CONSIDERATIONS FOR DESIGN OF NETWORK ARCHES

By Per Tveit1

ABSTRACT: Network arches can be defined as those with inclined

hangers with multiple intercetions. In network tied arch bridges, the
arch and the tie can be made more slender than in other arch bridges.
Downloaded from ascelibrary.org by Universidad De Sevilla on 05/11/15. Copyright ASCE. For personal use only; all rights reserved.

The tie is usually a concrete slab roadway with longitudinal prestress.

For load cases that relax a few hangers, network arches act very much
like trusses. Thus, they have little bending in the tie and the arches. To
avoid relaxation, the hangers should not be inclined too steeply. Too
small an inclination of the hangers will increase the bending moments
due to concentrated loads. Compared with similar spans of conventional
steel bridges, the network arch usually saves half the weight of steel.
During erection, the arch and hangers, supplemented by a light lower
chord, can support a concrete roadway while it is being cast. The
network arch is most appropriate when local conditions lead to accept-
ance of structural members above the roadway. This seems to occur
more often for railway bridges than for road bridges.

INTRODUCTION

Network arches have inclined hangers with multiple intersections. The

hangers connect the arch and tie at small intervals, leading to small
bending moments. Some railway bridges built in Saxony, Germany,
between 1860-1880 have skeleton lines like a network arch (Berger 1955).
(See Fig. 1.) Recently, several network arches and bridges related to
network arches have been built (Tveit 1964, 1966; Stein and Wild 1965;
Takagai, et al. 1970; Kikuno 1973; Snelling 1980). It is the purpose of this
paper to discuss optimal design of network arches, but some basic facts
will be presented first.
A simply supported span is always some type of beam with a tensile and
compressive flange and a web in between. The network arch in Fig. 2 is
also a simply supported beam, where the arch is the compressive flange,
the tie is the tensile flange, and the hangers are the web. The axial forces
in the tensile and compressive flanges are inversely proportional to the
distance between the flanges. In tied arches, aesthetic reasons limit the
distance between the arch and the tie. Thus, saving of materials depends
mainly on whether or not a design gives light chords and a light web.
Efficient load carrying of the inevitable axial forces in the span comes
from the small bending stresses in all members of the structure and good
lateral support of the arch in order to achieve high buckling stresses. Small
bending moments with good lateral support of the arch can be utilized for
making very slender spans. In the bridge shown in Fig. 2, the depth of the
arch is 0.42 m (1 ft 5 in.) and the depth of the tie is 0.5 m (1 ft 8 in.). If
slenderness of an arch bridge is the length of span divided by the sum of the
depths of the arch and the tie, then this network arch is probably the most
slender arch bridge in the world (Tveit 1980b).
•Docent, Agder Regional College of Engrg., N-4890 Grimstad, Norway.
Note. Discussion open until March 1, 1988. To extend the closing date one
month, a written request must be filed with the ASCE Manager of Journals. The
manuscript for this paper was submitted for review and possible publication on May
7,1985. This paper is part of the Journal ofStructural Engineering, Vol. 113, No. 10,
October, 1987. ©ASCE, ISSN 0733-9445/87/0010-2189/$01.00. Paper No. 21892.
2189

J. Struct. Eng. 1987.113:2189-2207.

 100m (328«)
Downloaded from ascelibrary.org by Universidad De Sevilla on 05/11/15. Copyright ASCE. For personal use only; all rights reserved.

FIG. 1. Skeleton Lines for Railway Bridge Built 1877-1880

Minimizing Bending in Chords. Two sets of inclined hangers (Fig. 2)

efficiently restrain the horizontal displacement in the plane of the arch. In
a good design of a network arch, any possible loading is distributed by the
hangers to the arch in such a way that the line of thrust deviates little from
the centerline of the arch, giving small bending in the upper and lower
chord.
Since the efficiency of a network arch largely depends on the suppres-
sion of bending moments in the planes of the arches, it is worth looking
more closely at the sources of the bending in the chords.
Four Sources of Bending in Network Arches. Distinguishing between
sources of bending moments in chords gives a better understanding of how
network arches should be designed.
The four sources are: (1) Bending due to concentrated loads and abrupt
changes in load intensity; (2) bending due to relaxation of hangers; (3)
bending due to distance between nodal points; and (4) secondary bending.
Regarding the first source, bending due to concentrated loads and abrupt
changes in load intensity, it is sometimes useful to look upon a network
arch as a series of individual trusses placed on top of one another. In Fig.
1, for instance, three individual trusses have the chords in common, which
will make the tresses deflect together. Figs. 3(a-b) show how a concen-
trated load gives rise to shear forces, indicated by arrows. These shear
forces give rise to local bending. Similar shear forces arise due to any
abrupt change in load intensity.
The steep hangers shown in Fig. 3(a) would give the shear forces in the
chords shorter lengths to travel in order to make all trusses deflect equally.
Thus, steep hangers normally lead to small bending due to a concentrated
load.
Three sets of hangers can also be used to reduce bending due to
concentrated loads. By comparing Figs. 3(a-b), it can be seen that three
sets of hangers shorten the distance that the shear forces in the chords
have to travel in order to make all trusses deflect equally.

FIG. 2. Network Arch Road Bridge Built 1963 at Bolstadstraumen i Norway (Tvelt
1966, 1980; Tveit et ai. 1978)

2190

J. Struct. Eng. 1987.113:2189-2207.

Downloaded from ascelibrary.org by Universidad De Sevilla on 05/11/15. Copyright ASCE. For personal use only; all rights reserved.

FIG. 3. Deflections and Shear Forces due to Concentrated Load in Network Arches
with Two and Three Sets of Hangers

Three and four sets of hangers (Tveit 1984a-c) are most advantageous in
spans with slender chords, because these spans have the greatest need for
reduction of bending. However, network arches with three or four sets of
hangers will not be discussed herein.
In terms of the second source, bending due to relaxation of hangers,
network arches with all hangers in tension act as trusses and have little
bending in the chords. However, live load on one side of the span can
make hangers relax. Fig. 4 (Tveit 1980a) shows bending and deflection in
a bridge discussed in appendix I of another paper by the writer (1984b).
Here, the left 54% of the span carries a live load equal to the dead load in
the tie. The cross section of the tie in this bridge is shown later in Fig.
11(b). The dotted hangers in Fig. 4 are relaxed due to live load. They are
numbered according to the sequence in which they relax.
The segments of the chords marked a belong to parts of the arch which
act like a truss, i.e., where all hangers are in tension. The segments of the
arch marked b are attached by hangers in tension to a section of the span
acting like a truss. The chords marked c are connected to each other by
one set of hangers in tension. This part of the bridge functions like a tied
arch with one set of hangers. The equilibrium of the bridge in zone c is
dependent on shear and bending in the chords, and this can lead to large
bending moments.
Zones a and b are more firmly held in place than zone c. Relaxation of
hangers causes significant increases in bending moments in the chords only
after zone c occurs, and even then bending moments do not increase as fast
as the moments in a tied arch with vertical hangers. This is because the
sloping hangers restrain the horizontal displacement of the arch and
because part of the network arch works like a truss.
Even if some hangers relax, moderate live load on part of the span gives
smaller maximum stresses in the arch than the same live load on the whole
span. This is because the partial live load gives a small axial force in the
arch. Considerable bending moments, due to relaxation of hangers, are
needed so that the two load cases cause equal maximum stresses. Fig. 5
shows how maximum stress in member 114 increases with increasing live
2191

J. Struct. Eng. 1987.113:2189-2207.

 Downloaded from ascelibrary.org by Universidad De Sevilla on 05/11/15. Copyright ASCE. For personal use only; all rights reserved.

"^

18.0
17.5

20.9
20.8
11.2
to 3it o • » i1?SI 5ffl
JF f iffli I 1915s HJFsPs s « Ms 8 «»¥t * H—2-H-
[ Dead l o a d {§)®Sk&l^n (4728 Eb par f t )

| Uv® Bead Cp) 8® kHfa 14728 Eb psf f t ) |

.^ - - - -1

HO :

J. Struct. Eng. 1987.113:2189-2207.

• 1

-2
S3
Ig-I-I /-
^~ ^ s " >L. " .'
-1 .:: C v>
~ 1 -

2 I
FIG. 4. Deflection, Bending, and Axial Forces due to Unrealistically High Live Load on Left-Hand Side of 200-m (656-ft) Span; Hangers Are
Numbered in Order in which They Become Slack with Increase of Live Load
 ^max
Downloaded from ascelibrary.org by Universidad De Sevilla on 05/11/15. Copyright ASCE. For personal use only; all rights reserved.

Ratio of live load to dead weight of lower chord

FIG. 5. Development of Maximum Stress in Arch Member 114 of Fig. 4

load. One line shows how stresses increase with increasing load intensity
on the left 54% of the span, while another line shows how stresses increase
due to live load on the whole span.
The stress in member 114 for the load on the left 54% of the span is
shown in Fig. 5, because this load case is the combination of loads which
gives equal maximum stress due to full and partial loading for the lowest
live load intensity.
For the span in Fig. 4, Fig. 5 shows stresses that have been calculated in
the deflected state (nonlinear calculation), assuming a constant modulus of
elasticity. Stresses due to partial load are almost linear until the first hanger
relaxes. When hangers 1 and 2 have relaxed, the maximum stress in
member 114 increases equally due to a partial load and to a load of equal
intensity on the whole span. For a live load equal to 61% of dead load,
partial load and full load give the same maximum stress in member 114.
For this load case, hangers 1-6 are relaxed.
The network arch where hangers relax is a discontinous system, where
each combination of relaxed hangers leads to new equations for calculating
forces and deflections. This complicates design because influence lines can
no longer be used. By giving the hangers a smaller angle with the lower
chord, their tendency to relax is reduced, and thus bending due to
relaxation is reduced. The smaller angle with the chords would, however,
increase bending due to concentrated loads. Since it is rather complicated
to calculate spans where hangers relax, it saves considerable calculations
if the hangers have the type of slopes from which it can easily be seen that
the chords' dimensions are determined by the loads on the whole span.
This strategy was used for railway bridge i (Tveit 1973) and for the bridges
described by Tveit (1980a) and in appendix I of another paper by the writer

2193

J. Struct. Eng. 1987.113:2189-2207.

 Cross-wet i o m M0» e-io*
of half brlQfh m2
tart© SI 28.1 11.3
Hangws 0019 1M
Arch: 5
-
.SM 210

ill
8-20 .317
8bm® 20 .311
Downloaded from ascelibrary.org by Universidad De Sevilla on 05/11/15. Copyright ASCE. For personal use only; all rights reserved.

Plot®
1m • 3.20(1
1m2 " H.lfl2
1m" • 11»l»4
IHton?- 1 4 S p i l
1 M N • 225 kips
l l ^ m •
7MH-Wp»

•
- - - - ' •
- _ - • -
1 1
\ \ |

FIG. 6. Deflection, Bending, and Axial Forces in Singls-Track Railway Bridge

(1984b), where prediction of relaxation of hangers and choice of slope of

hangers was discussed.
The third source, bending due to distance between nodal points, is due
to continuous curvature of the arch and to transverse load between the
nodal points of the lower chord.
This is particularly important in relatively short spans. Fig. 6 shows
bending and deflection due to dead load and 80 kN/m (5,483 lbf/ft) on the
full length of a 67-m (219 ft 10 in.-) single-track railway bridge. The arches
are made from single-rolled steel sections and have constant curvature.
The diagram gives forces for an arch and for half of a lower chord, as do
the diagrams in Figs. 4 and 7 (shown later).
2194

J. Struct. Eng. 1987.113:2189-2207.

 Bending due to distance between nodal points can be reduced by using
more hangers.
Next, the fourth source, secondary bending will be discussed. In Fig. 6,
it can be seen that positive bending predominates, which is due to bending
caused by deflection of the span as a whole. This bending causes stresses
that are disregarded in the usual calculation of trusses. The stresses are
proportional to the depth of cross sections and can therefore be reduced by
Downloaded from ascelibrary.org by Universidad De Sevilla on 05/11/15. Copyright ASCE. For personal use only; all rights reserved.

choosing slender chords.

In most network arches, the lower chord will be made of partially
prestressed concrete. The creep and shrinkage of the concrete will make
the span bend upwards, and this will counteract the secondary bending due
to deflection and contribute to a good appearance. In the writer's opinion,
the tie of a bowstring arch should, for good appearance, have a camber of
at least 1 %c of the span. A straight tie looks as if it is sagging badly.
Calculation of network arches in the deflected state will give smaller
moments in the lower chord than the ordinary linear calculation (Tveit
1980a). This reduction is due to tension in the deflected lower chord and is
analogous to reduction of bending moments in the stiffening trusses of
suspension bridges due to tension in the cables. The reduction will be
greatest for long spans with slender lower chords. For the 200-m long span
in Fig. 4 (Tveit 1980a), reduction of bending moment due to tension was up
to 25%. Similarly, calculation in the deflected state increases the bending
moments in the arch. This is due to the tendency for buckling in the arch.

OPTIMAL ARRANGEMENTS OF HANGERS

The optimal arrangement of hangers will depend on many factors.

Important considerations are slenderness of arch and tie; ratio of live load
to dead load; intensity of evenly distributed live load; magnitude of
concentrated live load, structural codes, form of arch, etc.
In the present paper, it has been assumed that the arch is part of a circle,
because this makes fabrication easy. Furthermore, the constant curvature
gives a more constant axial force in the middle portion of the arch and
contributes to even maximum bending moments along the tie.
Many goals must be considered at when optimizing hanger arrange-
ments. The following list is incomplete: (1) Small bending in arch and tie;
(2) equal cross section and maximum utilization of tensile capacity in all
hangers; (3) reasonable resistance against hangers becoming relaxed; (4)
reasonable minimum hanger diameter; and (5) aesthetic appearance.
The writer does not know how to achieve these goals but will try to
explain how one would try to find an optimal hanger arrangement. Along
the way, relevant reasoning will be presented. Fig. 4 gives an example of
a span having a hanger arrangement that is thought to be near the optimal.
The writer believes that the hangers should be placed equidistantly along
the arch. This arrangement gives the smallest bending due to local
curvature of the arch when the span is fully loaded. Two hangers at each
nodal point would give more bending in the arch due to local curvature and
less efficient support of the arch in buckling. When hangers relax, due to
partial loading, the distance between points of support doubles and
bending due to local curvature increases. Maximum stresses, however, do
not necessarily increase, because load cases that make hangers relax do

2195

J. Struct. Eng. 1987.113:2189-2207.

 not give maximum axial force in the arch. It seems best to make the
distance between nodal points small enough so that the load on the whole
span decides the dimensions of every point in the arch.
Fig. 7 shows the loads and forces that decide the dimensions of the arch
in the span in Figs. 6, 10, 11(e), 12, 15, and 16. Maximum stress in member
74 is 5% greater than the highest stress due to load on the whole bridge. It
Downloaded from ascelibrary.org by Universidad De Sevilla on 05/11/15. Copyright ASCE. For personal use only; all rights reserved.

is important to note that in this span the high stresses due to partial load are
due mainly to local curvature and not to relaxation of hangers.
The optimal number of hangers must be considered for each design. The
number of hangers in bridges discussed in previous papers (Tveit, et al.
1978; Tveit 1980a) and in the Bolstadstraumen bridge (see Fig. 2) (Tveit
1966) are believed to be reasonable choices. Many hangers will increase
the cost of labor in production, erection, and adjustment of hangers. This
increase will, however, be moderate since all the hangers have the same
cross section and hanger details are all alike. These repetitions tend to
reduce labor costs. Many hangers also mean that one can use small
hangers and light equipment for mounting and adjusting them.
An increase in the number of hangers will give lighter spans because it
gives less bending in the chords and less local shear force at the lower end
of hangers. Many hangers can also allow the use of single-rolled steel
sections for arches. Many hangers can, however, make the design around
the lower end of hangers more difficult due to crowding. More hangers will
usually allow a lighter temporary lower chord for erecting the span. For the
bridge in Fig. 4, the 3.9-m (12 ft 10 in.-) distance between the nodes in the
central part of the span was chosen so that there would be a hanger at each
transverse beam in the temporary lower chord (Tveit 1980a).
The use of many hangers makes it easy to replace defective ones without
interruption of traffic, since each hanger is light and the temporary removal
of each one causes less extra stress. If there are many hangers, there is a
decreased chance that the breaking of one or more hangers caused by a
vehicle will have catastrophic effects. The static effect of the breaking of
hangers is treated in other papers (Tveit 1959, 1966). It can be calculated
by ordinary computations by the removal of hangers.
Adjacent hangers at the deck are well spaced at the arch. Thus, the
network arch is less sensitive to the breaking of hangers than the usual tied
arch, and the writer is not aware that breaking of hangers is an important
cause of accidents in tied arches.
The hangers nearest to the ends of the arch usually have smaller
maximum forces than the other hangers. This phenomenon can, to some
extent, be counteracted if the distance between the hangers at the end of
the span is increased. This condition occurs naturally because the arch has
extra strength at the lower end of the wind portal.
The hangers should be placed equidistantly along the middle half of the
tie, giving the smallest bending in the tie and even maximum forces in the
hangers. The distance between the nodes in the middle of the tie should be
the same as, or slightly smaller than, the distance between the nodes along
the arch.
In the half of the tie situated nearest to the ends of the arch, hanger
distances should be varied in order to obtain nearly the same maximum
force in all hangers. Long distances between nodal points should be
2196

J. Struct. Eng. 1987.113:2189-2207.

 Downloaded from ascelibrary.org by Universidad De Sevilla on 05/11/15. Copyright ASCE. For personal use only; all rights reserved.

2197

J. Struct. Eng. 1987.113:2189-2207.

ars fa MM m
.f 4&&$ !®s^ is Yg " 12
V p n W la EneliMJ@c? In the !

(MNm)
Downloaded from ascelibrary.org by Universidad De Sevilla on 05/11/15. Copyright ASCE. For personal use only; all rights reserved.

FIG. 8. One Form of Buckling Mode for Bridge Shown in Fig. 4; More Hangers with
Correspondingly Smaller Diameters Are Placed along Chords

avoided, because the greatest bending moment in the tie often occurs at the
longest distance between nodal points.

BUCKLING OF ARCH

With usual arches and wind bracing, the buckling out of the plane of the
arch is more important than the in-plane buckling. Thus, a thorough
examination of in-plane buckling of the arch is usually not necessary.
In-plane buckling in network arches is best examined by advanced
computer programs that account for relaxation of hangers and calculate in
the deflected state, using nonlinear material properties. Due to lack of
space, only a few facts concerning buckling can be mentioned here.
Buckling is easiest to calculate when the whole span carries full load.
The decisive load case is likely to have the concentrated traffic load in the
middle of the bridge. A likely mode of buckling (Tveit 1959, 1966) is shown
in Fig. 8. How to compute the buckling load for modes like this is
explained by Tveit (1973). The arch should also be calculated as an axially
compressed bar elastically supported at specific points. This computation
can be done by a formula suggested by Engesser in 1884 (Gerard 1962).
The lowest buckling load is likely to come from a mode of buckling
combining traits from Fig. 8 and from buckling in elastically supported
bars. The minimum buckling load is, however, not likely to be much
smaller than the smallest buckling load calculated by the two methods.
When slender arches are calculated by FEMOPT (Andersen 1979), the
deflection of the arch between the nodal points can be conceptualized as
that indicated in Fig. 9. The transition from deflection line 2 to deflection
line 3 is sudden and followed by a great increase in stresses. The transition
can occur at loads well under the elastic buckling load.
If sophisticated computer programs are not available, large deflections
of the type shown in Fig. 9 can be prevented by avoiding slenderness
between points of support that can give large augmentation of transverse
bending moments.

FIG. 9. Lines of Deflection between Nodal Points; Numbering In Accordance with

Increasing Live Load

2198

J. Struct. Eng. 1987.113:2189-2207.

 CHOICE OF MATERIALS AND CROSS SECTIONS OF PARTS OF NETWORK
ARCHES

High-strength materials are best utilized in tension. Utilization is also

good in compression if stocky sections and low slenderness can be used.
Utilization is less efficient in bending. From these comments, it can be seen
that high-strength materials can be well utilized by network arches since
Downloaded from ascelibrary.org by Universidad De Sevilla on 05/11/15. Copyright ASCE. For personal use only; all rights reserved.

tension predominates in hangers and tie. Low slenderness and stocky

sections can be obtained in the arch; axial force predominates in the whole
structure.
Materials and shapes of the various members of network arches will not
be discussed.
Hangers. The hangers should normally be made from spirally wound
galvanized wire ropes. In the two Norwegian network arches (Tveit 1966),
the ropes were made of round wires, but perhaps locked coil ropes should
be used to improve corrosion resistance. Since the hangers can be easily
changed and the wire ropes are in excellent condition after 20 years, the
case for the locked coil ropes is not obvious. Where the wire ropes
intersect, they can be protected by a sheathing of slit open plastic tubes.
The socketed ends of wire rope can be fastened to the chords as shown
in Fig. 10. The ends of hangers are usually the only really labor-intensive
structural parts in a network arch. In the bridge shown in Fig. 4 (Tveit
1980a) and in appendix I of an earlier paper (Tveit 1984b), ends of hangers
only amounted to 2.5% of the steel weight. If the roadway is prestressed,
the lower ends of hangers can be passive anchors for the prestressing
tendons.
Arches. In network arches there is little bending, and the hangers and
wind bracing normally provide good lateral support of the arches. Thus,
there is less reason than usual to spread out the material in the cross
section to obtain a better slenderness ratio and more bending moment
capacity. Slender arches look best. It is advisable to try to make arches
slender since those above the roadway have a tendency to look massive
and domineering. Further, a slender arch rib offers much less area to wind
forces.
Single-rolled steel sections (see Fig. 10 or Tveit 1966) can be used for
arches in narrow road bridges up to about 100 m (330 ft) and for
single-track railway bridges up to approximately 70 m (230 ft). The
single-rolled section gives simple structural details and low fabrication
costs. In the two Norwegian network arches, triangular cross sections
were used (Tveit 1964, 1966). This enhanced the beauty of the bridges but
added to the costs. For spans over 150 m, hollow sections that are
weldable from the inside should be considered. For hollow arches, access
for maintenance may govern the size of the cross section. Increased cross
section of arches generally leads to fewer hangers.
For the bridge in Fig. 4 (Tveit 1984a) and appendix I of Tveit's earlier
work (1983, 1984b), the dimensions of the arch are decided by a load case
where 90% of the maximum stress is due to the normal force and 10% is
due to bending. Since compression is predominant, it would be natural to
make arches from reinforced concrete. In fact, concrete was used through-
out in the Nielsen arch bridges with inclined hangers built mainly in
Sweden between the two world wars (Bretting 1935). Normally, steel

2199

J. Struct. Eng. 1987.113:2189-2207.

Downloaded from ascelibrary.org by Universidad De Sevilla on 05/11/15. Copyright ASCE. For personal use only; all rights reserved.

FIG. 10. Upper and Lower End of Hanger for Single-Track Railway Bridge

should be used in the arches of network arches to keep the scaffolding

costs down. In cases of many equal spans, however, high-strength
concrete should also be considered, especially where the spans could be
prefabricated with pretensioned reinforcement and then floated to the piers
(Tveit 1980b).
Lower Chord. The great longitudinal tensile force in the tie is best
carried by prestressing cables because of the high strength-to-cost ratio.
Since the tie is subjected to both positive and negative bending moments
and to varying axial force, the writer has always found it best to use partial
prestress in the longitudinal direction of the span. In railway bridges,
fatigue in the longitudinal prestressing steel is counteracted by the choice
of a suitable degree of prestressing. For railway bridges, the use of
concrete in the tie will reduce traffic noise, particularly in bridges with rail
and sleepers on ballast.
The dead load of the lower chord adds to the stiffness of the span by
retarding relaxation of hangers. Thus, concrete is the best material for the
bridge deck. Orthotropic steel decks should probably only be used for
some wide bridges with spans well over 200 m (656 ft).
Figs. 11 (a-f) show a variety of edge beams. The edge beam in Fig. 11(a)
is enlarged to act as a traffic barrier to protect the hangers (Tveit, et al.
1978). Stiff edge beams reduce bending in the arch and generally do attract
more bending than slender ones.
2200

J. Struct. Eng. 1987.113:2189-2207.

 toad brigdes: Railway bridges:
Downloaded from ascelibrary.org by Universidad De Sevilla on 05/11/15. Copyright ASCE. For personal use only; all rights reserved.

gu « 1
8!

!
B.
lAmW-V)
1 E.

%,1 ^ ! 1

-1
- ' ••• I

1 !
• . i!
4m
|, 2.6m «'-6") %

FIG. 11. Cross Sections of Various Lower Chords: (a) L = 80 in (262 ft), Built 1963
(Tveit 1980a; Tveit, et al. 1978); (b) L = 200 m (6S6 ft), Calculated 1980 (Tveit 1980a);
(c) L = 100 m (328 ft), Calculated 1984 (Tveit 1984a); (d) L = 65 m (213 ft),
Calculated 1973 (Tveit 1983); (e) L = 67 m (219 ft 10 in.), Calculated 1981; (f) L =
67 m (219 ft 10 in.), Suggested 1983

MS- \ Contour of normal

• * cross-section of '
lower chord
Contourof concrete

L^02_J
Mu^-El '1

p ' T Jr
L»B \J for erectionlower chord /
Temporary Section B-B
Section A-A \ Elastomeric bearing pads *«J

FIG. 12. Ends of Arches in 67-m (219 ft 10 in.-) Single-Track Railway Bridge

The use of many prestressing cables gives smaller anchors and less
bending in the steel plate at the end of the arch (see Fig. 12). Many cables
will also reduce the need for transverse reinforcement to counteract bond
stresses on the surface of the cables due to local change in the cable force.
2201

J. Struct. Eng. 1987.113:2189-2207.

 A comparison of concrete slab roadways spanning up to 17.3 m (57 ft)
with and without transverse beams was carried out by Ronald T. Grimes
(Tveit, et al. 1978). It was found that the simple slab was preferable. For
more than three-lane widths, transverse post-tensioning became preferable
to the usual reinforcement.
For aesthetic reasons, the lower edge of the longitudinal beam must
Downloaded from ascelibrary.org by Universidad De Sevilla on 05/11/15. Copyright ASCE. For personal use only; all rights reserved.

cover the lower edge of the transverse beams. Thus, transverse beams lead
to unnecessary depth in the longitudinal beams.
When the tie is a concrete slab, the transverse moment decides the cross
section between the arches. The longitudinal moments of the slab are much
smaller. When full interaction between deck and arches is taken into
consideration, it has been found (Tveit 1984a) that the longitudinal bending
moment in the roadway and edge beam was roughly proportional to the
local longitudinal stiffness. The longitudinal bending in the lower chord will
rarely influence the dimensions of the edge beam. Minimum dimensions of
the edge beam are more likely to be influenced by the need for room for
longitudinal prestressing cables and by shear stresses around the lower end
of hangers.
Two more reasons for favoring the simple slab deserve a mention: (1)
The slab spreads out the concentrated loads, while the transverse beams
concentrate the loads and make them cause more bending in the edge
beam; and (2) the simple slab is more favorable as far as maintenance is
concerned.

STEEL WEIGHT IN NETWORK ARCHES

During 30 years of construction with network arches, it has been almost

embarrassing to point out the potential savings in steel weights. Partly for
this reason the writer's publications in 1959, 1964, 1966, 1973, 1980, and
1984 have contained many influence lines, tables, and diagrams suitable for
checking the validity of his statements. The steel weights for network

FIG. 13. Steel Weight per Unit of Bridge Deck in Various Types ©f Raod Bridges

J. Struct. Eng. 1987.113:2189-2207.

 Steei weights in railway bridges
Downloaded from ascelibrary.org by Universidad De Sevilla on 05/11/15. Copyright ASCE. For personal use only; all rights reserved.

FIG. 14. Comparison of Steel Weights of Various Types of Railway Bridges

arches in Figs. 13 and 14 were first published by Tveit in earlier works

(Tveit, et al. 1978; Tveit 1980a, respectively). They are based on worked
examples. The weight of reinforcement in tie and roadway is included in
the steel weights. The steel weights of other types of bridges are taken
from Herzog's 1975 paper. The comparison is, however, not quite fair,
because the weights for network arches represent a more conscious effort
to save steel. Further saving in steel can be obtained by making the arches
from concrete (Tveit 1980b) or by using inclined hangers with multiple
intersections where the arch in a through arch comes above the roadway.
The high-rise-span ratio, compatible with a good appearance in the
through arch, and the greatly reduced force in the horizontal chord
contribute considerably to the reduction of steel weight.
The lines in Fig. 14 represent steel weights of single-track railway
bridges (Herzog 1975). The steel weights for double-track railway bridges
are supposed to be twice as high.

ECONOMIC ERECTION
Network arches are inexpensive as far as material is concerned. If
efficient methods of erection could be used, network arches would make
most economical bridges. The most promising methods of erection utilize
the fact that the arch and hangers supplemented by a light temporary lower
chord can carry the deck while it is being cast (Tveit 1966, 1980a; Tveit, et
al. 1978). (See Figs. 15-16.) The same temporary lower chord might be
used for spans of varying length and width. During construction, most of
the tension in the lower chord is carried by a longitudinal prestressing
cable. This cable is tensioned from time to time during casting. The edge
beam must be cast before the roadway. Maximum wind forces during the
casting of the edge beam will often be important when designing the
longitudinal beam in the temporary lower chord. During erection and in the
final adjustment of hanger lengths, hanger forces can be calculated from
2203

J. Struct. Eng. 1987.113:2189-2207.

Downloaded from ascelibrary.org by Universidad De Sevilla on 05/11/15. Copyright ASCE. For personal use only; all rights reserved.

FIG. 15. Partial Plan of Temporary Lower Chord for Single-Track Railway Bridge
Shown in Fig. 11(e)

FIG. 16. Cross Section of Permanent and Temporary Lower Chord for Single-Track
Railway Bridge Shown in Fig. 11(e)

the deflection due to a transverse load on the steel hanger. The deflection
can be found by sighting along the rope above the transverse load and
measuring the distance to a point below the transverse load. The measure-
ments are easily made, except for the shortest hangers, which are usually
not fully utilized. Thus, exact measurements of their forces are not such a
necessity. In a paper by Takagai, et al. (1973), hanger forces were found by
measuring the eigenfrequencies of hangers.
In cold climates, the ice of a river or lake can be used for erecting or
moving the temporary structure (Tveit 1966, 1981). The ice can be
reinforced by wood or ribbed bars, which are placed on the ice (Cedervall
and Fransson 1979). Then, the thickness of the ice can be increased in
steps by pumping water onto the ice or by spraying water in cold air above
the ice. This thick ice can be used for casting and erecting the network tied
arch between the piers (Tveit 1984a-c).

SUITABLE SITES FOR NETWORK ARCHES

The low material and erection costs will make the network arch an
economic solution for many building sites, especially where spans of
80-300 m (250-700 ft) are required. The network arch is relatively light, has

J. Struct. Eng. 1987.113:2189-2207.

 vertical reactions, and is not sensitive to uneven settlements of the
foundations. Thus, the network arch can be extra competitive where soil
conditions are difficult.
The deck of a network arch is usually a simple slab. This makes ramps
in flat terrain short when a flow of traffic must be lifted to pass navigable
waters or other traffic. Generally, the network arch is most likely to be
built when local conditions lead to acceptance of structural members
Downloaded from ascelibrary.org by Universidad De Sevilla on 05/11/15. Copyright ASCE. For personal use only; all rights reserved.

above the lane. This seems to occur more often for railway bridges than for
road bridges.

ADDITIONAL RESEARCH

A consistent application of the ideas put forward in this paper would

make a considerable contribution to research and would perhaps lead to
more research into network arches, especially into the buckling problem.
Research into the dynamic behavior of network arches is also needed.
Takagai and coworkers' 1970 paper represents a beginning. Incomplete
observations of vibrations in the two Norwegian network arches can be
found in research by Tveit (1966). By comparing eigenfrequencies of
vibrations of tied arches with vertical hangers to vibrations of network
arches (Tveit 1981), the writer has concluded that wind-induced oscilla-
tions are not likely to be a problem in network arches. Further research in
this field would, however, be most welcome.

CONCLUSIONS

In network tied arches there is very little bending in the arch and the tie.
The hangers give the arch good lateral support. Thus, the buckling stress
of the arch is high. The axial forces in the upper and lower chords are
determined by the loads and by the rise of arch compatible with good
looks. The hangers make a very light web. Since all members in a network
arch efficiently carry forces that cannot be avoided in any simply sup-
ported beam, the writer concludes that the network tied arch is probably
the lightest possible, simply supported span. If suitable methods of
erection are used, the network arch must also be an economic solution,
because the building costs include that of materials, plus the expense of
forming and handling these materials.
The ends of hangers are the only labor-intensive structural members, but
they lend themselves to labor saving through repetition of details. Corro-
sion protection ought to be cheap, because the areas to be protected are
small and easily accessible.
Promising methods of erection utilize the fact that the arch and hangers,
supplemented by a temporary lower chord, can carry the roadway while it
is being cast. Light members and small scaffolding costs contribute to
keeping erection costs down.
The network arch is insensitive to uneven settlements of the founda-
tions, and the roadway is usually a simple slab. This property makes the
network arch especially competitive where soil conditions are difficult and
since it gives the shortest possible ramps when a flow of traffic must be
lifted to pass navigable waters.
2205

J. Struct. Eng. 1987.113:2189-2207.

 APPENDIX I. REFERENCES

Andersen, G. (1979). "Brugervejledning for FEMOPT. Program for optimenng af

baerende konstruktioner." ("Users' manual for FEMOPT. Program for the
optimizing of load-carrying structures," in Danish.) Note no. 7902, Inst. Bldg.
Tech. Struct. Engrg., Aalborg Univ. Centre, Aalborg, Denmark.
Berger, O. (1955). "Beurteilung des Tragvermogens einer grossen Eisen-
Downloaded from ascelibrary.org by Universidad De Sevilla on 05/11/15. Copyright ASCE. For personal use only; all rights reserved.

bahnbriicke auf Grand von Spannungsmessungen und Berechnung als

hochgradig (38 fach) statisch unbestimmtes System." ("Evaluation of the
load-carrying capacity of a large railway bridge by means of strain
measurement and calculation as a 38-times redundant system," in German.)
Bauplanung und Bautechnik, 9(12), Berlin, D.D.R., 526-536.
Bretting, A. E. (1935). "Inclined hangers impart slenderness to ribs of 470-ft
concrete arch." Engrg. News-Rec, 114(17), 577-580.
Cedervall, K., and Fransson, L. (1979). "The effect of reinforcement on the
carrying capacity of an ice cover." Publ. 79:1, Avd. for Konstrukttionsteknik,
H0gskolan i Lulea, Sweden.
Gerard, G. (1962). Introduction to structural stability theory. 1st ed.,
McGraw-Hill, New York, N.Y., 89-94.
Herzog, M. (1975). "Stahlgewichte moderner Eisenbahn- und Strassenbriicken."
("Steel weights of modern rail and road bridges," in German.) Der Stahlbau,
44(9), Berlin, B.R.D., 280-284.
Kikuno, M. (1973). "Ohnura Bay Bridge (Japan)." Acier-Stahl-Steel, 41(4),
Brussels, Belgium, 168-171.
Snelling, J. (1980). "Two unusual features of the Merivale Rail Bridge."
Presented at 11th IABSE Congress, held at Vienna, Austria, Final rep.,
IABSE, ETH-H6nggerberg, CH-8039 Zurich, Switzerland, 347-349.
Stein, P., and Wild, H. (1965). "Das Bogentragwerk der Fehmarnsundbriicke."
Der Stahlbau, 34(6), Berlin, B.R.D., 171-186.
Takagai, S., et al. (1970). "Report on experimental research on Aki-Ohashi."
Trans. JSCE, 2, part 1, 104-110.
Tveit, P. (1959). "Bogebruer med skra krysstilte hengjestenger." ("Arch bridges
with inclined intersecting hangers," in Norwegian.) Thesis presented to the
Tech. Univ. of Norway, at Trondheim, Norway, in partial fulfillment of the
requirements for the degree of Licensed Technician.
Tveit, P. (1964). "Nettverkbogar, ein ny brutype." ("Network arches, a new
type of bridges," in Norwegian.) 12(5), Bygg, Oslo, 105-113.
Tveit, P. (1966). "Design of network arches." Struct. Eng., 44(7), London,
England, 247-259.
Tveit, P. (1972). Discussion of "The design of inclined tied arch railway bridges
over the M56" by K. I. Majid, J. E. Spindel, and M. S. Williams, Proc. Inst.
Civ. Engrs., Paper 7392, 52, London, England, 181-184.
Tveit, P. (1973). "Network arch in double track railway bridge." Presented at
Nordic Resear. Days for Steel Struct., held in Oslo, Norway, Preprint V/13.
(The publication is identical to "Report 7205." Ren og Anvendt Mekanik,
Danmarks Ingeni0rakademi, Bygningsafdelingen, Aalborg, Denmark.)
Tveit, P. (1980a). "Network arches." Handout for poster session of IABSE's
11th Congress, held at Vienna, Austria, 45 pp. Publ. by Inst. Bldg. Tech.
Struct. Engrg., Aalborg Univ. Centre, Aalborg, Denmark.
Tveit, P. (1980b). "Network arches." 11th IABSE Congress, held at Vienna,
Austria, Final Rep., IABSE, ETH-H6nggerberg, CH-8039, Zurich, Switzer-
land, 817-818.
Tveit, P. (1981). "25 ar med nettverkbogar." ("25 years with network arches.")
Guest lect. at Tech. Univ. of Norway, printed at Inst, of Bldg. Tech. Struct.
Engrg., Aalborg Univ. Centre, Aalborg, Denmark.
Tveit, P. (1983). "Economic design of network arches." Rep. No. 8304, Inst, of
Bldg. Tech. Struct. Enger., Aalborg Univ. Centre, Aalborg, Denmark, 22 pp.
Tveit, P. (1984a). "A network arch with four sets of hangers designed to be cast
and erected on ice floating between piers." Handout for poster session of

2206

J. Struct. Eng. 1987.113:2189-2207.

 lABSE's 12th Congress, held at Vancouver, Canada, Publ. by Inst, of Bldg.
Tech. Struct. Engrg., Aalborg Univ. Centre, Aalborg, Denmark, 106 pp.
Tveit, P. (1984b). "Economic design of network arches." Handout for poster
session of lABSE's 12th Congress, held at Vancouver, Canada, publ. by Inst.
Bldg. Tech. Struct. Engrg., Aalborg Univ. Centre, Aalborg, Denmark, 32 pp.
Tveit, P. (1984c). "Building network arches on reinforced ice between piers."
Final Rep., lABSE's 12th Congress, held at Vancouver, Canada, publ. by
Downloaded from ascelibrary.org by Universidad De Sevilla on 05/11/15. Copyright ASCE. For personal use only; all rights reserved.

IABSE, ETH-Hongerberg, CH-8039, Zurich, Switzerland, p. 1130.

Tveit, P., et al. (1978). Network arches. 1st ed., 1, Civ. Engrg. Dept., Univ. of
Houston, Tex., 93 pp. (Revised edition reprinted at Inst, of Bldg. Techn.
Struct. Engrg., Aalborg Univ. Centre, Aalborg, Denmark, 1980.)

2207

J. Struct. Eng. 1987.113:2189-2207.