 1.61 73 47 68 66 95 <1 2.22 3.45 440 12
rectangular solid section T ?
6 Infant Henrique 2003 Portugal 280.0 25.0 8 20.2 4.50 16.8 57000 — v 1.50 18.1 4400 17.4 2500 11.2 3140 1.84 187 47 54 27 7 4 1.83 3.32 716 1 v voided slab section 1
7 Bloukrans 1983 South Africa 272.0 62.0 15 v 15.7 1.50 10.8 2300 — 3 4.60 ± 1.00 17.3 53000 4.7 2500 4.4 1190 1.79 59 45 60 69 96 5 2.25 3.48 335 12 T tee or slab-on-girder section
8 Sandö 1943 Sweden 264.0 40.0 t 22 T 12.4 1.23 5.7 540 — 3 3.95 ± 0.95 13.0 31000 0.9 31 6.6 1740 1.56 67 51 63 51 98 <1 1.62 3.12 352 15 n single or n-cell box-girder section
9 Tensho 2000 Japan 260.0 32.5 14 v 8.6 0.80 4.4 280 — 5.00 ± 1.00 17.0 54000 8.1 1500 8.0 2080 1.79 52 45 53 36 99 3 2.95 4.23 768 6
0
10 Sibenik 1966 Croatia 246.0 31.0 11 T 11.1 1.50 2.6 680 -- 3 3.30 ± 0.40 7.7 14000 0.6 3 7.9 1950 1.35 75 51 66 46 100 <1 0.98 2.48 239 27
11 Krk II 1980 Croatia 244.0 47.0 8 T 10.8 2.00 5.4 1700 -- 3 4.00 9.0 21000 1.7 300 5.2 1270 1.53 61 41 61 62 100 1 1.46 2.86 232 17 deck depth in m 6
12 Wilde Gera 2000 Germany 242.0 68.0 6 26.3 3.74 ? — 2 4.40 ± 1.10 11.3 34000 7.6 11000 3.6 860 ? 55 30 ? ? ? ? ?
13 Galena Creek 2011 United States 210.0 42.5 s 3 2 18.9 3.00 12.1 14000 — 3.70 6.1 13000 4.9 1040 2.10 57 31 38 40 48 1.00 2.58 129
14 Krka-Skradin 2004 Croatia 204.0 52.0 7 22.3 1.96 3700 — 2 3.00 10.9 15000 4.1 1700 3.9 800 1.31 68 41 63 79 80 9 2.16
5
15 Maslenica 1997 Croatia 200.0 65.0 7 T 21.2 2.00 13.3 4800 — 2 4.00 12.6 29000 5.2 3900 3.1 620 1.64 50 33 53 79 86 12 1.53 2.67 118 17
16 Pag 1968 Croatia 193.0 28.0 9 T 9.0 1.42 1.9 350 -- 3 2.65 ± 0.35 4.0 4500 1.4 77 6.9 1330 1.06 73 47 67 53 100 2 0.72 1.82 121 45
17 Modong Hongshui 1999 China 180.0 30.0 14 11.9 0.65 8.1 290 -- 3.50 5.7 9600 1.3 35 6.0 1080 1.30 51 43 52 46 100 <1 1.24 2.13 168 23
18 Nosslach 1968 Austria 180.0 45.0 s 12 ? ? ? ? ? — 3 4.00 ± 1.30 12.5 26000 ? ? 4.0 720 ? 45 ? ? ? ? ? ? ? ? ? 4
19 Hokowazu 1974 Japan 170.0 26.5 11 v 10.1 0.92 4.4 320 — 2 3.00 10.4 13000 3.0 560 6.4 1090 1.13 57 43 80 47 98 4 1.67 2.51 228
20 Houffalize 1979 Belgium 162.0 32.4 s 6 T 16.0 2.00 6.1 3100 -- 2 3.20 13.0 19000 4.5 1900 5.0 810 1.21 51 31 52 54 100 10 1.36 2.42 138 23
21 Pitan 1993 Taiwan 160.0 21.7 s 3 15.9 5.00 16.1 59000 — 6.25 17.5 97000 7.4 1180 2.99 26 14 20 15 62 2.17 4.90 319 3 2
22 Wertachtal ? Germany 156.0 34.0 11 T 12.7 1.42 6.7 1100 — 2 2.80 ± 0.80 8.5 9900 2.1 120 4.6 720 1.14 56 37 54 60 90 1 1.40 2.18 126 27
23 Neckarburg 1978 Germany 154.0 49.5 7 15.5 2.30 9.2 6300 — 2 3.00 7.2 9200 3.8 4800 3.1 480 1.47 51 29 45 67 59 31 1.41 2.29 84 21
24 Caracas 1953 Venezuela 151.8 32.0 12 T 20.8 2.00 5.4 1800 -- 3.10 9.4 13000 3.0 110 4.7 720 1.18 49 30 72 54 100 1 0.82 1.96 74 37 T
2 T T T T
25 Argentobel 1986 Germany 150.0 29.0 10 T 14.0 1.60 7.7 1500 — 2 2.75 ± 0.75 7.3 8520 2.4 200 5.2 780 1.17 55 34 49 50 85 2 1.22 2.05 118 28 T T
T 4 4
T
26 Cetina 2007 Croatia 140.3 20.6 7 T 10.2 1.40 5.4 870 — 2.50 8.0 7500 3.2 1200 6.8 960 1.02 56 36 51 40 90 14 1.44 2.14 172 19 T 2
T T v
27 Hundwiler 1991 Switzerland 138.4 35.8 13 T 9.6 0.82 4.3 200 — 2.15 ± 0.65 5.6 2200 1.2 36 3.9 540 0.65 64 47 86 109 92 2 1.23 1.44 82 46 rectangular solid section T ? T T T
v voided slab section T T
28 Maling 1999 Taiwan 138.0 33.0 7 v 16.0 1.10 7.6 1200 — 4 2.60 ± 0.60 20.2 16000 5.9 500 4.2 580 0.92 53 37 59 72 93 3 2.04 2.35 147 20 T TT v
T tee or slab-on-girder section 1 v
29 Big Qualicum 1996 Canada 132.0 24.0 s 6 T 24.0 1.73 7.2 2200 — 1.50 9.0 1700 7.2 860 5.5 730 0.66 88 41 76 73 44 22 0.81 1.25 73 26 T T T T
n single or n-cell box-girder section T T TT T T
v
T T
30 Crooked River 2000 United States 125.0 24.0 8 4 24.2 1.90 12.5 6400 — T 1.63 ± 0.38 11.0 2500 5.9 910 5.2 650 0.90 77 35 53 53 28 10 1.09 1.64 89 14 composite steel-concrete section
T v
31 Niesenback ? Germany 120.0 37.5 6 T 17.6 2.10 7.6 2300 — 2 2.50 8.9 8100 2.0 170 3.2 380 1.08 48 26 47 69 78 2 1.10 1.92 53 40
0
32 Siggenthal 2000 Switzerland 117.0 24.4 s 7 ? ? 1.40 ? ? — ? 1.20 ± 0.20 ? ? ? ? 4.8 560 ? 98 45 ? ? ? ? ? ? ? ?
33 Nanin 1967 Switzerland 112.0 24.5 t 8 9.9 1.00 4.1 400 — 1.20 ± 0.20 5.1 600 2.5 100 4.6 510 0.44 93 51 99 111 60 10 1.05 1.07 67 54
34 Bixby Creek 1932 United States 100.6 36.6 11 T 8.4 0.92 3.4 200 -- 2.00 ± 0.50 5.6 1900 1.5 120 2.8 280 0.58 50 34 78 126 100 6 1.46 1.39 50 67 year of completion 2020
35 Rhein 1963 Switzerland 100.0 20.9 ? 8 8.5 1.00 2.8 350 — 1.50 ± 0.25 6.1 1100 2.6 110 4.8 480 0.49 67 40 79 86 76 8 1.20 1.27 72 53
36 Wirrbachtal 2002 Germany 100.0 25.3 s 7 T 11.5 1.40 7.6 1300 — 1.10 ± 0.20 8.0 810 1.0 40 4.0 400 0.51 91 40 78 99 38 2 1.50 1.30 74 34 2010 s
37 Salginatobel 1930 Switzerland 90.0 13.0 t T 3.5 0.90 1.7 75 — 0.8 22 6.9 620 s
38 Wunshuei ? Taiwan 85.0 13.5 ? 14 v 11.0 0.45 4.3 81 — 2 2.00 7.8 4100 2.0 69 6.3 540 0.73 43 35 43 37 98 2 1.25 1.66 84 40 2000 s
s s
39 Gmundertobel 1909 Switzerland 80.0 26.0 t 18 T 6.9 0.70 1.9 60 — 1.65 ± 0.45 11.9 2700 1.2 37 3.1 250 0.48 48 34 72 108 98 1 2.67 1.69 82 49 s
s
40 Tiefetal 1952 Germany 77.7 12.6 t 10 T 10.0 1.03 3.6 310 — 1.50 ± 0.30 4.8 1400 1.1 25 6.2 480 0.60 52 31 75 42 82 1 0.91 1.27 55 s
1990
41 Guiers 2000 France 73.8 12.5 s 9 9.8 0.75 10.2 480 — T 1.25 ± 0.20 4.0 570 1.7 170 5.9 440 0.51 59 37 54 49 54 16 1.56 1.09 85 25
42 Russian Gulch 1939 United States 73.2 25.9 ? 15 9.7 0.55 3.5 45 -- 1.65 ± 0.35 4.3 970 1.2 64 2.8 210 0.47 44 33 98 109 100 7 1.15 1.06 30 1980
s
43 Mazzocco 2007 Italy 70.0 14.0 s 3 4 17.2 1.85 7.6 2700 -- T 1.50 6.2 1300 5.0 350 0.46 47 21 59 61 100 0.84 0.97 37
44 Stampfgraben 2003 Austria 70.0 21.0 6 T 9.3 0.91 5.4 350 — 1.33 ± 0.28 6.0 880 1.3 26 3.3 230 0.45 53 31 65 93 72 2 1.42 1.17 41 59 1970
t s
45 Rocky Creek 1932 United States 68.6 17.1 t 11 T 8.2 1.07 4.2 210 -- 1.77 ± 0.55 4.3 1100 1.0 64 4.0 280 0.51 39 24 54 68 100 6 1.23 1.17 42 69 s s
46 Fulton Road 1932 United States 64.0 12.7 t 5 T 23.5 1.17 7.7 820 — 1.50 5.7 1300 5.0 650 5.0 320 0.61 43 24 40 42 61 31 0.65 1.03 26 50 ?
1960
47 Matsesta 1938 Russia 63.6 23.6 ? 12 T 18.5 0.83 3.7 220 — 1.11 ± 0.21 6.6 680 3.2 120 2.7 170 0.37 58 33 78 128 76 13 0.81 0.84 17 126
48 Broadway 1932 Canada 54.7 13.9 t 11 T 13.7 1.86 6.3 1600 — 1.20 ± 0.30 11.7 1400 2.0 120 3.9 220 0.51 46 18 44 55 47 4 1.57 1.38 42 36 t
1950
49 Big Creek 1937 United States 54.1 23.0 t 11 T 8.2 0.76 3.3 110 — 1.20 ± 0.30 2.2 260 1.1 78 2.4 130 0.41 45 28 90 112 70 21 0.96 0.82 15
50 Lianlao ? Taiwan 50.0 12.5 ? 8 T 20.5 0.60 6.2 120 — 1.00 18.0 1500 1.8 54 4.0 200 0.30 50 31 67 83 93 3 1.35 0.98 34 70 t timber scaffolding / centering t
s steel scaffolding / centering 1940 ?
51 Wölkau ? Germany 47.5 5.1 s 3 T 4.5 0.60 2.1 59 — 0.53 ± 0.13 1.3 30 9.3 440 0.26 90 42 93 39 34 0.76 0.62 42 cantilever using cable-stayed t?
?
52 Elche de la Sierra 1927 Spain 40.0 4.0 20 T 8.0 0.49 2.8 42 — 1.01 ± 0.07 2.0 172 0.5 3 10.0 400 0.33 40 27 44 24 80 1 0.64 0.68 32 85 cantilever using effective truss t t t
pre-erected reinforcement arch 1930 tt t
53 Dodan Nallah 1995 India 40.0 14.5 s 7 8.3 0.51 2.5 28 — 0.53 ± 0.18 2.3 53 1.7 23 2.8 110 0.19 76 39 96 155 65 28 0.79 0.49 11 310 rotation method
54 Bronte Creek 1934 Canada 39.8 17.1 ? 11 16.4 0.23 3.8 17 — 0.81 ± 0.05 4.3 241 2.2 32 2.3 90 0.24 49 38 78 139 93 12 0.78 0.57 9 252 t
1920
55 Schwandbach 1933 Switzerland 37.4 6.0 t 9 T 4.2 0.85 1.0 31 — 0.20 1.3 4 0.8 2 6.2 230 0.16 187 36 85 73 12 6 0.64 0.47 19 104
56 Nan Ke ? Taiwan 25.0 3.3 ? 3 T 16.0 0.80 9.6 500 — 0.45 4.1 70 7.7 190 0.37 56 20 24 17 12 0.87 0.75 21
57 Ziggenbach 1924 Switzerland 20.0 4.7 t 8 T 5.2 0.88 1.5 89 — 0.41 ± 0.15 2.0 29 0.2 1 4.3 90 0.24 49 16 33 39 25 <1 0.75 0.65 8 103
58 Bohlbach 1932 Switzerland 14.4 2.7 t 9 T 4.5 1.16 1.3 160 — 0.16 1.0 2 0.9 2 5.3 80 0.40 90 11 14 13 1 1 0.58 0.76 6 17
This appendix gives additional details on the functions used in the computer program qult. This program,
written by the author, models the behaviour of ultra high-performance fibre-reinforced concrete slender
members under combined compression and bending (see Section 3.3 for an overview of qult). qult is or-
ganized as a hierarchy of functions that implement the general analysis method described in Section 3.2.
qult is written in matlab r2007a and is included with this thesis digitally as a set of matlab m-files.
Table B-1 lists all the major functions used in qult and describes their input parameters, calculation steps,
and output data. Most functions have an option to plot results in matlab. These plots are activated by sett-
ing the variable option to 1 near the beginning of most m-files. These plots allow the user to browse and
check results at each calculation stage. Samples of these output plots are shown in the bottom half of Fig-
299
B. Details of Computer Program QULT
sectional
member
geometry & sectional member
response
material analysis deflection
& capacity
response
input parameters stress-strain curve stress-strain curve stress-strain curve stress-strain curve
sectional geometry sectional geometry sectional geometry sectional geometry
member length member length
applied load applied load
eccentricity of load
FUNCTIONS GEOM
superscripts refer CENTROID1
to the sample output SECTPROP
diagrams below
MATMODEL2 SECTFORCE3
GETSTRESS NMBOUND
MFACTOR CONTOURCURV4 MOMCURV5 RESPONSE7
ULTIMATE
DEFLECTION6 CAPACITY8
depth
ENVELOPES
plane stress
width of strain
element strip
moment
axial moment
force
1. CENTROID
curvature
axial load
3. SECTFORCE ultimate load
stress
5. MOMCURV
contour of
axial equal curvature
force 7. RESPONSE
strain + eccentricity
column
6. DEFLECTION
moment
8. CAPACITY
300
B. Details of Computer Program QULT
geom(up) 1. Input up=1 for upside geometry or up=-1 for upside down geometry.
2. Input sectional geometry (defined within function) by entering depth in
mm, width in mm, and number of strips per element. Up to four stacked
rectangular elements can be defined.
3. Calculate lever arms of strips with respect to bottom of section.
Calculate heights and widths of strips.
4. Reverse geometry if up=-1.
5. Output n×3 matrix. Row i is for strip i of the section. Column 1 is for lever
arm from the bottom of the section in mm, Column 2 is for strip height
in mm, Column 3 is for strip width in mm.
centroid(up) 1. Input up=1 for upside geometry or up=-1 for upside down geometry.
2. Call geom(up) to get geometry data.
3. Calculate strip area and strip moment area.
4. Calculate centroid of section.
5. Calculate strip lever arm with respect to centroid
6. Output n×2 matrix. Row i is for strip i of the section. Column 1 is for lever
arm from the centroid of the section in mm, Column 2 is for strip area in
square mm.
301
B. Details of Computer Program QULT
sectforce(curv, etop, up) 1. Input curvature curv, top strain etop and up=1 for upside geometry
or up=-1 for upside down geometry.
2. Call geom(up) to get geometry.
3. Call centroid(up) to get lever arm and area data.
4. Call mfactor to get material factor.
5. Calculate strain, stress, force, and moment about centroid of each
strip in the section.
6. Calculate axial force by summing forces, and moment by summing
moments about centroid.
7. Output 1×2 matrix: axial force in MN and moment in MN·m.
8. Plot plane of strain, stress profile, and force profile (optional).
nmbound(up) 1. Input up=1 for upside geometry or up=-1 for upside down geometry.
2. Find limiting compressive strain from matmodel.
3. Find distance from centroid to top fibre.
4. Find curvatures corresponding to planes of strain defined by top strain
at limiting compressive strain and the neutral axis crossing at the centre
of each strip.
5. Call sectforce to find axial forces and moments of each plane of strain.
6. Output n×3 matrix: axial force in MN, moment in MN·m, and curvature.
contourcurv(up) 1. Input up=1 for upside geometry or up=-1 for upside down geometry.
2. Find maximum curvature and curvature due to balanced point, using
nmbound, geom, centroid, and matmodel.
3. Define increments of contours of equal curvature to calculate.
4. Calculate sets of axial force and moment for each contour using
sectforce.
5. Output n×3×k matrix to global variable. Each row corresponds to one data
point of axial force in MN, moment in MN·m, and curvature. Each matrix
corresponds to one contour of equal curvature.
6. Plot contours of equal curvature in axial force-moment space (optional).
302
B. Details of Computer Program QULT
envelopes(Qmax, Lmax, n) 1. Input maximum compressive load Qmax in MN, maximum length of
member Lmax in mm, and number of shorter columns to calculate n.
2. Set column lengths to be calculated.
3. Call capacity to get for ultimate load and ultimate moment for each
column length.
4. Output m×3×n matrix. Row m corresponds to one data point of ultimate
load in MN, ultimate mid-length moment in MN·m, and end moment in
MNm. Matrix n holds data for one column length.
5. Plot member capacity envelopes for each column length considered
(optional).
303
B. Details of Computer Program QULT
response(Qmax, L, eccen) 1. Input maximum compressive load Qmax in MN, maximum length
of member L in mm, and eccentricity of load in mm.
2. Define intervals of load for calculating column deflection curves.
3. Call momcurv to get moment-curvature diagram for current Q.
4. Find maximum moment and set mid-length deflection conditions for
each column deflection curve.
5. Calculate column deflection curves for current load Q.
6. Record list of mid-length moment and end eccentricity for current load Q.
7. Interpolate mid-length moment for given eccentricity of load eccen.
Repeat for all load Q.
8. Reorder list of load Q and mid-length moment M such that moments are
from lowest to highest.
9. Calculate deflection from load Q, moment M, and eccentricity eccen.
10. Output n×3 matrix. Each row corresponds to a load stage in a deflection
controlled test. Column 1 is load Q in MN, Column 2 is deflection w in
mm, and Column 3 is mid-length moment M in MNm.
11. Plot load-deflection response of column (optional).
ultimate(Qmax, L, eccen) 1. Input maximum compressive load Qmax in MN, maximum length
of member L in mm, and eccentricity of load in mm.
2. Call response to get load-deflection-moment response of column.
3. Find first and last instance of ultimate load and their indices.
4. Average the moments from the first and last instance.
5. Output ultimate load in MN, ultimate deflection in mm, and ultimate
moment in MNm.
304
Appendix C: Slender Column Drawings
This appendix contains drawings and photographs of the slender column load tests described in Section
3.4. Specimens made from high and ultra high-performance fibre-reinforced concrete were prepared and
tested according to these set of drawings and tested at the University of Toronto between September 2007
and December 2007. Specimens E and F, which were made from high-performance fibre-reinforced con-
crete are not specifically dealt with in the body of thesis, but are included here for reference.
Figure C-1 shows the general arrangement of the six specimens and their connection to the mts testing
machine, as well as the list of materials and fabrication. Figure C-2 shows the detail of the steel end plates
that were used to form the ends of each column. Studs were welded at the centre of each plate to provide
shear resistance between the concrete columns and their end plates. Figure C-3 shows the double tongue
assembly that receives the pin at each end of the column. It connects to the steel end plates of the columns
with four nuts and bolts. The larger Specimens E and F use threaded holes instead of nuts because of the
space requirements of the column. The eccentricity of load is created by offsetting the centreline of the pins
from the centreline of the column. Figure C-4 shows the plate gripped by the mts machine head that is
pinned to the top double tongue assembly. The drawing also shows the welded plate that is pinned to the
bottom double tongue assembly. This welded plate is bolted to the circular base plate, which is fixed to the
reaction frame. Figure C-5 shows the general arrangement of the plywood forms used to cast the six speci-
mens. Forms were designed to support the concrete specimens during erection and connection into the
305
C. Slender Column Drawings
test frame. Figure C-6 shows typical sections of the plywood forms and how they were nailed together. The
drawing also shows the positioning of the steel end plates relative to the ends of the forms. End plates were
clamped and secured with additional pieces of plywood to hold them in the correct position. Holes in the
end plates were temporarily plugged to prevent the concrete from spilling out during casting. Figure C-7
shows the types of instrumentation used during the load test. Surface strain gauges with lengths of 60 mm
were glued to the flexural compression face of each column. Tooling balls with 6 mm diameters were
glued along the centerline of the adjacent column face and used as targets for the cohesive laser radar
scanner. The horizontal displacement at mid-length of each column was measured by a linear variable di-
fferential transformer. Figure C-8 shows all the as-built measurements of the specimens including average
compressive strengths (from cylinder tests), eccentricities of load, and lengths, depths, and widths of
columns. These measurements are used as input parameters for the computer program qult (see Ap-
pendix B), which calculates the load-deflection response of the column according to the general method
306
C. Slender Column Drawings
307
C. Slender Column Drawings
308
C. Slender Column Drawings
309
C. Slender Column Drawings
Figure C-4. Top and bottom tongues. Scale reduced to 80% of original.
310
C. Slender Column Drawings
311
C. Slender Column Drawings
312
C. Slender Column Drawings
313
C. Slender Column Drawings
314
Appendix D: Design Calculations for Concept 80
This appendix contains a set of manual design calculations pertaining to Concept 80. Concept 80 is one of
the seventy-two ultra high-performance fibre-reinforced concrete arch bridge concepts proposed proposed
in Chapter 6. Concept 80 is described in greater detail in Section 6.2.3. The purpose of these calculations
are: (1) to demonstrate the use of the simplified arch analysis method presented in Chapter 4, which is used
for calculating maximum sectional forces in arch systems, (2) to demonstrate the use of the simplified
design method presented in Section 3.5, which is used for calculating capacities of ultra high-performance
fibre-reinforced concrete members in compression and bending, (3) to demonstrate how each of the sev-
enty-two design concepts presented in Section 6.2.1 were designed, and (4) to serve as a means of checking
There are a total of eighteen pages of calculations, each labelled from D1 to D18. Page D1 shows the trial
cross-sections of the arch and deck girder, along with the given material properties and global geometry.
Page D2 lists system properties that are used frequently in design calculations. Page D3 tabulates dead load
quantities from all primary and secondary structural components of the bridges. Page D4 shows calcula-
tions for all load quantities, including dead load, live load, shrinkage strain, and uniform drop in temper-
ature. Pages D5-D12 shows sample design calculations for ultimate limit state load combination 2, for
which sectional forces at midspan are critical. System properties are calculated on pages D5-D6. Long-
term sectional forces, short-term sectional forces, and second-order deflections and moments are calcu-
315
D. Design Calculations for Concept 80
lated on pages D7-D10. Page D11 shows calculations of fixed system moments. Pages D13-D15 shows
sample design calculations for ultimate limit state load combination 1, for which sectional forces at the
quarter-point are critical. Reduced, member capacity n-m interaction diagrams for the arch and deck
cross-sections are shown on page D12. Critical sectional force demands for ultimate limit state load com-
binations 1 and 2 are plotted on to the diagrams. Because these points lie within the envelopes, the pro-
Rudimentary shear force design calculations are shown on pages D16-D18. For simplicity, an equivalent
distributed load applied over half the span (calibrated to cause the same second-order moment shown on
page D14) is assumed. From this, bending moment, shear force, and axial force diagrams are calculated.
On page D17, Mohr’s circle of stress calculations are done for the arch cross-section at x=0.1L. The result-
ing principal stress is less than the factored cracking stress of the material. Shear stress calculations for di-
fferent values of x are done on a spreadsheet. Resulting critical principal stresses along the arch member
are plotted on page D18 (solid line). Principal stresses are most critical within the first 10% of the span, but
are still below the factored cracking limit. A secondary (dashed) curve is shown on the diagram, corres-
ponding to results for a section that has twice the web thickness and twice the top and bottom slab thick-
nesses. This is done because increased cross-section proportions are expected near the springing lines. The
secondary curve shows a significant reduction in critical principal stresses, as compared to the primary
curve.
316
D. Design Calculations for Concept 80
317
D. Design Calculations for Concept 80
318
D. Design Calculations for Concept 80
319
D. Design Calculations for Concept 80
320
D. Design Calculations for Concept 80
321
D. Design Calculations for Concept 80
322
D. Design Calculations for Concept 80
323
D. Design Calculations for Concept 80
324
D. Design Calculations for Concept 80
325
D. Design Calculations for Concept 80
326
D. Design Calculations for Concept 80
327
D. Design Calculations for Concept 80
328
D. Design Calculations for Concept 80
329
D. Design Calculations for Concept 80
330
D. Design Calculations for Concept 80
331
D. Design Calculations for Concept 80
332
D. Design Calculations for Concept 80
333
D. Design Calculations for Concept 80
334
Curriculum Vitae
Jason Angeles Salonga jason.salonga@utoronto.ca
Education
B.E.Sc. in Civil Engineering 2004 University of Western Ontario, London, Ontario, Canada
B.A. in Visual Arts 2004 University of Western Ontario, London, Ontario, Canada
Selected Awards
nserc Canada Graduate Scholarship (Doctoral) May 2007 to April 2009
Ontario Graduate Scholarship in Science and Technology January 2007 to April 2007
sion-making in the Red River Basin. Canadian Water Resources Journal. 29(4): 251-266.
Salonga, J. A. 2008. Rational analysis of slender ultra high-performance fibre-reinforced concrete columns. 7th fib
Salonga, J. 2007. Parallels in teaching visual arts and engineering design. Engineering Teaching and Learning Prac-
tices, ASEE St. Lawrence Section Conference in Toronto, Canada, October 19-20.
Non-referred reports
Salonga, J. 2007. The state of modern bridge design and construction in Taiwan. Report for nserc Summer Program
Salonga, J. 2004. Aggregation methods for multi-objective flood management decision making: Red River Basin case