You are on page 1of 12

Comparative Study of the Proportions, Form,

and Efficiency of Concrete Arch Bridges


Jason Salonga, A.M.ASCE1; and Paul Gauvreau2
Downloaded from ascelibrary.org by Massachusetts Institute of Technology (MIT) on 11/20/15. Copyright ASCE. For personal use only; all rights reserved.

Abstract: This article presents a comparative study of 55 concrete arch bridges built over the last century. For each bridge considered in the
study, section properties and geometrical ratios were determined and compiled into a database. These data were used to identify empirical trends
related to important attributes such as proportions, stiffness, slenderness, and efficiency. These trends were then expressed as functions of span
length. The database and empirical trends together provide a unique and comprehensive quantitative characterization of the state of the art of
concrete arch bridges. By situating new design concepts within the trends compiled by this study, designers of arch bridges can gain insight into
the implications of specific design decisions with regard to structural behavior and efficiency and will have a quick and reliable means of es-
timating the nature and extent of the effort required for validation. DOI: 10.1061/(ASCE)BE.1943-5592.0000537. © 2013 American Society of
Civil Engineers.
Author keywords: Bridges, arch; Bridges, concrete; Reinforced concrete; Comparative studies; Databases; Conceptual design; Deck-
stiffened arches; Slender arches; Shallow arches; Design aides.

Introduction indication of the implications of deviating from the recommended


values.
Bridge designers rely extensively on the knowledge of previously These difficulties can be addressed by a more systematic and
completed works. When designing a bridge to satisfy a given set of better-documented study of existing works. Providing engineers
requirements, for example, knowledge of the structural system, with actual values of structural characteristics taken from individual
dimensions, and details of a completed bridge that satisfies similar bridges within a representative set not only provides a logical basis
requirements can be used as a starting point for the generation of for the definition of representative values (e.g., the average value of
design concepts. This is an efficient process, since a new design a given characteristic) but also gives designers a basis on which to
concept produced in this way is effectively prevalidated by reference assess the implications of deviating from the representative values.
to the antecedent work. Links to previously built works can be Documenting the origin of each individual characteristic within the
identified in practically every bridge. Designers of average ability set gives credibility to the data.
may choose to keep their designs close to reference works, whereas One of the best known studies of this type is a compilation of
designers of considerable creativity may depart markedly from construction cost data for a set of 19 prestressed concrete bridges
precedent. built in Switzerland before 1979 (Menn 1990). In the originally
In addition to the insight available to designers from the detailed published version of that study (Menn 1979), scaled longitudinal and
consideration of individual reference bridges in their entirety, there cross-section drawings were included. For each bridge, the study
is also much to be gained from consideration of specific structural expressed the construction cost of primary components such as
characteristics within a group of similar reference bridges. For superstructure concrete and superstructure prestressing steel as
example, designers rely extensively on span-to-depth ratios rec- percentages of total construction cost. Designers can use this in-
ommended by standard texts [such as Menn (1990), Leonhardt formation to estimate the effect on overall cost of a change in a given
(1979), and Mondorf (2006)] or other sources to proportion structural quantity.
structural members in preliminary design. These ratios are par-
ticularly important because primary dimensions are often chosen Objective and Organization of Article
on this basis alone. Because these values have generally not been
justified in any formal way, they can take on the character of rules The primary objectives of this study are to characterize the state of
of thumb. Furthermore, these ratios alone do not provide any the art of concrete arch bridges by means of a comprehensive da-
tabase and to identify and describe relevant trends in geometrical
properties across the set of bridges considered. The database that
was compiled for this article describes each structure graphically and
1
Bridge Engineer, International Bridge Technologies, Inc., 9325 Sky quantitatively. These descriptions, together with the identified em-
Park Court, Suite 320, San Diego, CA 92123; formerly, Postdoctoral Fellow, pirical trends among primary quantitative parameters, can be used in
Dept. of Civil Engineering, Univ. of Toronto, Toronto, Canada M5S 1A4 the preliminary assessment of feasibility and efficiency of a given
(corresponding author). E-mail: jason.salonga@utoronto.ca design concept, and thus provide valuable guidance to designers.
2
Associate Professor, Dept. of Civil Engineering, Univ. of Toronto,
This article is divided into two parts. The first presents the da-
Toronto, Ontario, Canada M5S 1A4. E-mail: pg@ecf.utoronto.ca
Note. This manuscript was submitted on October 9, 2012; approved on tabase of concrete arch bridges and describes the process used to
July 16, 2013; published online on July 18, 2013. Discussion period open select bridges for the database and the method used to deal with
until May 9, 2014; separate discussions must be submitted for individual ambiguous geometrical data. The second part describes trends re-
papers. This paper is part of the Journal of Bridge Engineering, © ASCE, lating to proportions, stiffness, slenderness, flatness, shallowness,
ISSN 1084-0702/04013010(12)/$25.00. and efficiency arising from the data.

© ASCE 04013010-1 J. Bridge Eng.

J. Bridge Eng., 2014, 19(3): 04013010


Database of Concrete Arches There is at least one significant precedent for comparative
studies based on a relatively small set of bridges. Menn’s study
of bridge construction costs of prestressed concrete bridges
Scope of Database (Menn 1990), briefly described in the introduction, has been
The database is restricted to highway bridges built entirely of con- regarded as extremely useful by designers (Schlaich and Scheef
crete and for which the arch supports the bridge deck from below. 1982), despite the fact that it is based on data from only 19
Limiting the scope in this way results in a reasonable balance be- bridges.
tween consistency and diversity, which enables meaningful infer-
ences to be drawn from the database.
The database was not intended to be an exhaustive compilation of Content of Database
Downloaded from ascelibrary.org by Massachusetts Institute of Technology (MIT) on 11/20/15. Copyright ASCE. For personal use only; all rights reserved.

record spans or a definitive chronicle of historical development, but


rather a source of documented information that could be used for Every bridge that was found in the search that satisfied the scope was
quantitative analysis. It was therefore necessary to ensure that, for included in the database. No bridge that satisfied the scope was
every bridge in the database, numerical values could be defined excluded on the basis of any subjective criterion. This yielded a set
for a minimum set of geometrical properties. This implied that of 55 bridges.
a given bridge could be included in the database only if drawings Elevations and typical sections of the bridges in the database are
were available of precision and detail sufficient to extract arch span, shown in Fig. 1. Bridges are identified by name, year of completion,
arch rise, and complete dimensions of arch and girder cross-sections. and bridge ID numbers, which have been assigned in decreasing
As a result, many bridges that appear prominently in online image order of span length. The elevations in Fig. 1 have all been drawn to
repositories and inventories of record spans were not included in the same scale.
the database. The typical sections include a cross section through the girder, an
elevation of a typical spandrel column, and a cross section through
the arch. The same scale was used to draw these views for the entire
set of bridges. The arch cross sections are cut perpendicular to the
Sources of Information
arch axis and are drawn with dimensions averaged from sections at
An extensive search was conducted to populate the database. Given the crown and springing line. Girder cross sections are cut vertically.
the need for precise and detailed drawings, the search concentrated A cross section through the spandrel column is given immediately to
primarily on descriptive articles published in technical journals with the right of each typical section. In some cases, dimensions of
high editorial standards. An initial keyword search in Engineering spandrel columns were inferred from available elevation and section
Index and a review of bibliographies included in articles by Chen views. Unless otherwise specified in references, spandrel columns of
(2007), Ewert (1999), and Savor and Bleiziffer (2008) and a book thickness 1.5 m or greater were assumed to be hollow, with wall
chapter by Mondorf (2006) yielded a number of bridges and helped thickness of 300 mm.
to identify the following journals as sources that warranted closer Fig. 2 gives, for each bridge, data obtained from the reference and
attention: computed section properties for arch, girder, and columns. A visual
• Journal of Bridge Engineering: 1996 to 2009; definition of the primary dimensions is shown in the elevation and
• Journal of Structural Engineering: 1983 to 2009; section views of the Krk I Bridge (ID 2) in Fig. 1. The following
• Journal of Structural Design and Construction: 1996 to 2009; assumptions and guidelines were used in compiling the information
• Canadian Journal of Civil Engineering: 1974 to 2009; for Fig. 2:
• Proceedings of the ICE – Bridge Engineering: 2003 to 2009; • Slab-on-girder sections were assumed to be composite;
• Structural Engineering International: 1991 to 2009; • Arch depth and girder depth are measured perpendicular to the
• Der Bauingenieur: 1985 to 2009; longitudinal axis of the respective members;
• Bautechnik: 1985 to 2009; • For variable depth arches, the dimensions used for calculating
• Beton- und Stahlbetonbau: 1985 to 2009; and section properties were averaged from sections at the crown and
• Schweizer Ingenieur und Architekt: 1985 to 1996. springing line; and
The range of years provided for a given journal reflects online • Regarding girder continuity (abbreviated as “Cont.” in the
availability through the University of Toronto library. Each volume figure), the term “discontinuous” refers to girders that span
in the ranges indicated was searched in detail. between spandrel columns as simply supported beams, and
The proceedings of the first six International Conferences on “continuous” refers to girders that are continuous over spandrel
Arch Bridges, held from 1995 to 2010, were searched in detail. columns.
Design drawings of several Canadian and American concrete arch Fig. 3 shows text citations referring to the source document
bridges, which were procured from contacts of the authors, were also references followed by parameters computed from the basic geo-
examined. metrical data compiled in Fig. 2. These quantities, which charac-
These sources are by no means exhaustive. Despite this, how- terize the bridges in the database in terms of proportion, efficiency,
ever, it is expected that the set of bridges complied from these and system behavior, will be defined and discussed in detail in the
sources will provide a reasonably faithful impression of the state of following section.
the art of concrete arch bridge design, and for this reason, insights In Fig. 2 and Fig. 3, the “?” symbol indicates values that were not
obtained from the database will be generally useful. Subsequent available in the source document or that could not be determined
sections of this article will demonstrate that there are rational with sufficient accuracy. The solid circle symbol indicates values
explanations for all of the empirical trends arising from the database that are not applicable owing to the unconventional form of a given
and that these trends are generally consistent with observations bridge. For example, spandrel column properties cannot be calcu-
made by other authors. This would likely not be the case if the set of lated for the Galena Creek Bridge because it has none. The Salgi-
bridges was populated by exceptional examples not representative natobel Bridge, with its novel design, has an arch that changes
of the state of the art. A future expansion of the database and review considerably in depth and shape and fuses with the deck for half its
of the observed trends has been planned. span. Although average values for system parameters could be

© ASCE 04013010-2 J. Bridge Eng.

J. Bridge Eng., 2014, 19(3): 04013010


Downloaded from ascelibrary.org by Massachusetts Institute of Technology (MIT) on 11/20/15. Copyright ASCE. For personal use only; all rights reserved.

Fig. 1. Elevation and section views of the 55 concrete arch bridges considered in this study

© ASCE 04013010-3 J. Bridge Eng.

J. Bridge Eng., 2014, 19(3): 04013010


Downloaded from ascelibrary.org by Massachusetts Institute of Technology (MIT) on 11/20/15. Copyright ASCE. For personal use only; all rights reserved.

Fig. 2. Geometrical data recorded for bridges considered in this study

estimated for this bridge, comparisons with other bridges in the Flatness (Span-to-Rise Ratio)
database would not be meaningful for certain parameters because of
The span length L and rise f of the arch are usually the first two
differences in form and overall system behavior.
dimensions to be determined in the design process. The span-to-rise
ratio L=f not only provides a simple quantitative characterization of
Empirical Trends the overall visual form of the arch, it also plays an important role in
determining many aspects of structural behavior. Fig. 4 depicts L=f
This section presents observations on the design and structural be- as a function of span length L for the bridges considered in this study.
havior of concrete arch bridges arising from relationships among the The values of L=f range between 2.3 and 11.2. The mean value is 5.3
geometrical quantities collected in the database and parameters com- and the coefficient of variation is 0.37. All but three of the bridges
puted from these quantities. (95% of the sample) have L=f between 2.3 and 8.0.

© ASCE 04013010-4 J. Bridge Eng.

J. Bridge Eng., 2014, 19(3): 04013010


Downloaded from ascelibrary.org by Massachusetts Institute of Technology (MIT) on 11/20/15. Copyright ASCE. For personal use only; all rights reserved.

© ASCE
04013010-5

J. Bridge Eng., 2014, 19(3): 04013010


Fig. 3. System parameters computed based on geometrical data given in Fig. 2

J. Bridge Eng.
database, the Infant Henrique Bridge, allude to the fact that the
locations of the springing lines were determined by tight site con-
straints (Adão da Fonseca and Mato 2005). The significant varia-
tion in the observed values of L=f as a function of L thus appears
consistent with the important role played by aesthetic considerations
and given roadway and site geometry on the choice of the springing
lines.

Arch Depth and Girder Depth


Downloaded from ascelibrary.org by Massachusetts Institute of Technology (MIT) on 11/20/15. Copyright ASCE. For personal use only; all rights reserved.

Arch depth and girder depth are chosen early in the design process.
Both dimensions have a significant effect on the visual qualities of
a given bridge as well as structural response, since they are directly
related to the axial and bending stiffness of the respective mem-
bers. Leonhardt (1979) makes no specific quantitative recom-
mendations regarding member depth. Menn (1990) recommends
that girder depth be constant over the arch and approaches and that
girder span-to-depth ratio be chosen between 12:1 and 15:1 in the
Fig. 4. Span-to-rise ratio as a function of span length
approach spans. Menn makes no specific recommendations for
arch depth.
The upper two graphs of Fig. 5 plot average arch depth and girder
These observations are generally consistent with the recom- depth as functions of arch span length L. It is apparent that arch depth
mendations of Menn (1990), who proposed that L=f be chosen and girder depth both tend to increase with increasing values of L.
between 2 and 10. According to Menn, values of L=f below 2 are The bridges within the database can be divided into two groups:
visually awkward and difficult to build, presumably because of the those with arch depth greater than girder depth, referred to as deep-
need for the arch section to be completely enclosed by formwork arch systems, and those with girder depth greater than arch depth,
when it is cast in place. As L=f approaches 10, sensitivity to imposed referred to as deep-girder systems. The deep-arch systems, which
deformations due to creep, shrinkage, and change in temperature make up 43 of 55 bridges in the database, cover the entire range of
increases significantly. The data are also consistent with recom- spans. The deep-girder systems make up 12 of 55 bridges in the
mendations made by Leonhardt (1979). Recognizing the link be- database. With the exception of the Infant Henrique Bridge, which
tween stresses due to imposed deformations and the degree of has a span of 280 m, these bridges cover arch spans up to only 132 m.
statical indeterminacy, Leonhardt proposed the following ranges for Regression lines for girder depth and average arch depth were
L=f : 5 to 12 for three-hinged arches, 4 to 12 for two-hinged arches, computed separately for the two groups of bridges. The lines and
and 2 to 10 for fixed-end arches without internal hinges. Neither respective equations are given in Fig. 5. (The Pitan Bridge, a clear
author provides detailed quantitative justification for the proposed outlier, was excluded from these calculations.) The regression lines
ranges of L=f . The highest span-to-rise ratio from the database, 11.2, for the deep-arch and deep-girder systems present a relatively strong
exceeds the proposed upper limits of both Menn and Leonhardt for reciprocal relationship, in the sense that the regression line for girder
fixed-end arches, demonstrating that feasible solutions can be de- depth for the deep-arch systems is similar to the regression line for
veloped outside these recommended ranges. The data indicate, arch depth for the deep-girder systems, just as the regression lines for
however, that designers have generally not ventured into the upper arch depth for deep-arch systems and girder depth for deep-girder
ranges of span-to-rise ratios proposed by these two authors. systems bear strong similarities. This suggests that there may be
No obvious relation defining span-to-rise ratio as a function of significance to the sum of girder depth and average arch depth. This
span length can be identified from Fig. 4. This observation is con- quantity will be called system depth, dsys .
sistent with the negligible coefficient of determination (0.082) of the The bottom graph of Fig. 5 shows system depth plotted as
linear regression line. The lack of correlation of L=f with L can be a function of span length. The regression line and its equation are
understood in relation to the factors considered in the selection of given. (The Pitan Bridge was again excluded from the linear fit.) The
L=f in design. For given profiles of roadway and terrain to be relatively high coefficient of determination of the regression line,
crossed, the choice of arch span length generally leads to a relatively 0.88, confirms that it is a faithful representation of the data.
small number of possibilities for the rise (and sometimes only The strong correlation of dsys with L stands in contrast to the lack of
a single credible possibility). This is a consequence of the common correlation of L=f with L. In the previous subsection, it was proposed
practice of setting the chord joining the springing lines to be parallel that site-specific geometrical constraints severely limited the range of
to the roadway and of locating the crown of the arch as close as possible choices for arch rise once the location of the springing lines
possible to the girder. The bridge elevations of Fig. 1 are generally had been determined. In contrast, site-specific geometry has relatively
consistent with these principles. For a given site, therefore, suitable little impact on primary member depth. This gives designers con-
values of arch rise can generally be expressed as a single-valued siderable freedom to select these dimensions based on considerations
function of arch span length. related to structural efficiency and economy. This is consistent with
The choice of a suitable (L, f ) pair within this set depends on the much stronger relationship between member depth and span
a number of factors. Although lower values of L=f reduce the forces length observed from the data.
in the arch for a given load and minimize the effects of shortening in The strong empirical relationship between dsys and L can help
the arch, it is clear from the observed data that these simple statical designers select a suitable system depth once arch span has been
considerations are not the primary factor driving the choice of L=f . In determined. The data also indicate that designers have considerable
many cases, the given topography can suggest locations of springing freedom to apportion dsys to arch and girder, at least for spans up to
lines that are clearly better than any others. For example, the about 130 m, i.e., for the range of spans in which there is good
designers of the arch with the greatest span-to-rise ratio in the representation from both deep-arch and deep-girder bridges.

© ASCE 04013010-6 J. Bridge Eng.

J. Bridge Eng., 2014, 19(3): 04013010


Downloaded from ascelibrary.org by Massachusetts Institute of Technology (MIT) on 11/20/15. Copyright ASCE. For personal use only; all rights reserved.

Fig. 5. Arch depth, girder depth, and system depth as functions of span length

Although the Infant Henrique Bridge demonstrates that deep- Igirder I


girder systems can be built for long spans, the absence of such Mgirder ¼ Mtot and March ¼ Mtot arch (1)
Isys Isys
systems in the database between 132 m and 280 m raises the question
of whether deep-girder systems are suitable for long spans. This where Mtot 5 total bending moment due to nonuniform live load;
question will be discussed in greater detail later in this article. Igirder and Iarch 5 moment of inertia of girder and arch, respectively;
and Isys , which will be referred to as system moment of inertia,
5 Igirder 1 Iarch . (It is assumed that the modulus of elasticity of
Bending Stiffness
concrete is the same for both girder and arch.) Bridges with Iarch =Isys
Because most concrete arch/girder systems are highly statically close to 0.0 are called deck-stiffened arches. Bridges with Iarch =Isys
indeterminate, member stiffness plays an important role in de- close to 1.0 will be referred to as self-stiffened arches.
termining structural response. Bending stiffness is particularly (Quantity Igirder is understood to be a property of both the girder
important in this regard because arches must rely on bending to cross section and the overall structural system. For girders that are
resist nonuniform live load. This behavior is described mathe- continuous over the spandrel columns, Igirder is calculated as a section
matically in the following equations developed by Robert Maillart property. For girders that are discontinuous over the columns, Igirder is
(Billington 1973): assigned a value of zero, recognizing that a discontinuous girder

© ASCE 04013010-7 J. Bridge Eng.

J. Bridge Eng., 2014, 19(3): 04013010


cannot contribute to the overall bending stiffness of the arch/girder figure that, although Iarch =Isys is reasonably well distributed between
system. This definition of Igirder is used consistently throughout this 0 and 1 for relatively short spans, the minimum value of Iarch =Isys
paper and affects the computation of Isys and hence parameters deff and tends to increase with increasing values of L.
rsys defined later in the article.) The Infant Henrique Bridge, the only long-span deck-stiffened
It is therefore of benefit to designers to have a reliable basis for arch in the database, stands as an outlier. It confirms that the full
validating initial choices of bending stiffness. There are two primary range of Iarch =Isys can be indeed be used for relatively long spans. As
considerations in this regard. The first is the total stiffness to be stated previously, the high span-to-rise ratio of the Infant Henrique
provided for the overall system and the second is the distribution of Bridge was a response to unique site constraints. Arches of extreme
this stiffness to arch and girder. Although bending stiffness is flatness are particularly susceptible to the effects of shortening,
strongly influenced by member depth, depth alone gives only an which are proportional to bending arch stiffness. It is therefore likely
Downloaded from ascelibrary.org by Massachusetts Institute of Technology (MIT) on 11/20/15. Copyright ASCE. For personal use only; all rights reserved.

incomplete picture of stiffness. Two sections of identical width and that the designers chose a thin arch to minimize these effects.
depth, for example, can have significantly different moments of A suitable explanation of the dominance of self-stiffened arches
inertia depending on whether the section is solid or hollow. It is for long spans requires consideration of the load path of the partially
therefore important to consider bending stiffness explicitly. completed bridge during construction. The temporary structure used
Total bending stiffness of the system is characterized mathe- to support arches before they can carry load on their own is a sig-
matically by Isys . It is helpful to express this parameter in terms of nificant component of construction cost. For long spans, this cost
effective stiffness depth, deff , defined as follows: can be minimized by building arches in two halves, proceeding
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi outward from each of the springing lines. Until the two cantilevers
deff ¼ 3 12Isys =bdeck (2) meet at the crown, they must be supported with temporary stays.
The cost of these stays can be minimized by ensuring that the arch
alone can carry the weight of spandrel columns and girder at all
The quantity deff is thus the depth of a solid rectangle with width stages of construction. This would not cause a problem if these
equal to that of the bridge deck and moment of inertia equal to Isys . components could all be constructed in a single operation, since
The use of deff removes the effect of bdeck and thus provides a arches are normally dimensioned to carry the entire dead load of the
consistent basis for comparing total stiffness across a diverse set of structure. When the girder is built in stages from one side of the span
bridges. to the other, however, a significant nonuniform load is applied to the
Fig. 6 shows computed values of deff plotted as a function of L for arch. If the bending capacity of the arch is insufficient, the capacity
bridges in the database, together with a linear regression line. (The (and hence the cost) of the temporary structure must be increased
Pitan Bridge was not included in the regression calculation.) The accordingly.
coefficient of determination is 0.91. The strong linear trend is not For long-span arches, the area of concrete required to carry dead
a direct mathematical consequence of the strong linear relationship load is generally large enough to enable a cross section of high
between dsys and L observed in Fig. 5. Two arch/girder systems with bending capacity (i.e., a box) to be provided with minimal increase
identical dsys , for example, can have values of deff that differ by in permanent materials. Self-stiffened systems thus provide rela-
a factor of 2 or more. tively economical solutions with regard to the permanent structure.
For a given value of Mtot , Maillart’s equations are independent of The high bending strength of the arch also enables it to carry non-
arch span L. This implies that, at least with regard to load path for live uniform loads during construction on its own, thus minimizing the
load, there is essentially no restriction on the relative bending cost of the temporary structure. On the basis of this reasoning,
stiffness of arch and girder, for any given value of L. On this basis, therefore, it is reasonable to conclude that long-span deck-stiffened
therefore, one would expect to find values of Iarch =Isys well dis- arches are likely to be less economical than self-stiffened options
tributed between deck-stiffened and self-stiffened over the full range when the cost of temporary structures is considered.
of spans. Fig. 7, a plot of observed values of Iarch =Isys as a function of
L, shows, however, that this is not the case. It is apparent from this

Fig. 7. Ratio of arch moment of inertia to system moment of inertia as


Fig. 6. Effective stiffness depth as a function of span length a function of span length

© ASCE 04013010-8 J. Bridge Eng.

J. Bridge Eng., 2014, 19(3): 04013010


Concrete Quantity of dead and live load. The slenderness ratio l 5 kL=r is commonly
used for assessing the significance of second-order effects. For
The economical use of materials is an important consideration in
arches, it is convenient to define a system slenderness ratio, given by
preliminary design. For concrete girder bridges, it is common to
the following expression (adapted from Galambos 1998):
express the quantity of concrete in the superstructure in terms of
an effective slab thickness teff , defined as the total volume of concrete l ¼ kS=rsys (5)
in the superstructure (excluding diaphragms and other secondary
components) divided by the area of the bridge deck. This parameter
where k 5 effective arc length factor; S 5 arc length of arch; and rsys
provides a consistent basis for comparing the consumption of
5 system radius of gyration, defined as follows:
concrete across a set of bridges with dissimilar span lengths and
widths. For example, Menn (1990) proposed the following ex- qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Downloaded from ascelibrary.org by Massachusetts Institute of Technology (MIT) on 11/20/15. Copyright ASCE. For personal use only; all rights reserved.

pression (slightly simplified for this article) for estimating teff (in rsys ¼ Isys =Aarch (6)
meters) for posttensioned concrete girder bridges:
where Aarch 5 cross-sectional area of arch (taken as an average value
teff ¼ 0:35 þ 0:0045l (3) for variable depth arches). This formulation of rsys is reasonable,
since it accounts for the entire bending stiffness of the system (Isys )
where l 5 girder span length in meters. This equation is a linear but includes only the area of the arch, since the girder does not resist
regression line corresponding to data obtained from a set of com- axial compression. Austin (1971) proposed the following values for
pleted bridges in Switzerland. k: 0.35 for fixed arches, 0.50 for two-hinged arches, and 0.54 for
A similar parameter can be defined for arch bridges using the three-hinged arches.
following equation: System radius of gyration is plotted versus span length for the
bridges in the database in Fig. 9, together with a regression line,
teff ¼ V=ðL × bdeck Þ (4) which was computed excluding the Pitan Bridge. The coefficient of
determination is 0.90.
where V 5 total volume of concrete in the arch, girder, and spandrel System slenderness ratio l is plotted as a function of L for the
columns; bdeck 5 width of deck; and L 5 arch span length. bridges in the database in Fig. 10. The two lines plot values of l
Computed values of teff are plotted as a function of L in Fig. 8 for calculated for fixed arches based on values of rsys computed from the
the bridges in the database. Menn’s regression line for posttensioned regression equation for rsys . The upper and lower lines represent L=f
girders is also drawn, for spans up to 250 m, beyond which concrete of 3 and 8, respectively.
girders are extremely rare. Values of teff for the arch bridges tend to For short spans, observed values of l are reasonably well dis-
increase with L, although the scatter is significant. Although most of tributed between the minimum of 14 and the maximum of 99 for the
the arch points lie above Menn’s line for girders, a small number of Nanin Bridge. As L increases, the distribution of the points tends to
points lie below. This implies that although it is possible to design tighten from above and below toward the computed lines, i.e., an
concrete arch bridges that consume less concrete than girder bridges, average value of about l 5 61. Although the arch of the Infant
most concrete arch bridges have consumed significantly more con- Henrique Bridge appears slender when considered on its own, the
crete than girder bridges. This would imply that it is not only the cost of combined slenderness ratio of the arch/girder system, 54, is about
temporary structures that has hampered the competitiveness of arch one half that of the Nanin Bridge.
bridges in recent years, but also a high consumption of materials. The graph reveals that although arches of considerable slender-
ness have been built, designers of long-span arches have tended
Slenderness toward structures of more modest slenderness. The relatively broad
distribution of l for short-span arches deviates significantly from the
Menn (1990) discussed the importance of second-order effects in the lines computed on the basis of the regression equation for rsys . This
calculation of bending moments in arches under the combined action

Fig. 8. Effective slab thickness as a function of span length Fig. 9. System radius of gyration as a function of span length

© ASCE 04013010-9 J. Bridge Eng.

J. Bridge Eng., 2014, 19(3): 04013010


Downloaded from ascelibrary.org by Massachusetts Institute of Technology (MIT) on 11/20/15. Copyright ASCE. For personal use only; all rights reserved.

Fig. 11. Shallowness ratio as a function of span length


Fig. 10. Slenderness ratio as a function of span length

Bending moments in arches due to dead load can be reduced


would indicate that designers should exercise discretion in using this through the introduction of temporary hinges and/or jacking of the
equation as a basis for validating design concepts. arch at the crown. The data have been analyzed assuming that these
measures were not implemented during construction. They therefore
Shallowness represent an upper bound for bending in the arch due to dead load.
Shallowness ratio is plotted as a function of L in Fig. 11. Only
Arches that are sensitive to the effects of axial shortening will be
three bridges have b greater than the threshold value of 0.05. This
referred to as shallow arches. Axial shortening in arches affects
confirms that shallowness generally need not be considered for
structural behavior in two ways. First, it causes the crown to deflect
arches that have been designed with conventional proportions. The
downward, which reduces the rise of the arch and hence induces
three shallow arches in the database have maximum span-to-rise
additional compressive forces in the arch. Second, axial shortening
ratio of 7.7, which indicates that they have achieved shallowness on
induces redundant forces associated with restraint of rotation in
the basis of high Isys rather than flatness. The flattest arch considered,
statically indeterminate systems. These forces are proportional to the
Infant Henrique, has b 5 0:037, which is well below the proposed
bending stiffness of the system. For this reason, fixed arches always
threshold of 0.05. This implies that, at least with regard to shal-
resist dead load in some combination of compression (which will be
lowness, it may be possible to design even flatter arches than this.
referred to as arch action) and flexure (which will be referred to as
beam action). For a given arch, the relative proportion of dead load
resisted by beam action will be referred to as the shallowness of the Conclusions
system. Shallowness thus provides a means of quantifying the po-
sition of a given arch bridge on the spectrum bounded by pure arch This article describes a quantitative study of 55 concrete arch
behavior and pure beam behavior. bridges. For each bridge, dimensions of primary structural compo-
Salonga and Gauvreau (2010) proposed that the relative pro- nents were used to compute a small set of fundamental geometric
portions of arch action and beam action are governed by the shal- parameters. Values of each parameter were plotted as functions of
lowness ratio b, which is given by: arch span length. These plots revealed trends that can provide useful
guidance in the preliminary phase of the design process.
b ¼ rsys =ðh f Þ (7)
Flatness (Span-to-Rise Ratio)
where h 5 fixity factor that accounts for the support conditions of the
arch (i.e., pinned or fixed); f 5 arch rise; and rsys 5 system radius of The graph of L=f as a function of L exhibits significant scatter, which
gyration defined previously. A greater proportion of arch action implies that designers have generally chosen L=f in response to site-
(smaller values of b) is thus expected with increasing f , decreasing specific conditions of topography and roadway profile rather than to
Isys , and increasing Aarch . Salonga and Gauvreaup(2010)
ffiffiffiffiffi proposed the minimize compressive force. Fifty-two of 55 bridges had values of
following values of h: 2 for fixed arches and 24 for two-hinged L=f within the range 2.3 to 8.0. This is consistent with recom-
arches. Three-hinged arches are determinate and therefore do not mendations made by eminent designers such as Menn (1990) and
develop redundant forces owing to shallowness (b 5 0). Leonhardt (1979). The highest value of L=f within the database,
Salonga and Gauvreau (2010) proposed 90% as the minimum 11.2, exceeded these recommended ranges.
proportion of dead load that must be resisted by the arch in com- These observations imply that span-to-rise ratios within the range
pression for systems to be considered true arches. They observed that 2.3 to 8.0 should not pose undue problems. Although the presence
systems with shallowness ratios less than 0.05 generally satisfy this of one bridge in the database with L=f of 11.2 indicates that it is
requirement, and thus defined b 5 0:05 as a threshold value for true possible to exceed recommended ranges for flatness, the relatively
arch behavior. As b increases above 0.05, systems rapidly lose their small number of such bridges within the database would imply that
ability to carry permanent load in pure axial compression and behave this choice should always be informed by a sound understanding of
increasingly like curved, fixed-end beams. the structural behavior of flat arches.

© ASCE 04013010-10 J. Bridge Eng.

J. Bridge Eng., 2014, 19(3): 04013010


Arch Depth and Girder Depth owing to their high bending stiffness rather than flatness. The Infant
Henrique Bridge, the flattest arch within the set of bridges con-
The database included significant numbers of both deep-arch sys-
sidered, has a shallowness ratio of 0.037, which is well within the
tems and deep-girder systems. Although deep-arch systems were
threshold value. This would indicate that arches of even greater
present over the entire range of spans considered, deep-girder sys-
flatness might be feasible.
tems were restricted (with one exception) to spans of less than 140 m.
The trends identified in this article characterize primary aspects
In both types of system, both arch depth and girder depth increase
of the design of concrete arch bridges in terms of quantities that can
with increasing arch span length. Notwithstanding the considerable
be computed from dimensions of primary structural members. They
variation in the relative magnitude of arch depth and girder depth,
provide designers with a means of quickly assessing the validity of
there is a strong linear relation between system depth dsys (average
concepts. The identified trends cover most aspects of structural be-
arch depth plus girder depth) and arch span length. This relation,
Downloaded from ascelibrary.org by Massachusetts Institute of Technology (MIT) on 11/20/15. Copyright ASCE. For personal use only; all rights reserved.

havior of importance in preliminary design and are based on well-


together with the presence of one deep-girder system with L 5 280 m,
documented data obtained from a significant number of completed
would imply that deep-girder systems can be feasible for long-span
structures. This should give designers confidence in using these
arch bridges.
trends as a basis for taking a given concept forward into final design,
as well as an indication of what, if any, additional analytical work
Bending Stiffness may be required.
There is a strong linear relation between total bending stiffness of the
arch/girder system as characterized by effective stiffness depth deff
and arch span length L. This trend cannot be inferred mathematically Acknowledgments
from the linear relationship observed for dsys and L. Although the
distribution of system bending stiffness to arch and girder is The authors are grateful for the financial support provided by the
mathematically independent of L, the data indicate a trend toward Natural Sciences and Engineering Research Council of Canada.
self-stiffening with increasing values of L. The dominance of self-
stiffened systems for long-span arches can be explained by the need Notation
for flexural strength to resist nonuniform loads during construction
while minimizing the cost of temporary structures. The Infant The following symbols are used in this paper:
Henrique Bridge, the only exception to this trend, confirms, how-
Aarch 5 cross-sectional area of arch;
ever, the basic feasibility of long-span deck-stiffened arches.
bdeck 5 width of bridge deck;
deff 5 effective stiffness depth;
Concrete Quantity f 5 rise of arch;
There is considerable scatter in the computed values of effective slab I 5 moment of inertia of arch, girder, or system (arch
thickness (which includes the quantity of concrete in girder, col- 1 girder);
umns, and arch) plotted as a function of L. The general trend, k 5 effective arc length factor;
however, is an increase in quantity of concrete with increasing L. L 5 arch span length;
When compared with a familiar empirical expression for effective l 5 girder span length;
slab thickness of posttensioned concrete girder bridge super- M 5 bending moment of arch or girder, or total moment of
structures, the arches in the database generally consume more system;
concrete for a given span. This suggests that the competitiveness of rsys 5 system radius of gyration;
arch bridges is hampered not only by the cost of temporary structures S 5 arc length of arch;
but also by a relatively inefficient use of materials. A small number of teff 5 effective slab thickness;
bridges, however, confirm that it is possible to design arch bridges V 5 volume of concrete in arch, girder, and spandrel
that make a significantly more efficient use of materials than average columns;
girder consumption as defined by the empirical equation. b 5 shallowness ratio;
h 5 fixity factor; and
Slenderness l 5 system slenderness ratio.
The observed values of system slenderness ratio l are well dis-
tributed between 14 and 99 for short spans. With increasing values of References
L, the distribution tightens from both the top and the bottom toward
a value of approximately 61 within increasing values of L. This Adão da Fonseca, A., and Mato, F. (2005). “Infant Henrique Bridge over the
behavior cannot be inferred mathematically from previously iden- River Douro, Porto.” Struct. Eng. Int., 15(2), 85–87.
tified trend lines for the parameters used to calculate l. The data give Aigner, F. (1968). “Stahlbeton-Bogenbrücken auf der Österreichischen
no indication whether slender arches would be feasible for long Brenner-Autobahn.” Bauingenieur, 43, 91–95.
spans. Austin, W. J. (1971). “In-plane bending and buckling of arches.” J. Struct.
Div., 97(5), 1575–1592.
Baxter, D. J., and Balan, T. A. (2008). “Design of the Fulton Road Bridge
Shallowness precast segmental concrete arches.” J. Bridge Eng., 10.1061/(ASCE)
1084-0702(2008)13:5(476), 476–482.
The plot of computed shallowness ratio b as a function of span length
Baxter, J. W., Gee, A. F., and James, H. B. (1965). “Gladesville Bridge.” ICE
indicates that only three of 55 bridges in the database exceed the Proc., 30(3), 489–530.
proposed threshold value of 0.05, beyond which arches cannot be Bill, M. (1955). Robert Maillart. Trans. W. P. M. K. Clay. Girsberger,
assumed to carry permanent load in pure compression. This confirms Zurich.
that shallowness need not be considered for arches with standard Billington, D. P. (1973). “Deck-stiffened arch bridges of Robert Maillart.” J.
geometrical proportions. The three arches achieved shallowness Struct. Div., 99(7), 1527–1539.

© ASCE 04013010-11 J. Bridge Eng.

J. Bridge Eng., 2014, 19(3): 04013010


Blinkov, L. S., Cosolo, E., and Valiev, S. N. (2001). “Rehabilitation of a bridge Menn, C. (1990). Prestressed concrete bridges, Trans. P. Gauvreau, Birkhäuser
over the Matsesta River, Russian Fed.” Struct. Eng. Int., 11(3), 181–183. Verlag, Basel, Boston, Berlin.
Bouchet, A. (1964). “L’échangeur la Araña du croisement des autoroutes Mondorf, P. E. (2006). Concrete bridges, Taylor & Francis, New York.
nord-sud et est-ouest à Caracas (Vénézuela).” La Technique des Travaux Mörsch, E. (1909). “Die Gmündertobel-Brücke bei Teufen im Kanton
(Belgique), 40(7–8), 229–240. Appenzeil.” Schweizerische Bauzeitung, 53(7), 81–85.
British Columbia Ministry of Transportation and Highways (BCMOTH). Ostrander, J. R., and Oliver, D. C. (1987). “Construction of the Broadway
(1995). “Contract plans.” Big Qualicum River Upstream Bridge No. Bridge at Saskatoon, Saskatchewan, in 1932.” Can. J. Civ. Eng., 14(4),
3051, R6-V433-3051, British Columbia Ministry of Transportation and 429–438.
Highways, Victoria, BC, Canada. Petrangeli, M. P. (2007). “Prefabrication of medium span arch bridges.”
CALTRANS. (1932a). “Bridge across Bixby Creek (as built plans).” Doc. Proc., 5th Int. Conf. on Arch Bridges, Multicomp, LDA Publishers,
No. 50001001, State of California Department of Public Works Division Madeira, Portugal.
Downloaded from ascelibrary.org by Massachusetts Institute of Technology (MIT) on 11/20/15. Copyright ASCE. For personal use only; all rights reserved.

of Highways, Sacramento, CA. Podolny, W., and Muller, J. M. (1982). Construction and design of pre-
CALTRANS. (1932b). “Bridge across Rocky Creek (as built plans).” Doc. stressed concrete segmental bridges, Wiley, New York.
No. 50001081, State of California Department of Public Works Division Radic, J., Bleiziffer, J., and Tkalcic, D. (2005). “Maintaining safety and ser-
of Highways, Sacramento, CA. viceability of concrete bridges in Croatia.” Bridge Struct., 1(3), 327–344.
CALTRANS. (1937). “Bridge across Big Creek (as built plans).” Doc. No. 
Radic, J., Savor, ^
Z., Bleiziffer, J., and Kalafatic, I. (2008a). “Sibenik Bridge—
50001093, State of California Department of Public Works Division of Design, construction and assessment of present condition.” Proc., Chinese-
Highways, Sacramento, CA. Croatian Joint Colloquium: Long Arch Bridges, Univ. of Zagreb, Zagreb,
CALTRANS. (1996). “Bridge across Russian Gulch.” Earthquake Retrofit Croatia, and Univ. of Fuzhou, Fuzhou, China.
Project No. 714.0, State of California Department of Public Works 
Radic, J., Savor, 
Z., Prpic, V., Friedl, M., and Zderic,  (2008b). “Design
Z.
Division of Highways, Sacramento, CA. and construction of the Maslenica Highway Bridge.” Proc., Chinese-
Chen, B. (2007). “An overview of concrete and CFST arch bridges in Croatian Joint Colloquium, Univ. of Zagreb, Zagreb, Croatia, and Univ.
China.” Proc., 5th Int. Conf. on Arch Bridges, Multicomp, LDA Pub- of Fuzhou, Fuzhou, China.
lishers, Madeira, Portugal. Salonga, J., and Gauvreau, P. (2010). “Span-to-rise ratios in concrete arches:
Cheng, K. M. (1994). “The Pitan Bridge, Taiwan.” Struct. Eng. Int., 4(4), Threshold values for efficient behavior.” Proc., 6th Int. Conf. on Arch
231–234. Bridges, SECON HDGK, Croatia.
“Communications techniques Belges 8.” (1978). Annales des travaux publics 
Savor, Z., and Bleiziffer, J. (2008). “Long span concrete arch bridges of
de Belgique, no. 1–2, Fédération Internationale du Béton (fib), Lausanne, Europe.” Proc., Chinese-Croatian Joint Colloquium Long Arch Bridges,
Switzerland. Univ. of Zagreb, Zagreb, Croatia, and Univ. of Fuzhou, Fuzhou, China.
Dajun, D., and Weiqing, L. (1999). “Neue Entwicklungen bei Hochhäusern 
Savor, Z., Mujkanovic, N., Hrelja, G., and Bleiziffer, J. (2008). “Re-
und großen Brücken aus Beton in China.” Beton- und Stahlbetonbau, construction of the Pag Bridge.” Proc., Chinese-Croatian Joint Col-
94(4), 178–185. loquium: Long Arch Bridges, Univ. of Zagreb, Zagreb, Croatia, and
Eilzer, W., Schmidtmann, W., and Jung, R. (2005). “Die Wirrbachtalbrücke.” Univ. of Fuzhou, Fuzhou, China.
Beton- und Stahlbetonbau, 100(3), 236–240. Schlaich, J., and Scheef, H. (1982). Concrete box-girder bridges, In-
Ewert, S. (1999). “Betonbogenbrücken: Mehr als 200 m Spannweite.” ternational Association for Bridge and Structural Engineering, Zurich.
Beton- und Stahlbetonbau, 94(9), 377–388. Schwaab, E., and Gattner, A. (1953). “Straßenbrücke über das Tiefe Tal bei
Fritzell, G. (1960). “Deflection measurements on the Sandö Bridge 1942- Rosshaupten, Allgau.” Beton- und Stahlbetonbau 11.
1958.” Proc., 6th Congress of the IABSE, International Association for Sinotech. (2007). Drawings of Wunshuei, Lianlao, Nan Ke Bridges.
Bridge and Structural Engineering (IABSE), Stockholm, Sweden. Sinotech Engineering Consultants, Inc., Taipei, Taiwan.
Galambos, T. V. (1998). Guide to stability design criteria for metal State of Nevada Department of Transportation (Nevada DOT). (2000).
structures, Wiley, New York. B-1948-N/S Galena Creek Bridge, Carson City, NV.
Goodyear, D., Smit, N., Heacock, J., and Starkey, S. (2001). “Respect for Standfuß, F. (1990). “Nationalbericht: Brücken in der Bundesrepublik
tradition: The New Crooked River Gorge Bridge.” Troisième Confer- Deutschland.” Beton- und Stahlbetonbau 85(5), 106–113.
énce Internationale sur les Ponts en Arc, Presses des Ponts, Paris. Stellmann, W. L. O. (1966). “Brücke über den Rio Paraná in Foz Do Iguaçú,
Hünleim, W., and Ruse, P. (1985). “Ein neues Verfahren für den Bau von Brasilien.” Beton- und Stahlbetonbau 61(6), 145–149.
Bogenbrücken dargestellt am Bau der Argentobelbrücke Würgau.” Der Taiwan Area National Expressway Engineering Bureau (TANEEB). (1999).
Bauingenieur, 60(12). The second freeway: An anthology of particular bridges, Ministry of
Igarashi, T. (1976). “Die Hokowazu-Brücke in Japan, ein Stahlbetonbogen Transportation, Taiwan District, Taipei, Taiwan (in Mandarin).
im freien Vorbau.” Der Bauingenieur, 51(8), 281–284. Tandon, M. (1995). “Arch Bridge at Dodan Nallah.” Proc., 1st Int. Conf. on
Inaudi, D., Rüfenacht, A., von Arx, B., Noher, H. P., Vurpillot, S., and Arch Bridges, Thomas Telford, London.
Glisic, B. (2002). “Monitoring of a concrete arch bridge during con- Tanner, P., and Bellod, J. L. (2005). “Widening of the Elche De La Sierra
struction.” Proc., SPIE 4696, Smart Structures and Materials 2002: Arch Bridge, Spain.” Struct. Eng. Int., 15(3), 148–150.
Smart Systems for Bridges, Structures, and Highways, International Tonello, J., and Dal-Palu, P. (2001). “Le Pont du Triple Saut (the Triple Jump
Society for Optical Engineering (SPIE), Bellingham, WA. Bridge).” Troisième Conferénce Internationale sur les Ponts en Arc,
Kamimura, K., Kouno, M., Okada, K., and Tsuka, T. (2001). “Design and Presses des Ponts, Paris.
construction of the Tensho concrete arch bridge.” Troisième Conferénce Viet Tue, N., Dehn, F., Schliemann, T., Reintjes, K. H., and Tauscher, F.
Internationale sur les Ponts en Arc, Presses des Ponts, Paris. (2005). “Anwendung von Hochleistungsbetonen bei der Bogenbrücke
Koppel, A. J., and Walser, R. (1991). “Hundwilertobelbrucke: Ein bemer- Wölkau im Zuge der BAB A17.” Beton- und Stahlbetonbau 100(11),
kenswerter Neubau.” Schweizer Ingenieur und Architekt, 109(11), 931–938.
250–256. Wang, J. J., and Au, F. T. K. (2001). “Modong Hongshui River Bridge,
Leonhardt, F. (1979). Vorlesungen über Massivbau. Sechster Teil: China.” Struct. Eng. Int., 11(2), 101–103.
Grundlagen des Massivbrückenbaues, Springer-Verlag, Berlin, Hei- Wößner, K., Gebhardt, H., Schnabel, R., and Wörner, H. (1979). “Die Tal-
delberg, New York. brücke Rottweil-Neckarburg.” Beton- und Stahlbetonbau 74(10), 237–243.
Liebenberg, A. C., and Latimer, M. G. (2001). “Bloukrans Bridge.” Troisième 
Zderic,  Runjic, A., and Hrelja, G. (2008). “Design and construction of
Z.,
Conferénce Internationale sur les Ponts en Arc, Presses des Ponts, Paris. Cetina Arch Bridge,” Proc., Chinese-Croatian Joint Colloquium on
Ministry of Transportation of Ontario (MTO). (1935). Bronte Creek Long Arch Bridges, Univ. of Zagreb, Zagreb, Croatia, and Univ. of
Bridge. As-built plans. Ontario Department of Highways, Toronto. Fuzhou, Fuzhou, China.
Menn, C. (1979). Brückenbau I, ETH Abteilung für Bauingenieurwesen, Zimmermann, W., and im Gailtal, N. (1999). “Der Bau der Stampf-
Zurich. grabenbrücke.” Beton- und Stahlbetonbau 99(4), 304–310.

© ASCE 04013010-12 J. Bridge Eng.

J. Bridge Eng., 2014, 19(3): 04013010

You might also like