BUCKLING OF
NETWORK ARCH
BRIDGES
FRANK SCHANACK
BIOGRAPHY
Dr. Schanack is professor for
bridge and structural
engineering at the Universidad
Austral of Chile. Prior to his
appointment at UACh, he
completed his doctorate degree
at University of Cantabria in
Spain and served as a bridge
engineer at APIA XXI, Spain
and Highway Construction
Office of Saxony, Germany. He
completed his engineering
degree at Dresden University of
Technology, Germany. His
primary areas of research are
static and dynamic structural
behavior of steel bridges,
network arches and theory and
analysis of structures.
Contact: Instituto de Obras
Civiles, Universidad Austral de
Chile, General Lagos 2086,
Valdivia, Chile
schanack at yahoo.de
SUMMARY
Recently, many new network
arch bridges have been built.
When introducing new
structures it is important to
understand well its structural
behavior in order to assure the
usual safety and quality level. In
arches, the inplane buckling
behavior is of particular interest,
whose general solution for
network arches was developed
by the author and is presented in
this paper.
Based on a general parametric
study by means of 3D nonlinear
FEM analysis, the critical elastic
inplane buckling load and
mode shape are determined. It is
pointed out that the network
arch shows a nonlinear load
displacement behavior and a
geometrically nonlinear load
step analysis is necessary.
Based on the analysis of the
results a concept is found,
which allows considering the in
plane buckling problem as that
of a curved compression strut
with radial elastic support.
Following this concept the
author derives a simple analytic
design formula for the
calculation of the critical
buckling load for inplane
buckling of network arch
bridges. This formula has an
error of less than 5%.
From the results of the
presented study it may also be
concluded that the hanger net
constitutes a very effective
buckling support of the arch in
its plane. The inplane buckling
safety is much higher than the
one out of plane, which favors
unusual arch cross sections as
for example rolled Hbeams.
No Photo Available
Page 1 of 10
Fig. 1. Mangamahu network arch bridge in New
Zealand, span 85 m, 2008, courtesy of Michael
Chan.
INPLANE ARCH BUCKLING OF NETWORK ARCH
BRIDGES
Paper for the 2009 World Steel Bridge Symposium by
Frank Schanack
Instituto de Obras Civiles, Universidad Austral de Chile, General Lagos 2086, Valdivia, Chile
Network arches are tied arch bridges with inclined hangers, which have multiple intersections. In the last five
years there has been built a significant number of network arch bridges, maybe initiating a turning away from
the traditional tied arches with vertical hangers. When introducing new structures it is important to understand
well its structural behavior in order to assure the usual safety and quality level. In arches, the buckling
behavior is of particular interest. The existent solution does not explain the inplane buckling of network
arches to a satisfying extent, limiting its application range to a special case. A general parametric study of the
inplane buckling leads to a concept, which allows considering the problem as a curved compression strut
with radial elastic support. Then, a simple analytical formula is derived that predicts the critical buckling load
and buckling mode for inplane buckling of network arch bridges with an error of less than 5%.
Introduction
Until several years ago only a few network arch bridges had been built, although it was its development
started in the 1950s. But, as it can be seen in Table 1, in the last 5 years many network arches have been built
in the whole world.
Name Country Span Year
Bechyně Bridge Czech Republic 41,0 m (134,5 ft) 2004
Bridge over the Deba River Spain 110,0 m (360,9 ft) 2006
Providence River Bridge USA 121,2 m (397,6 ft) 2007
Blennerhassett Bridge USA 267,7 m (878,3 ft) 2008
Mangamahu Bridge New Zealand 85,0 m (278,9 ft) 2008
Rosenbachtal Bridge Germany 90,0 m (295,3 ft) 2009
Bridge over the Carbón River Peru 120,0 m (393,7 ft) under construction
Bridge over the river Ob in Novosibirsk Russia 360,0 m (1181,1 ft) planned
Table 1. International examples of new built and planned network arch bridges.
Network arches are defined by its inventor, Per
Tveit, as tied arch bridges with inclined
hangers, that cross each other at least twice [7]
(Figure 1). There are principally two reasons
for today’s turning away from the traditional
tied arch with vertical hangers towards the
network arch. First, the reduction of maximum
bending moments in arch and tie by a ratio of
≈1:10, which promises important material
savings [5]. And second, the nowadays
common use of powerful FEM software that
allow the analysis of the highly hyperstatic
structure within seconds.
Page 2 of 10
The structural behavior of network arches is very different to that of conventional tied arches. Not only the
bending moments are reduced, but also, for example, the deformations and the dynamic properties are
changed [4]. Among these interesting structural effects there is also a change of the inplane buckling
behavior. Consequently, it is desirable to conduct a general study of the inplane buckling of network arches.
This study was conducted by the author, who found the correct critical buckling mode shapes and derived a
formula for the calculation of the buckling load. In this paper, the study and its results are described.
General Procedure
When studying the arch stability, two structural characteristics need to be considered. First, the load level that
causes the system’s instability. This critical load may be expressed as the sum of the exterior loads or, much
more practical, as the axial arch force.
Second, it is necessary to determine the buckling mode shape, i.e. the eigenmode shape of the system under
buckling load.
The whole study is limited to ideal elastic material properties; because the arch’s crosssection will always be
assessed in such a way that plasticization starts at ultimate limit state. Consequently the structural behavior
under design loads is elastic.
Also, this study refers to network arches that do not show hanger relaxation in any load situation.
Determination of the Buckling Load
Many engineers might be inclined to use linearized eigenvalue analysis in order to determine the buckling
load and mode shape of network arches. But, the network arch shows nonlinear loaddisplacement behavior.
Consequently, this procedure cannot be applied.
The correct analysis type is the geometrically nonlinear load step analysis. In such analysis the elastic
buckling load is calculate by means of a series of subsequent geometrically nonlinear analyses, in
which the load is increased incrementally. The nonlinear calculation will not find a solution of the
equilibrium equations any more, if the load level reaches the bifurcation point. With an appropriate
iteration it is possible to determine the load level that corresponds to the elastic buckling load.
Determination of the Critical Mode Shape
The critical elastic eigenmode shape is defined as the vector of deformation, which belongs to the smallest
solution the eigenvalue problem of a structure. However, the eigenvalue analysis is only applicable, if the
loaddisplacement behavior between the base state and the bifurcation point is linear.
In network arches, this is approximately true only for very small load increments. Hence the last stable load
increment before the bifurcation point, as in section 2.1, must be used as the base state.
Another possibility is the determination of the system’s vibration mode shapes under buckling load. Under
this load, the stiffness of the systems for deformations according to the buckling mode shape converges to
zero. Hence the frequency of the vibration mode converges to zero, too. Consequently, under buckling load
lever, the fundamental vibration mode shape is identical to the critical eigenmode. This second method was
applied for the described study.
Existing Method
Based on model test and
theoretical considerations Per
Tveit developed the
assumption of the buckling
mode shape shown in Figure
2 [8]. The distances of the
Fig. 2. Assumption of the inplane buckling mode shape of network
arches [8] (special case).
Page 3 of 10
points of inflection along the chords are established according to virtual triangles within the hanger net. By
means of the geometry of this buckling mode shape, the flexural stiffness of the chords and the elongation
stiffness of the hangers, Per Tveit derived an analytical formula for the calculation of the inplane arch
buckling load. This was the only analytical method known to the author.
But, the buckling mode shape of Figure 2 is merely a special case, which only occurs in certain combinations
of bridge geometry and crosssections. Consequently, the method proposed in [8] cannot be recommended for
general application.
Parametric Study
In order to determine their influence on the critical buckling load and mode shape, several parameters are
studied. These are:
1. Type of load
2. Hanger number
3. Archhanger angle
4. Arch moment of inertia
5. Tie moment of inertia
As basis of the parametric study a 100 m (328.1 ft) railway bride is chosen, which is described in detail in [1].
For traffic load the European Load model UIC71 is used (Figure 3).
Influence of the Type of Load
Different load distributions may lead to different buckling mode shapes and corresponding critical buckling
loads. In the following, 4 different load cases are studied, which are representative for the possible bridges
loads.
Fig. 3. European train load model UIC71.
Page 4 of 10
1. Selfweight and permanent load.
2. Selfweight, permanent load, and
traffic load on the whole span.
3. Selfweight, permanent load, and
traffic load on half of the span.
4. Selfweight, permanent load, and
single load at midspan.
For load case number 1 the load factor
is applied to selfweight and permanent
load. For the rest of the load cases only
the traffic load is increased. In this way
the influence of the different load
positions on the buckling load is more
evident.
The first two load cases (Figure 4) lead
to very similar buckling mode shapes
and an almost identical buckling load of
58,600 kN (260,600 kip). Due to the
higher initial load of load case 2, the
load factor α
crit
of 3.100 is much smaller
than in load case 1 (α
crit
= 4.639). The
arch buckling shape consists of a
symmetric wave with 9 wave crests. It is
not related to the bending moment
distribution under these loads. Since
arch and deck are connected to each
other by the hangers, the deck also
shows a wavelike deformation. Hence,
the amplitudes are so small that this
deformation is negligible.
Under asymmetric load, like in load
case 3, the axial force in the arch is not
constant. At buckling load level it
increases continuously from 45,500 kN
(202,400 kip) to 66,000 kN (293,600
kip). As it can be seen in Figure 4, the
buckling mode shape is not symmetric
either and shows bigger deformations in
the area of the load. In this area, the
axial force in the arch is 62,000 kN
(275,800 kip), which is bigger than in
load cases 1 and 2. The increment of the
buckling load is possible, because the
wave length of the buckling mode shape
and therefore the buckling length is
shorter (Figure 4). In terms of buckling
safety, load case 3 is note decisive,
because the load factor 4.266 is bigger
Fig. 4. Critical load N
cr
and buckling mode shape for different
types of load (load cases 1 to 4).
Fig. 5. Critical load N
cr
and buckling mode shape for a change
of the number of hangers, load case 1.
Page 5 of 10
than the one of load case 2.
The single load of load case 4 at mid
span causes local bending moments in
the arch, which are positive where the
hangers transmit the single load to the
arch. These bending moments change
the buckling mode shape and force the
arch into a wavelike deformation with
less wave crests and therefore longer
buckling length (Figure 4). Although
the buckling load decreases to
32,540 kN (144,700 kip), the load
factor 8,459 is the biggest of all load
cases. Consequently, the single load is
accompanied by the biggest buckling
safety.
According to the results of this section,
the decisive load case for arch buckling
is a uniform load on the whole bridge
length. In order to simplify the
following studies, only the load case 1
will be considered.
Influence of the Hanger
Number
In order to study the influence of the
hanger number on the buckling load,
the example bridge was calculated with
20, 32, 48, and 80 hangers, while the
hangers’ diameter remains constant.
Figure 5 shows the different buckling
loads and buckling mode shapes.
It is may be seen that the buckling load
increases with a higher number of
hangers. This increase is possible,
because the number of wave crests of
the buckling mode shape also increases
with the number of hangers. This leads
to shorter buckling lengths. I propose to
classify the wavelike buckling mode
shapes into orders, according to the
number of wave crests.
Influence of the Archhanger
Angle
The bridge example has a radial hanger
net according to Brunn&Schanack [1]
and [6]. In the radial hanger
arrangement the archhanger angle is
Fig. 6. Critical load N
cr
and buckling mode shape for a change
of the hangerarch angle, load case 1.
Fig. 7. Critical load N
cr
and buckling mode shape for a change
of the moment of inertia of the arch, load case 1.
Page 6 of 10
practically the same for all hangers. Usually, an angle between 45º and 65º is chosen. These limit values were
also used for the present parametric study. Figure 6 shows the buckling loads and mode shapes for different
cross angles.
The buckling load of the arch with an archhanger angle of 45º is about 60% smaller than for an archhanger
angle of 65º. Again, the change of the order of the buckling mode shape, i.e. the decrease and increase of the
number of wave crests, makes this difference possible.
Influence of the Arch´s Moment of Inertia
According to the buckling theory of Euler the buckling load is linearly proportional to the bending stiffness.
The influence of the bending stiffness on the buckling load is studied in this case by means of a variation of
the moment of inertia of the arch.
Figure 7 shows the buckling loads and modes shapes of the bridge example for different moments of inertia
of the arch. Although the buckling load increases and decreases corresponding to the moment of inertia, the
relation is not linear. This result is still compatible with Euler’s buckling theory, because at the same time the
buckling mode shapes change as well. But, in contrast to the rest of the parameter variations, the buckling
load is now increasing, although the order of the buckling mode is decreasing.
Influence of the Tie´s Moment of Inertia
In the preceding studies deformations
of the tie were detected in the
buckling mode shapes, which are
negligibly small. But, since the tie is
an elastic support of the hangers and
therefore of the arch, a change of the
tie’s bending stiffness shall be looked
at. Figure 8 compares the buckling
loads and mode shapes of different
moments of inertia of the deck.
Again, the buckling load and the
mode shape order increase with a
higher moment of inertia. However,
there is a certain moment of inertia of
the deck, from which it may be
considered totally stiff compared to
hangers and arch. A further increase
of the moment of inertia does not
lead to any further increase of the
buckling load.
The bending stiffness of the deck of the bridge example is of usual dimension and bigger than this limit
stiffness. Therefore, in the preceding studies almost no deck deformation was detected.
Fig. 8. Critical load N
cr
and buckling mode shape for a change of
the moment of inertia of the deck, load case 1.
Page 7 of 10
Conclusions of the Parametric Study
Generally it may be concluded that the bending stiffness of the arch has the biggest influence on the buckling
load. Less important are the archhanger angle and the hanger number. The bending stiffness of the tie has the
smallest influence and is even negligible, up to certain value.
The parametric studies lead to the following more specific conclusions about the stability behavior of network
arches in its plane.
1. The decisive load case is full load on the whole bridge.
For different load distributions the buckling mode shapes and loads are similar. Consequently that load
distribution that in its initial state is closed to the buckling load, has the smallest buckling safety. This load
distribution is full load on the whole bridge.
2. The deck may be considered as bending stiff.
A variation of the bending stiffness within usual dimensions does not lead to any change of the buckling
behavior.
3. The critical mode shape is a wavelike deformation of variable order.
Under decisive load almost the same critical mode shape is obtained for all cases studied. It is a wavelike
deformation of the arch, whose amplitudes increase towards midspan. The only variable is the number of
wave crests, according to which it is possible to classify the buckling mode shapes into orders.
The order of a buckling mode shape increases with the stiffness of the hanger net, i.e. with higher hanger
number, a bigger archhanger angle, and a bigger moment of inertia of the deck. But, it decreases with a
bigger moment of inertia of the arch.
Design Formula
The development of the analysis of Per Tveit (section 3) is based on the assumption of a buckling mode shape
(Figure 3) that is very similar to the ones calculated in section 4. However, it is assumed that this mode shape
only changes with the archhanger angle. As shown in the preceding sections, this is not correct. Therefore, I
felt motivated to derive a new analytical method for the determination of the buckling load and buckling
mode shape of the network arch in its plane.
According to the conclusions of section 4.6, the following simplifications can be made:
1. The decisive load is full load on the whole bridge, which
causes a constant axial compressive force with negligibly
small bending moments in the arch [4].
2. The deck can be assumed to be bending stiff and
consequently all hangers to be fixed at their lower end.
Consequently, the calculation problem is reduced to a curved
compression bar, with constant axial force and radial elastic
support in its plane.
The solution of this problem for straight bars with elastic
foundation is known (Figure 9), e.g. [2]. Furthermore, the
stability behavior of this system is very similar to the one
observed in section 4. According to the stiffness relation of
foundation and bar, the mode shape order is bigger or smaller.
The formula for the straight bar is the following:
Fig. 9. Straight and curved
compression bar on elastic
foundation.
Page 8 of 10


.

\

⋅

.

\

⋅
+ ⋅

.

\
 ⋅
= k
n
l
EI
l
n
P
cr
2 2
min
π
π
(1)
where
P
cr
Critical buckling load,
n Buckling mode shape order (number of wave crests),
l Length of bar,
EI Bending stiffness, and
k Foundation stiffness per unit length.
The left term of equation (1) describes Euler’s buckling load of a pinned bar, which is smallest for first order
buckling mode shape. The right term describes the increment of the buckling load due to the elastic
foundation. The elastic foundation has its biggest effect for first order buckling mode shape, because then the
amplitudes are biggest.
With increasing mode shape order the left term is increasing while the right term is decreasing. The critical
buckling load of the system is obtained for that mode shape order, which results in the smallest sum of both
terms.
In order to transfer formula (1) to the network arch, the left term has to be transformed into equivalent for
circular arches and the right term into its equivalent for radial foundation. The radial foundation is constituted
by the hanger net, whose stiffness depends on the cross angle, the hanger number and the elongation stiffness
each hanger.
According to this concept I derived the formula (2) for the calculation of the critical elastic inplane buckling
load of network arches, under load on the whole bridge. The derivation is described in detail in [4].





.

\

−


.

\
 ⋅
⋅ ⋅
⋅ ⋅
+
(
(
¸
(
¸
−


.

\
 ⋅
=
1
sin
2
1 min
2
0
3
0
2
2
0
α
π
α
α
α
π
n
f
R n EA
R
EI
n
N
p p
y
cr
(2)
where
f
s
f
R
⋅
+
=
2
4
2
2
(3)

.

\

⋅
=
R
s
2
arcsin
0
α (4)
f Rise of the arch,
EA
p
Elongation stiffness of one hanger,
EI
y
Bending stiffness of the arch,
s Span,
α Archhanger angle,
Page 9 of 10
n Mode shape order, and
n
p
Total number of hangers per arch.
The result of formula (2) depends on the mode
shape order n, such as shows, for example,
Figure 10. For certain value n the smallest
buckling load is obtained. This is the critical
elastic buckling load and the corresponding
number n is the mode shape order of the critical
mode shape.
Evaluation of the Design Formula
On the one hand the formula (2) neglects the
bending stiffness of the deck, an assumption that is on unsafe side, but whose error is very small considering
usual deck dimensions. On the other hand, the assumption of a twohinged arch is on the safe side. Therefore,
the buckling load as from formula (2) should be a bit smaller than the real buckling load.
In order to evaluate the correctness and precision of the new formula (2), it was compared with the results
from the FEMcalculations of section 2. For example, for the 12 different calculations presented in the
parametric study, in which the parameters are varied beyond the practical limits. Here, the mean deviation is
3.5%, with a peak value of 6.6%. In only three cases the result of the formula is on unsafe side.
Furthermore, the formula was checked in some network arch bridge projects, for example, in the
Brandangersound Bridge, Norway [3] with a span of 220 m. In this bridge the numeric analysis results in a
buckling load of 24,300 kN (108,100 kip) while formula (2) gives 24,260 kN (107900 kip). Another example
is the Åkviksound Bridge, Norway [9] with a span of 135 m, where the FEManalysis results in a buckling
load of 34,000 kN (151,200 kip) while formula (2) gives a value of 34,100 kN (151,700 kip).
These results confirm the correctness of the described concept of the stability behavior in the arch plane. The
precision of formula (2) is best for network arch bridges with usual dimensions. In this case its error is much
smaller than 5%.
Summary
The network arch is increasingly used for bridge construction. Therefore it is even more important to
understand properly its structural behavior. One part of it is the inplane stability behavior of the arch. The
existing solution of this problem was limited to a special case; consequently a new study was required.
Network arches show nonlinear loaddisplacement behavior. Therefore, the correct method for the
determination of the critical buckling load is a geometrically nonlinear loaddisplacement analysis with
iterative load incrementation. The critical mode shape is obtained from eigenvalue analysis of the last stable
load increment.
The decisive load case for inplane buckling is full load on the whole bridge, because in this case the buckling
load factor is smallest. The buckling load increases with a higher hanger number, with steeper hangers and
with a bigger moment of inertia of the arch. In network arch bridges with usual dimensions the deck can be
considered as bending stiff in terms of the arch stability.
The critical mode shape is characterized by a wavelike deformation of the arch of variable order and with
increasing amplitudes towards midspan.
Conceptually spoken, the inplane buckling of network arches can be considered as the buckling of a curved
bar under constant axial force and with radial elastic foundation. According to this concept, formula (2) was
derived that serves as a simple analytical tool for the determination of the critical buckling load as well as the
Fig. 10. Example for buckling load vs. mode shape
order as obtained with formula (2).
Page 10 of 10
critical mode shape. An evaluation of the formula confirms the correctness of the concept and shows that the
formula has an error of less than 5%.
Acknowledgements
Nonlinear finite elements analyses were performed with SOFiSTiK Analysis Software under academic
license.
References
[1] Brunn, B., Schanack, F.: Calculation of a Double Track Railway Network Arch Bridge applying the
European Standards. Graduation thesis, TUDresden, 2003.
[2] Brush, D., O., Almroth, B., O.: Buckling of Bars, Plates and Shells. MCGrawHill Book Company, ISBN
0070085935, 1975.
[3] Larssen, R., M., AasJakobsen, K.: Brandangersundet Bru – Verdens slankeste. Nyheter om Stalbyggnad
3/2006, ISSN 14049414.
[4] Schanack, F.: Puentes en Arco Tipo Network – Network Arch Bridges. Asociación Científicotécnica del
Hormigón Estructural, ISBN 9788489670, 2009.
[5] Schanack, F., Brunn, B.: Analysis of the structural performance of network arches. The Indian Concrete
Journal, Vol. 83, Nr. 1 (2009), ACC limited, pp. 713.
[6] Schanack, F., Brunn, B.: Netzgenerierung von Netzwerkbogenbrücken (Generation of the hanger
arrangement of network arches). Stahlbau 78 (2009).
[7] Tveit P.: Design of Network Arches. Structural Engineer 44(7) (1966), pp. 247259.
[8] Tveit P.: Network Arches for Railway Bridges. Report 7205 (1973). Danish Academy of Engineers,
Aalborg, Denmark.
[9] Tveit P.: Comparison of Steel Weights in Narrow Arch Bridges with Medium Spans. Stahlbau 68 (1999).
pp. 753757.