INTERNATIONAL CENTRE FOR MECHANICAL SCIENCES

COURSES AND LECI'URES- No. 316

ADVANCED PROBLEMS
IN BRIDGE CONSTRUCTION

EDITED BY

[Link]
UNIVERSITY OF VENICE
[Link]
UNIVERSITY OF ROME

SPRINGER-VERLAG WIEN GMBH

 Le spese di stampa di questo volume sono in parte coperte da
contributi del Consiglio Nazionale delle Ricerche.

This volume contains 243 illustrations.

This work is subject to copyright.

All rights are reserved,
whether the whole or part of the material is concerned
specifically those of translation, reprinting, re-use of illustrations,
broadcasting, reproduction by photocopying machine
or similar means, and storage in data banks.
© 1991 by Springer-Verlag Wien
Originally published by Springer Verlag Wien-New York in 1991

In order to make this volume available as economically and as

rapidly as possible the authors' typescripts have been
reproduced in their original forms. This method unfortunately
has its typographical limitations but it is hoped that they in no
way distract the reader.

ISBN 978-3-211-82318-7 ISBN 978-3-7091-2614-1 (eBook)

DOI 10.10071978-3-7091-2614-1
 PREFACE

The volwne collects the lecture notes of a course held at the Department of Structural
Engineering and Geotechnics of the International Centre for Mechanical Sciences dealing
with the most and modern topical problems of bridge design .

.The topics presented allow to tackle both theoretical-analytical as well as technical-

constructive aspects of the design problem, pointing out how in the case of bridges,
specifically for long span bridges, the two aspects are absolutely inseparable.

In modern bridges, reasons of technical and economic feasibility oblige an extreme

parceling of the construction process, with the consequent need to revise, with to respect to
the past, both design concepts as well as the theoretical apparatus of analysis that governs it.

All this can clearly be derived from reading the present volwne, in which the different
contributions stress theoretical and technical questions of particular interest and topicality,
without claiming to approach them systematically, but offering clear procedural rules and
trend indications.

With reference to the theoretical approach, some topics of particular importance are
reviewed, such as the possibility of using limit analysis, the simplification of the design
process for bridges, durability, and computer aided design.
 These topics are presented with the aim of providing design tools and not specific
analytical verification procedures, for which it is necessary to consult specialized texts.

For what concerns the bridge typologies and the corresponding construction
problems, the emphasis is mostly on the ones still in an evolutionary phase, that is long span
suspended/stayed bridges and cantilever built bridges with prefabricated segments.

Giuseppe Creazza
Michele Mele
 CONTENTS

Page
Preface

Bridges: some problems and some solutions

[Link] ........................................................................... ................... !

New trends and developments in bridge construction

R. Walther ........................................................................... ................. 1 1

Cantilever built bridges with prefabricated segments

R. Lacroix ............................................................................ ................. 2 7

Composite bridges: new developments in Europe

[Link] ........................................................................... ................... 55

Structural systems for cable suspended bridges

N. J. Gimsing ........................................................................... ............ 6 7

Some basic problems in the design of long span cable stayed bridges
[Link] ........................................................................... ............ 9 1

On the durability of reinforced and prestressed concrete structures

M. Mele, E. Siviero ............................................................................ ... 1 2 1

Cracking due to thermal effects on bridges

H. Falkner ............................................................................ .............. 16 1

Structural effects of time dependent behaviour of concrete

[Link] ........................................................................... ................... 17 5

Technological factors in the design of foundation piles

C. Viggiani ............................................................................ ............. 2 2 3

Computer based structural analysis of bridges

S. Odorizzi ....... ... , ...... ................................._........................................ 2 57
 BRIDGES
SOME PROBLEMS
AND SOME SOLUTIONS

G. Macchi
University of Pavia, Pavia, Italy

This introduction to the course is not emphasizing such

applications of the modern technique which are at the frontier of the
present possibilities as the bridge of the Messina strait or the
Gibraltar strait, fascinating problems certainly reserved to very
exclusive groups of engineers.
Nevertheless, a vaste field of activities for the engineers is
appearing in new roles: both the repair works and the reconstruction
of new bridges in the special environment formed in the last decades
require a substantial change of the disciplinar bases for our
profession.

1. A NEW ROLE FOR THE ENGINEERS: DIAGNOSIS AND STRENGHTENING

The new roles do not require a further sofistication of statics

and dynamics of the great structural systems, but rather the
improvement of the knowledge of the properties of construction
materials,their behaviour in time, particulary on the chemical attack
of the environment and under the fatigue effects. They require
the analysis of structural elements much more complex than the
usual ones, the study of different associations between steel and
concrete, and a more sys~ematic use of the non-linear analysis of
structures in which steel and concrete are associated.
The engineer is often called to study existing structures and
therefore he needs an education on such diagnostic techniques which
have been so far rarely useful to him, simply called to design new
works whith perfectly known materials and with freely selected
techniques.
2 G. Macchi
The 3 Arch Bridge in Venice is a personal experience in the field
of restoring bridges, it is for me a symbol of this new role prepared
for engineers; the monument is in this case not only a "monument", but
also an important resisting structure to be analyzed and to be
strenghtened according to the rules of our scientific and technical
knowledge.
The tools to be used in such occasions are not rigorously the
same of·our daily profession. It is necessary a further development of
the knowledge of materials nearly unknown to us until yesterday, as
the ancient masonry, improve the knowledge of the degradation
processes, analyze structures having a so high complexity that our
usual analysis methods cannot be used: structures nevertheless very
familiar to the engineers living four hundred years ago.
The problems are not very different when often we are called to
assess the stability or the residual life of reinforced concrete works
built in the first decades of this century or even in the fifthies or
sixties. In such cases the traditional strength of materials helps
very little and the need arises of improving our knowledge of the
corrosion processes of reinforcement within concrete, transformed by
chemical processes due to the environment, and extrapolate from our
still scarce knowledge of the development in time of such phenomena a
prognosis and proposals for strenghtening.
On the bow-string bridge of the Polyclinic in Pavia I had the
opportunity of completely live the problems posed by this new
situation created by a rapid increase of the effects of the
environment on the structures.
The discussion of the alternatives between demolition and repair,
unfortunately took place in the absence of precise parameters allowing
to judge (in a suitable cost-benefit analysis ) the advantages of one
or the other of the two alternative solutions.
The chemistry of materials is helping, but the development in
time of these phenomena, their dependency on several environmental
quantities, the real transformation of the resistance mechanisms in
presence of reinforcements in advanced phase of corrosion, are still
basically unknown. The engineer is therefore often forced to take a
decision with insufficient information.
Nevertheless, this will be our fundamental role in the years to
come, all the owners of roads and railways in the world agree in
thinking that the highest percentage of available resources will be
devoted in the next decades to the maintenance and repair, or even to
replacement of existing structures instead of construction of new
infrastructures.
The International Association for Bridge and Structural
Engineering IABSE will held next September in Lisbon an International
Symposium on this subject, in general on the subject of the durability
of structures, which is today one of the fundamental items for the
structural engineering.
I will report there on the events of the bridge in Pavia, which
led to the decision of demolition and rebuilding of the works, and I
will indicate the decisional process which took place; this type of
decision is now current practice even on our highways.
Fatigue stresses monitoring under traffic has been recently
performed on the Pecora Vecchia Viaduct of the Autostrada Florence-
Bologna; the measurements took place in view of the demolition of the
bridge.
Some problems and some solutions 3
Such studies are performed under the belief that the degradation
phenomena which took place on the viaducts of our highways and which
ace generally attributed to the chemical attack (particularly from the
salts used to avoid frost effects) are in reality more complex
physical and chemical phenomena on concrete. On such a degradation an
important role seems t~ have the mechanical effect of the dynamic
repeated action due to the transit of more and more heavy vehicles at
more and more higher speed. The results of such measurements are not
yet complete and have not been published. In cooperation with
Professor Radogna and his co-workers ,nevertheless, it seems to us
that the degradation of the concrete structures takes place under a
combined and synergetic action of chemical degradation and mechanical
action.
The transits of vehicles are considerably increased since the
time of design of such works; the load per axle of the vehicles is
also considerably increased, most of the heavy vehicles are very near
to the limit of the bearing capacity of the bridges; the transit speed
is considerably higher than the speeds usual at the time of the
construction. From measures recorded in few hours it is possible to
extrapolate to several millions the number of the significant
excursions in order to study the effects of fatigue during a life of
thirty years; they explain the high level of degradation presently
observed and the necessity of essential intervention or even the
demolition of the structure.
A difficult and sophisticated professional work is open now for
the engineer: it is dealing with the interventions for strengthening
and repair, an extremely difficult profession. A multiplicity of
materials is on the market, not always fit to the case, and not always
easy to control. Further problems are posed by interaction processes
with the existing materials of the structure.

2. HOW BUILD DURABLE BRIDGES IN THE FUTURE?

The experience done in the field of degradation leads to the

problem of techniques suitable for building in the future structures
more durable than done thirty years ago. In lack of more sophisticated
scientific approaches we may learn from the most frequently observed
phenomena. The most damaged structural elements are the slabs: they
are thin, they are reinforced with small diameter steel, directly
subject to the action of chemical corrosion and to the fatigue action
of heavy traffic.
Carbonation of concrete reaches to 50 or 100 mm in the thin webs
of structural elements and produced the corrosion of the transversal
reinforcement before that of the longitudinal reinforcement.
Those considerations should lead to conceave compact structures,
with reinforcements concentrated at a considerable distance from the
external surface and with reinforcement less sensitive to corrosion,
that is large diameter bars.
A second important cause of degradation have been the anchorages
of prestressing reinforcement when they are applied outside the
structure. The elimination, as long as possible, of such anchorages is
surely a good measure for higher durability. Finally, expansion joints
and moving bearings showed to be among the most vulnerable elements of
our bridges.
4 G. Macchi

The joints between adjacent decks did not find adequate solutions
for water-proofing, and were a cause of penetration of water, of de-
icing salts, and therefore of considerable corrosion.
Even the construction joints, which have been then closed ( so
that through them the prestressing bars have a continuity) have been
incorrectly executed for the lack of precompression in the joint
itself, and then formation of craks and further corrosive actions.
The supports of decks on the piles were also subject to frequent
and systematic damages with necessity of substitution. The question
arises whether, instead of creating ideally rational and simple
boundary conditions, particularly difficult or excessively expensive,
is it not preferable to create a continuity between supports and deck.
In such a way supporting devices and expansion joints may be reduced,
and used where they are strictly necessary, and where it is possible
to guarantee the best quality and durability by adequate (even
expensive) devices.
At the FIP Congress in New York, nearly twenty years ago, I
indicated such developement lines for the durable construction of
bridges. After a tiring and difficult work of restoration on just
executed works on the Autostrada between Venice and Trieste, I
succeeded in convincing the owners to realize a first viaduct along my
new ideas: the bridge on the river Meduna. It was a solid prestressed
slab, continuous along its complete length; this bridge has been a
prototype which has been later on repeated several times with
excellent results, even from the economic point of view.
One of such works is the viaduct Cadore, here in Udine.
Then, when possible, in the central part of the work the deck has
been built solidly connected with the piles, so that the supporting
devices have been eliminated.
The prestressing reinforcement is done with 40 mm diameter bars,
continuous for the complete length, without external intermediate
anchorages, therefore without possibility of chemical attack.
A further element of knowledge has been achieved more recently:
carbonation of concrete,and all the corrosion phenomena of the
reinforcement, may be considerably reduced by a more compact external
surface of the structure. The porosity of concrete may be reduced in
such a way that the penetration of water and C02 may be sensibly
reduced with the new technique of high strenght concrete. This option
has to be seriously considered today, not in view of reducing the
resisting sections of the structures, but in order to realize more
durable concrete.
Some problems and some solutions 5

3. BRIDGES WITH UNBONDED CABLES

If considered from the point of view of durability, the solutions

for bridges recently suggested by our French colleagues are
surprising: they suggest truss structures with extremely thin members,
caissson decks with very thin webs, prestressing cables out of the
section.

Such solutions are justified in the following way:

i) : prestressing cables within the concrete section hinder the

perfect execution and the perfect compaction, so that to put
outside the large sections of prestressing cables may be
better

ii) prestressing cables may better be protected outside by

different means, and may be subject to an easier control of
their possible deterioration; substitution is easy when
their corrosion appears.

Such a philosophy, which is just opposite to that previously

described, has its own logic and its own conception of maintenance.
The merits of the two opposite attitudes may be judged only case
by case; the French solution, which is based on an efficient
maintenance and on a continuous diagnosis of the degradation of
reinforcement, is not suitable for cases well known to our experience.
Continuous monitoring and sudden intervention in case of
necessity are certainly possible in the case of great engineering
works; more difficult is to imagine that such a perfect organization
would work in that net of roads of tens of thousand kilometers which
is the ordinary net.
The bridges with external unbonded prestress lead me to the
second subject which was intended to be treated in this lecture; that
is the necessity of new methods of analysis for new types of
structures, in particular the systematic use of nonlinear analysis .I
am confident that no Italian engineer is convinced to be allowed to
calculate the ultimate safety of such bridges with the usual methods
suggested by our national code. The bond between steel and concrete,
which is the essential assumption of our method of verification, is
lacking in this case;it is therefore necessary to take account of the
independence of the two materials in a nonlinear analysis which takes
rigorously account of the compatibility of deformation.
The consequence of the lack of bond is that concrete deforms
locally without a similar local strain in the steel; the strain of
steel is distributed along a great. length and therefore the stress in
steel may increase very little (or nothing at all) at the moment of
collapse of the structure. This mechanism is not difficult to be
followed for relatively simple structures, but requires to abandon the
old methods of analysis and to use rather sophisticated nonlinear
methods.
-6 G. Macchi

4. SLENDER MtMBERS
Another problem more and more frequently appearing in the design
of bridges is that of slender members, and also in this field
nonlinear analysis is a technique without alternatives. The problem
arises for the tall and very flexible piles of viaducts and bridges
and for the very long stays of the most modern cable-stayed bridges;
they are the most specific field of application of the nonlinear
methods of analysis of reinforced concrete.
Classic methods of analysis for instability cannot be applied to
such structures, and the analysis may only be performed with methods
taking into account the geometrical nonlinearity and nonlinearities of
materials.
Is this field the limit states analysis of reinforced concrete
led in the last decades to substantial improvements. Slenderness may
be extremely high, so that the check of stability would be
impossible without adequate consideration of two fundamental factors
having an impact on the second order effects: the value of the axial
force and the quantity of longitudinal reinforcement. The Eulerian
slenderness is not a sufficient parameter even for preliminary design
of slender members. Effects of reinforcement on second order effects
cannot be neglected.
Furthermore, under the action of permanent loads it is not
allowed to neglect the effects of creep; it seems therefore worth to
remember how in the field of such problems a precious help may be
found in the use of parametric methods, which have been developed in
the last 10 years for reinforced concrete sway frames. The available
numeric testing is very sophisticated and allows the extension of
Pesults to a large number of cases, taking into account: tensile
strenght, craking of the section, tension-stiffening, nonlinear
behaviour of concrete, creep: for all of them the most modern models
have been used.
The first diagram shows how the second order effects may be
reduced, at a costant value of the axial load, by increasing the rei
nforcement, which is represented by the geometrical percentage :it is
very easy to understand that under an increase of flexural moments
the best way to resist them is an increase of reinforcement.
The second diagram shows how even very slender members may be
subject to rather limited second order moments if the axial load is
limited. This concept is easy to understand. Very slender members may
result well acceptable, if subject to small axial loads. All these
concepts may be synthetized in the third diagram here presented, in
which the abscissa is a newly proposed equivalent slenderness A*
in which the Eulerian slenderness A is modified taking into
account the geometrical percentage of reinforcement Q , and the
value of the reduced axial load, 11 , that is the ratio between the
existing axial load and the maximum which may be substained by the
section.
Such two parameters, together with a correction taking into
account the resistance of concrete (in comparison with ~ conventional
resistance) may easily be assumed as criteria for design of very
slender elements to be subject afterwards to a more complete and
sophisticated analysis when the final project has to be established.
Some problems and some solutions 7

5. NONLINEAR ANALYSIS

The mentioned problems are not the only ones in which a nonlinear
analysis is useful. I will only mention the necessity to know the real
safety margins of the great cable-stayed bridges, which usually are
analyzed in extremely sophisticated way for the service conditions,
but for which the safety is very rarely analyzed.
According to the usual criteria of limit states, it would be
nece~sary to contemporarily operate the reduction of the strength of
materials, an increase of the external actions and a correction on the
unfavorable way of the state of coaction introduced by the tensioning
of the stays, which are subject to a very important model uncertainty.
It is certain that the great designers of the most important of
such bridges are perfectly aware of this necessity, but in the
technical literature very seldom this kind of considerations are
found.
I think that more care should be given in the future to these
problems, and that ultimate state verifications of cable-stayed
bridges should be systematically required and performed.

REFERENCES

1. [Link], [Link] - Assessment of a Bow-String Bridge 50 Years

Old in Pavia - Estratto dai Proc. of IABSE Symp. on Durability of
concrete structures, Lisbon, 1989

2. [Link], [Link], [Link] Synergetic Effects of

Environmental Actions and Fatigue - Estratto dai Proc. of IABSE
Symp. on Durability of Concrete Structures, Lisbon, 1989

3. [Link], [Link] - Definition modifiee d'elancement des

colonnes dans les ossatures en beton arme a noeuds dep1a~ables -
Estratto da FESTSCHRIFT Prof. Dr. Bruno Thlirlimann zum 60 -
Geburstag 1983

4. [Link] - A Model for Unbonded Tendons Prestressed Structures -

Int. Symp. on Fundamental Theory of Reinforced and Prestressed
Concrete, NIT, Nanjing, Cina, Sept. 18-20, 1986, pp.46-53

5. [Link] - Design Philosophy of the CEB-FIP Model Code With

Respect to Non-Linearity - Estratto dai Proc. of FIP Symp.,
Jerusalem, Sept. 4-9, 1988
8 G. Macchi

·~

Fig.l. Three Arch Bridge Fig.2. Three Arch Bridge

in Venice (16th century). in Venice ( FEM model)

Fig.3. Bow-string r.c. Fig.4. Corrosion of prestressing

bridge in Pavia(l935) strands in a bridge to be
corrosion of reinforcement. demolished.
Some problems and some solutions 9

Fig.S. Continuous slab Fig.6. View of Cadore

viaduct in Udine Viaduct.
(Viadotto Cadore).

Fig.?. Truss bridge with

external unbonded
prestress.
10 G. Macchi

[Link] unaau
211,U._
I

D
-
Cau~• AA Coupe BB
H:n"16 H,.·s

Fig.S. A slender structure:the

mast of a cable-stayed bridge.

Fig.9. Effect of reinforcement

(Q) on second order load (v) on second
effects of slender r.c. order effects.
frames.
~
2.1 "'•
2.0 • • fed • 1?.:5 Ml=la
-35.0
[Link]. Second order effect
Mz + Mil
Mr
as a function of the 1.7t---+----1f---++.H+-7'--+--- l
equivalent slenderness [Link]----+-----11---+-f+l+--+---1
* vo.b fed )o.s I

A. "' A. 1+ 15"Q ( 17.S ( MPa)

NEW TRENDS AND DEVELOPMENTS
IN BRIDGE CONSTRUCTION

R. Walther
EPFL, Lausanne, Switzerland
12 R. Walther

1. General remarks.
In recent years a great many innovative developments have taken place
in modem bridge construction, as will be described in the following. However,
it has to be pointed out that important and noteworthy as they are, they hardly
constitute an outright revolution as for example the invention of prestressing
did some fifty years ago. The most striking development can undoubtedly be
found in the surge of cable-stayed bridge construction with concrete decks and
towers, which are nowadays being built in great numbers almost anywhere in
the world. The idea of supporting structures by stays is indeed quite old the first
application dating back some 200 years and it seems rather surprising that the
actual break-through had to wait for so long a time. These important aspects of
modem bridge construction will be dealt with at the end of this presentation.
New developments which are likely to improve and innovate bridge design in
the coming years will now be discussed.

2. Desi&D philosophy and anahsis.

Our generation has certainly been decisively marked by the explosive
entry of modem computer techniques used in the design process. Highly
sophisticated and reliable programs of Computer Aided Design (CAD) are now
operational and widely used in practice. While this does indeed greatly facilitate
the daily work and allows complicated structures to be analysed relatively easily
and precisely, it may be rightly questioned, whether in the long run it will
really help to further the innovation and creation of new structures. One
should never lose sight of the fact that by far the most decisive and challenging
part of good engineering, which governs all other aspects, is and will always
be, the search for the most appropriate basic concept, an art which no
computer can assume and one which is incidentally very difficult to teach and
to master. As an example the still beautiful Brooklyn Bridge built over one
hundred years ago by John Roebling may be cited. At that time the available
theoretical tools did not permit the analysis of such a complicated highly
indeterminate static system correctly, yet due to his intuition and ingenuity
New trends and developments 13

Roebling was able to imagine and localize a novel load carrying concept made
up of a skilful combination of suspension and stay cables.
For the design he used a greatly simplified but appropriate approach
which today would be termed the statical method of the theory of plasticity. He
attributed one part of the loads to be carried by the suspension cables, which
could already then be analysed by the catenary equation and the other part by
the stays, the forces of which he determined by simple equilibrium
considerations.
It took nearly a century to recognize the potential advantages of the
theory of plasticity, which is the simplest and most appropriate method of
assessing the inherent structural safety of statically indeterminate systems.
Most modern codes today require separate verifications, one to guarantee the
serviceability (SLS; serviceability limit state according to the CEB terminology)
and the other to satisfy the safety requirements (ULS, ultimate limit state).
The clear distinction of the purpose of these two verifications permits us
to choose the appropriate method for each, which was not really the case with
the ancient concept of allowable stresses. Since at service loads prestressed
concrete structures behave more or less elastically, it seems only logical to rely
on the theory of elasticity. In many cases, especially in bridge design, the SLS is
the determining criterion and the fear that the application of the theory of
plasticity might lead to unsatisfactory or even dangerous structures is thus
wholly unwarranted. In the case of the free cantilever or incrementally
launched bridges the governing criterion are the prevention of cracks and the
limitation of deformations during the critical erection phases, both of which
have to be checked on the basis of elastic or quasi-elastic analysis by taking
into account all the relevant phenomena including shrinkage, creep,
relaxation and temperature effects.
It is however of prime importance to recognize that those particular
effects just mentioned have in general no or only insignificant influence on
structural safety, as the theory of plasticity or any other non-linear analysis
clearly reveals and what can also be verified experimentally. Thus there is
really no need to consider the effects due to shrinkage, creep and temperature,
sometimes even increased by arbitrary load factors, for the ULS check.
The recognition of these facts often permits structural elements such as
columns, piers or pylons to be built more elegantly and more importantly to
 14 R. Walther

avoid expansion joints and mechanical bearing devices, which are always
susceptible to premature deterioration.
The theory of plasticity gives also a clear answer to the often discussed
question of how to treat the hyperstatical moments of prestress for the ULS
check. Even today many codes stipulate formulae of the type:

where:
Mu ultimate moment at the section considered
"rQMQ moment due to load effects increased by the load factor (1.6-1.9)
Mph hyperstatical (secondary) moment of prestress.
'Yp load factor of Mph (varying between 0.8 and 1.3 depending on
the code considered).
Again it has to be stressed that the hyperstatical moments pertain to the
elastic state and can no longer be clearly defined at the ultimate limit state due
to the considerable moment redistribution which will have taken place. While
the formula just mentioned may yield satisfactory results in most cases, there
are others (inverted T-beams for example) fig 1, where the shifting of the
moment closure line stipulated by the term 'YpM ph goes in the wrong direction.
The resulting sectional safety factors g, calculated backwards by assuming 'Yp= 1
(first line in the table of figure 1) and 'Yp = 1,3 (second line) clearly show the
inconsistency of such approaches. Furthermore it is sometimes recommended to
take 'Yp=0.8 if Mp acts favourably and 'Yp=1.2 in the unfavourable case, which
means that the global system would not even satisfy the equilibrium conditions
and the resulting global safety, which is identical for a T -beam and an inverted
T-beam (last line in the table of figure 1). No such ambiguities are met if one
performs the ULS check by the statical method of the theory of plasticity, which
clearly assesses the effective distribution of internal forces.

3. Materials.
By applying advanced concrete technology it is now possible to obtain
very high strength concretes of about 100 N/mm2 or even more under field
conditions, and without exorbitant costs. It is true that compressive strength
values of 120 N/mm2 have been obtained experimentally under laboratory
New trends and developments 15

STATIC SYSTEM, CABLE PROFIL, LOADS CR08& SECTION (j) CROSS SECTION ®

MA, g + q"'" -10'000 kNm

"'
~ ~~~
Mr. 9 + q = ~ 5'000kNm

SERVICE LOAD MOMENTS

<D ®
SPAN SUPPORT SPAN SUPPORT

~-ii8l r.z~ [!~ l2:!8J

r---·-t----t
~ [2;121 ~~] !(7~
1----
~:

GLOBAL SYSTEM GLOBAL SYSTEM

Fig 1
conditions before, but it is only recently that these concretes have become
commercially available.
For concrete bridges of short or medium span the necessary dimensions
are often dictated by serviceability and detailing requirements ( deflection,
cover, durability etc ) rather than by the concrete compressive strength and
therefore one might feel that the use of high strength concrete is not really
relevant for bridges, except for members heavily subjected to compression such
as columns, piers, pylons or compression chords of cantilever bridges.
There are however, other properties of these concretes which are potentially of
great interest in structural engineering, namely :
-very high strength concretes can reduce creep and shrinkage up to
50% compared with normal concrete.
-considerable improvement to the resistance to chemical aggression,
notably greater protection against corrosion of the reinforcement.
-improved freeze-thaw resistance.
-greater abrasion resistance.
The main problem with high strength concretes is that their ductility is
decreased appreciably and this creates many problems (fig 2 ).
 16 R. Walther

COMPRESSION STRENGTH DIAGRAM

llpo i<:n.J

- - AOGAEGATE' AROA&. WITHOUT Su::A

- · -- [Link]!"AROAL1·1~
- - - - - IASALT AaGREClA7E IS" [Link]

Fig. 2. Stress strain diagram for concretes of different strengths.

This fact is all the more crucial as modern design concepts are based on
the theory of plasticity or other non-linear methods, which are only applicable
for ductile materials. However it should be noted that one way of increasing the
ductility of high strength concretes is by proper detailing. A good example of
this is tightly wrapped transverse reinforcement. Also it has been shown
experimentally that decreasing the stirrup spacing means that the lateral
deformations which a column could sustain without any loss of compression
capacity increases substantially. There is on the other hand, a limit to the
amount of transverse reinforcement lest not to impair the necessary
compaction.
Composite construction is another way of producing the concrete as
ductile as possible. By placing the deformable steel section at the surface of the
concrete member and then encasing the two as a monolithic section the
ductility of the concrete is increased.
This solution was recently proposed by Prof. Schlaich and the author in
their prize winning project of the new Williamsburg Bridge in New York (fig.
3). The slender pylons were considered as a composite stiff steel casing filled
with very high strength concrete.
A good example of composite construction is high strength concrete filled
steel tubes. In this case the ductility, load bearing capacity and compacting of
New trends and developments 17

Fig. 3. Proposal for the new Williamsburg Bridge.

Fig. 4
concrete are all improved. French engineers have recently adopted this idea in
bridge construction by using a concrete filled steel tube as the lower cord of a
bridge section as shown in fig 4.

4. Skew slabs.
Urban conditions for overpasses and underpasses quite often require that
skew bridges of moderate spans be used for which solid prestressed concrete
slabs are usually best suited. For such slabs the customary distinction between
primary and hyperstatical load effects due to prestress cannot be applied since
their internal hyperstaticity is infinite.
18 R. Walther

In principle, the serviceability limit state is checked by an elastic analysis.

However, even small micro-cracking can already lead to considerable
redistribution of stresses which has to be accounted for in the analysis. The real
deflections often become larger than the calculated values, while theoretical
moment peaks, e.g. in obtuse angles, are of little significance since neither are
they necessary for equilibrium, nor do they appreciably affect the general
crack development. A minimum reinforcement according to detailing rules will
in general be sufficient in these regions.
Unfortunately, the statical method of the theory of plasticity does not
readily lend itself to ULS-checks of skew slabs. Even relatively modem finite
element calculations yield only the elastic distribution of stresses, which do not
adequately reproduce the actual conditions of the ultimate limit state. In reality,
this can only be achieved satisfactorily by using non-linear finite elements,
which are not readily available and whose application for daily practice is still
somewhat tedious and expensive. In the meantime one can easily use the
kinematic method, also called the yield line theory. Since this results in an
upper bound limit, thus leading to an over estimation of the load-carrying
capacity, one should be very prudent in choosing the assumed mechanism. The
true mechanism can easily be found by relatively small and inexpensive model
tests.
As an example, some pertinent results of a large series of tests conducted
at the Swiss Federal Institute of Technology at Lausanne are presented in Fig. 5
together with a sketch of the appropriate mechanism. Among others, the
following conclusions can be drawn from this type of investigation:
- by selecting a realistic mechanism, the ultimate load can be
accurately predicted.
- transverse prestressing can greatly improve the behaviour of simple
skew slabs at the ultimate, as well as at the seviceability limit state.
- no significant passive reinforcement is needed at obtuse angles or at
any other singular point.
New trends and developments 19

DEFORMATIONS OF THE FREE EDGES NEAR TO ULTIMATE LOAD

• $Vt8 Wifl1,.AIIRAI..Ul. Coi8U:.S (uniformly di stributed load)
0
... 20

60

THE THREEOIFFERENT [Link] PREST'RBSING TBCICNi

O'.liSIDEREOFOA MSlABSOF 15° AHGI..EOFS¥.EW

Fig (a) Fig (b)

~ Fig (c)
TOP SURFACE H ==:= BOTTOM SURFACE(+)
YIELD LINE YIELD LINE

Fig (d)
(+) POSITIVE YIELD LINE
( - ) NEGATIVE YIELD LINE

Fig 5 . Skew slabs in micro-concrete tested at the Swiss Federal Institute

of Technology, Lausanne.
a) cable arrangements and dimensions.
b) deflection at the free edge under service load (the transverse
prestressing reduces deflections significantly).
c) yield lines.
d) failure mechanism.

S. Cable stayed brjd~:es with slender concrete decks.

As has been mentioned above, cable stayed bridges undoubtedly represent one
of the most rapidly developing fields in modern bridge design. The structural
configuration of this type of bridge has lead to the more recent development of
cable stayed bridges with slender concrete decks.
20 R. Walther

For multiple cable stayed bridges with close cable spacing the longitudinal
bending moments in the deck due to permanent loads remain very small, while
the ones due to live loads increase with the bending stiffness of the deck. Hence
for small and medium span cable stayed bridges there is no need to choose stiff
decks, on the contrary a relatively slender concrete slab may often prove to be
the best solution. The main advantage of this type of bridge is its structural

Fig. 6 Diepoldsau Bridge, Switzerland.

Section A-A Section B-B

Fig. 7 Elevation and cross-sections of the Diepoldsau Bridge.

New trends and developments 21

simplicity and easy erection which renders it very economical and elegant. The
first time these ideas were put into action was with the Diepoldsau Bridge over
the Rhine in Switzerland Fig 6 and 7. This bridge consisted of a main span of
97m supported by a multiple cable stayed system, the average deck thickness
being 0.45m. In spite of the deck slenderness, the global stiffness and the
sensitivity of the system to vibrations are quite similar to those observed for
more traditional bridges.
Theoretical and experimental work has also shown that these concepts
are equally applicable to longer spans. An example of this is the project of the
La Dala bridge in Switzerland, with a span of some 210m and an average deck
thickness equal to 0.42m and with a width of 13m, Fig 8 and 9.

Fig. 8. Project of the La Dala Bridge, Switzerland.

The main problem with such a bridge is its resistance against flutter due
to high wind forces. A thorough dynamic analysis and wind tunnel tests have
shown that the initially conceived cross-section ( fig 9 left ) did not guarantee a
sufficient safety against flutter; thus it proved necessary to shift the cable-stays
further out to the edges (fig 9 right).

6. Experjmental work.
Even though the static and dynamic behaviour of cable stayed bridges can
nowadays be analyzed by modem non-linear computer programs, it was deemed
desirable to verify the feasibility of choosing very slender concrete decks by a
 22 R. Walther

I l__j. ___ ~~---'!04=.50"'--------:----.1-l

J.-------'----""104 50 _ _ _ I

! !Jt7TJ t20900
I
l!J0840
I 47.70

f
-JJOm

Elevation.
INITIAL DESIGN MODIFIED CROSS SECTION

0:19

·'
-~q_- l- L!Q__ fiO 840

1~10

Cross sections

Fig. 9. Elevation and cross-section of the La Dala Bridge, Switzerland.

thorough experimental investigation on a fairly large scale model. Such a study

has been successfully completed at the Swiss Federal Institute of Technology.(4).
The main dimensions and the set-up of these tests are shown in the
figures 10 and 11. In addition the aerodynamic properties of this type of bridge
were determined on sectional and global models in wind-tunnel tests.
New trends and developments 23

Fig. 10. Test set-up of the experimental investigation.

!- '
I

HAUBANS
\
\
r::t=====P =f'"
t 700(mm}
:!
Fig. 11. Main dimensions of the test bridge.
24 R. Walther

The following main conclusions can de drawn from the theoretical and
experimental investigations:
The buckling stability of bridges with relatively narrow spacing of the
stays ( 5 to 10 m) becomes hardly ever critical. The failure load is usually
determined by the yield strength of the stays.
The horizontal forces introduced into the deck by the stays effectively
prevent cracking of the concrete deck. Small cracks occurred only at mid-span
and close to the end support.
Such bridges are rather flexible structures and the displacements are
relatively important. For the main model the ratio between maximum deflection
and span length is about 1/600 under service load distributed over the entire
structure (1/300 for loads distributed over the central span only). As can be
seen from figure 12 the non-linearity of geometry and material have clearly to
be considered in calculating the displacements.

interruption of the test

3.15 I sz
3.0 c+··-----~'---r--~
last measured values
'7
* •
·------ calculated

• measured

measured values, comiCied to take

* into account the observed vertical
displacements at the end supports

~
0 20 40 60 80 100 120 [mm)

Fig. 12.
New trends and developments 25

As far as the dynamic and aerodynamic behaviour is concerned four

phenomena have to be considered:
- vibrations under traffic loads.
- galloping.
vortex-shedding
- flutter.
Vibrations due to traffic loads do not impair the safety of such structures
( the stress-variations are way too small to produce fatigue problems ), but they
may cause undesirable physiological effects for the pedestrians. The study has
however shown that vibrations remain in the fully acceptable range according
to all the pertinent recommendations.
Galloping of the deck can only occur if the diagram of vertical loads
induced by wind versus angle of attack have regions of negative slopes
(SCN /Sa<O ), however, this is never the case for slender decks.
Vortex -shedding may in certain cases give raise to potentially dangerous
vibrations even though one of the determining parameters, that is the ratio of
the deck width (B) to the depth (h) is usually greater than B/h > 10 and thus in a
rather favourable range. In order to limit the vortex-shedding phenomena the
deck should be given a streamline shape, if necessary deflectors have to be
provided to channel the air flow around angles. At any rate sectional wind
tunnel tests have to be conducted for important structures beyond the known
range of experience.
However, by far the most critical aerodynamic phenomenon is
undoubtedly the flutter, that is a combination of bending and torsional
movements with rapidly increasing amplitudes, which may occur if the
torsional and flexural natural frequencies are similar. Unfortunately the best
method to prevent flutter, choosing a high torsional rigidity, is contrary to the
very concepts of bridges with slenders decks. This fact limits the range of
feasible applications of this type of structure to bridges of small and medium
spans ( up to 250m) or to bridges of greater deck width.
In the case of the Skarnsund Bridge, actually under construction in
Norway, with a central span of 530m and a width of only about 12m this concept
was clearly no longer feasible and for this reason a torsionally rigid deck
(triangular box-section) had to be chosen (fig 13 ).
26 R. Walther

I I

01~, (rJ-- ~· ·::... ~·~ uf! r';l

~~· ~~~~~~1o5~.crn~u4~~
~:
_,
~: -- ' I~
i
L-
I
LOO .. -t 4 60 ••o20 460 t .1~~
+ 1?51 5 25 525 t 12~ ~
1300
t

Fig. 13.
Cross section of the Skamsund Bridge in Norway with a central span of 530m.
Referepees.
( 1) Walther R.,Houriet B.,Isler W.,Moia P.:
Cable-Stayed Bridges.
Thomas Telford, London,1989.
(2) Almeida J.:
Non-linear problems in Cable-Stayed Concrete Bridges.
Swiss Federal Institute of Technology
Prestressed and Concrete [Link] September
1988.
(3) Klein J.F.:
Comportement et stabilite des tabliers minces dans les ponts
haubanes.
These de doctoral en cours de redaction a l'IBAP-EPFL.
The behaviour and stability of cable stayed bridges with slender
decks.
PhD Thesis, in the process of completion at the Swiss Federal
Institute of Technology, EPFL, Lausanne, Switzerland.
( 4) Cable-Stayed Bridges with Slender Decks.:
Test Report.
Swiss Federal Institute of Technology
Prestressed and Concrete [Link] September
1988.
 CANTILEVER BUlL T BRIDGES
WITH PREFABRICATED SEGMENTS

R. Lacroix
Freyssinet, Boulogne, Francia
28 R. Lacroix

1. llmrruTICJJ
Bridges with cast-in-situ segm:nts care up against certain limitations, particularly
as far as large structures are concerned , or at least have certain drawbacks.
The first consists of a relatively lengthy construction schedule' caused by the
casting cycle of segm:nts : the no:nml cycle lasts one week, in order to allow for the
hardening of concrete on days off ; an average seg-a:mt leQgth of 2 •.::JJ m, WliliiEEIIl a
construction of 5 m a week on each side of the pier, and for a very long bridge, the
contractor nay rnve to increase the mnnber of carriage form travellers to nmrit:ain an . ·~"'
adequate cprogram.
In certain cases, a cycle of two segm:nts a week could be attained, by IIE8IlS of either
a high strength concrete, or a the:nml treat:Irent to increase the concrete's early strength.
However, such achievaiE!lts are not easy, and the difficulty in controlling the arms'
deflections is increased, due to the concrete being so recently cast at the tine of the
prestress tensioning.
Fran a quality point of view, the cantilever built bridge with cast-in-situ segm:nts
runs another risk due to the fact that the strength of the concrete is only koo.-m after 28
aays ; if the results of the control samples prove to be insufficient, one must then choose
between either the ClaiDlition of all the concrete already cast, or the acceptance of a
derogation , provided that a few extra precautions are taken.
Fiually, statistically, a higher frequency of disorders has been noticed on bridges
with cast-in-situ segm:nts because of early stressing of the concrete and the obviously
difficult casting conditions.
The design of prefabricated segm:nt bridges IIEkes these problans avoidable, and is
therefore an actual 1mprovaiE!lt despite a few inconveniences which will be described below.

2. 'IHE DESIGN PRil'CIPlE RR PREFAPRICATED SEJJ1ENIS

Prestressing is An excellent llB'lllS of assenbly, thanks to the canpression exerted on
the faces in contact with the prefabricated pieces ; but to achieve this tightening without
any problans, the nutual contact must be perfect. ·
The solution to dris problan was thought up in the late fifties by Campenon-Bemard,
for the construction of the Oloisy-le-Roi Bridge on the Seine River tnear Paris).
Cantilever built bridges 29

This solution consh;ts of casting the nth seerent with its cootact face with (n-1)
segm:mt cast directly m contact with the latter·, A perfect similarity of the cootiguous
faces is therefore achieved! ; these faces are said to be rratch-cast, and provided that
incidental defamations do not appear between the casting of concrete and the assenbling of
Beg!!El1ts, an excellent mutual contact at the joints is obtained.
Generally, :[Link] prior to assaTJbling two contiguous segr'E11ts, the contact faces
are crnted with a rrrulti-purpuse epoxy resin which ;
- firstly guarantees the joint's watertightness, an important characteristic,
especially if prestressing ducts tmbedded in concrete pass through the joint ;
- secondly, ensures a better contact by evenning out any slight :imperfections in the
segtlE!1ts' rntch casting, with the thidc1ess of the resin ;
- lastly, fctCilitates assanbly with the lubrication of the contact faces, the friction
coeffiLient of the resin before polymerization being quite small.
In sane cases the segiiEilts have been assenbled without resin. For reasons [Link]
earlier, t.11is method would seen inadvisable.
When the joint between two segments is obliquely crossed by prestressing reinforcE!lEI1ts,
apart fran the exceptional case when the canponent of the prestressing force in the joint's
plane is strictiy balanced by the shear force exerted at the joint, a tangential force is
applied along the contact area, and because of the small value of the friction ccefficient
of the resin, the segiil:!flts tend to slide one against the other. This sliding is generally
avmded by fitting the faces in c6ntact with shear ~reys (ref. fig 1 ), thick enough to
withstand the shear force.
These keys, useful in avoiding the lJRltual sliding of two contiguous segments, are
also placed in the slabs, where their purpose is different : they insure the rrechanical
assemblmg of two pieces of slab belonging to adjacent segments (ref. fig 2 ) , and
allow a better distribution of the traffic load.

3. FABRICATICN, sra:KIN:; AND EREITICN OF SEll1ENIS

Che of the rrain advantages of construction with prefabricated segiiEilts is in the
quality achieved, thanks tu the industrial casting of concrete • Especially as far as large
bridge are COHcerned, the fabrication unit for segments nay be conceived with a stiff steel
frame, with rrams of important vibration, and if the need arises, heating equiiJIE11t for
concrete, without the ~ght of the installation being a drawback, as is the case with
cast-in-s1tu se~nts.
This unit can be located near the hatching plant, thus shortening the actual casting
time, and facilitating the use of superplasticizers (since the effect of superplasticizers
on fresh concrete is short tenn, and can disapear 30 to 45 minutes after mixing).
As explained above, the nth ~t is cast directly in contact with the (n-1) segment,
the face of the latter ,simply being cootcd [Link] Cl form oil type product.
A well equipped unit can produce one segment every other day with a setting and
hardening time of the concrete of .:h hours, and even one segment per day, provided that the
early strength of concrete is sufficient.
30 R. Lacroix

Incidentally, oue rrust note that the requ:[Link]!ent of a very short fabrication cycle
justifies the use of a high strength concrete, far nnre them nechanical [Link]!IEnts :
for bridges, the concrete usually considered nrust have a characteristic strength on
cylinders of about 35 M!2 at 28 days, and it::i,s:·_quite rare that the structural analysis
danmds a stren,<>th higher than 40 or 45 MPa • ..Uthough for a long sp311 bridge, it !!light be
nnre ecooanica1 to :increrule the concrete quality, rather th311 :increase the nunber of
fabrication cells ; for example, such was the case with theRe Island &idge, over 3 km
long. This structure was bwlt with a coocrete with a characteristic strength of 55 MPa,
whereas the project only required 40 MPa. The use of high strength concrete therefore
enabled a considerable anotmt of tine to be saved, the opening of the bridge could ..take
place before the start of the tourist season (the bridge was canpleted within ,18 llDilths,
including futmdations). A11other llllpOrtant side effect of the use of strengh concrete nrust
a1fD be noticed : this mterial, due to its CXII1Jl3Ctness, actually gives the structure a
better durability : the action of superplasticizers ori concrete is achieved through an
:impressive reduction of its water content ; the W/C ratio, i.e. water \\eight over carent
~t, the value of which anotmts to 0.45 to 0.50 for a typical concrete, d:iminishes to
0.32 or 0.35 for a plasticized concrete, her;!ce a very smll free water content in the
hardened mterial, t1my tests proved that the water or air penreability of a p1asticized
coucrete was 10 to 100 tines smller than that of a norna1 concrete. The <.:arbonation.,. or
the ingression of chlorides in a lll3rine atiiDSphere, is therefore much sla..e:r.
The last but not least advantage of the prestressed concrete bridge nrust be added, as
far as. high .strength OCXICI't!te lS concerned : again because of Its lcM content of free
water, its creep is XI to 40 %less than that of a nonm1 concrete, which also cl:innnishes
tk prestressed losses, and the re-allocation of forces.
After fabrication, the stomge of the segl!E!1ts, which are often quite cumbersc:ne, it;
in ~ open air. The only precaution to take concerns the stacking of segl!E!1ts on top,.of
each otlE" ; the early concrete my be subnitted to an :important creep, and it is then
advisable to avoid any flexions caused by the \\eight they carry.
It also useful to set up a data lase M1ich records the fabrication date, the
respective position in the storage area, and the waiting period of each seg11E11L; this will
ensure the correct mtch-casting of .the joints, and also keeping accotmt of the concrete
creep J:m811Eters, ([Link], age at first looding), for the detenuination of the canber
to be given .to the arn5 during construction. The sane data is a1fD useful for calculating
the re-allocation of forces due to creep.
The handling of segl!E!1ts can be done in various ways : when the deck is to be built
entirely over a navigaole waterway, or an accessible area, a transportation on lm"ge of
flat bed trailer can be considered ; the end of the ann is then equipped with a derrick in
order to lift the seg11E11ts, and to bring than into cmtact with the previOus ones.
~er, when the ground belcM deck is not unifonn, \\hlch is generally the case, it is
strongly advised to u::,e a lauOChing gantry resting on the piers [Link] to erect the segl!E!1ts
progressively. launching gantries are costly invest:nents, but, 1f well conceived, can be
redeemad ru1d used again on several successive sites.
The actual adjustment of the arn5 1 profile during their erection , up to the profile
desired for the service life, lllll::it be carried out before hand, prior to the segnents'
fabrication, and cannot be subjected to any toodification during the construction of the
deck.
Cantilever built bridges 31

In othP..r \[Link], prefabriU'Ition illlplies, cmtrary to cast-:i.n-situ bridges, a perfect

previous knowledge of the concrete's rheological properties, and of the [Link] schedule.
An interruption of \[Link], calli:ied by a strike for exanple, nams a m:xlificatim of the
segtlBlts 1 bonding age, and therefore m:xlifies the alignnEilt of the profile.
After the ;erection of the prefabricated segrents, the arnE 1 continuity is obtained by
casting in situ a special closing piece, which dOE'S not cause any rmtirular difficulty.

4. Cm:::EPI'ICN AND INSI'[Link] OF PRESIRESSJ1{; nNJ:lN)

As for.,bridges with cast-in-'-Situ segnents, the prefabricated segnent bridge contains

isostatic tendons, namt to ensure the stability of the structure during constructioo, and
continuous tendons.
As far as the declt':s analysis hipothesis is concerned, one nrust note that prefabrica-
tion [Link] [Link] : with cast-:i.n-situ segnents, and with a rather
canplicated alteratioo of the folm\'Ork, SCliE reinforcing lm-s can be laid across the
segnents' joints ; this allows tensile stress under the effect of short lasting actions.
This is ruled out with prefabricated "'segnents.
The anchoring of isostatic tendons also differs :: with cast-:i.n-situ bridges, these
tendons are usual!y anchored at the ends of the webs. This disposition is inadvis:ible m the
case of prefabricated segnents, because the :innaliate and long tenn deflection of canpressed
concrete behind the anchor plates tends to destroy the gearetrical mt:cb-<:asting of the
segnents' contact faces.
The tendons are preferably anchored m concrete blocks or ribs protr¢ing fran the
concrete, lilside the box-gitder fonred by the deck.
'Ihe watertightness of the ducts' coupling across the jomts is another delicate mtter,
for prefabricated segnents as well as those .cast-:i.n-situ•. The ducts, I!Bde of spiral .steel
sheets, are joined end to end with tbreaded~.sleeves scr~ to the dutts ·fillets. The
watertightness is achieved by applying a supple, canpressible product around the opening of
the duct, similar to that used in building coostructioo. Whatever the precautions taken, it
is important to check, by blowing air for example, that watertightness is perfect, and that
the ducts are not blockeS by resin.
As far as all cantilever built bridges are concerned, the areas behlnd the
prestressing anchor plates nrust be very carefully treated, since their concrete is subnitted
to an often considerable tensile stress ; after tensioo:ing, tensile cracks are [Link]::ines
noticed, of a SIBil width (about 0.1 nm), along the ducts' outline ; the starting point of
these cracks is located a few <lec:iJretres behind the prestressing anchor plate, and their
length my vary fran a few <lec:iJretres up to one tretre. These cracks do not jeo}BI"dize
the structure's safety, and often vanish after canpletion of the \[Link], due to the sheer
force exerted by the leads applied to the ann~ They can be avoided, espec:ially If the
detailing rules prescribed by the French prestressed concrete code are follal\ed : Code tiPFL83.
lastly, one must note the risk of disorders induced by the IDstallation of longitudinal
prestressing tendons m the lower slab of a box girder of-,variable height : the height of .~,
the girder being variable, the slab is curved, which induces a radial force fran the
prestressing terdons ; during tensioning, no chsorder arises, since this radial force: is
OOlanced by an opposing force, induced by the curve of the canpressed concrete.
32 R. Lacroix

:But when the bend:ing IIDlEllt increases, due to the effect of live loods, to a thermll
gradient, or to the re-allocation of lood effects due to creep, then the concrete of the
101\e!' slat> is no longer canpressed, and the radial force of prestress:ing tendons is no
longer opposed, and could trigger the rupture of the slab, through excessive flexion.

5. EliTERNAL p~:m:;

As explained above, the layout of prestressing tendons and anchorages damnds IIBil}!
special precautions. There are other cmstraints related to the project itself : for ·
example, for the checking of sheer force, the web's thiclmess cannot be taken into account
as a \Jlole : one [Link] deduct either the d:ianeter or half the d:ianeter of the enbedded
prestressing ducts, according to the code follo.-e:I.
For all of these reasons, and tmder the leadership of French engineers, a real
"external prestressing school" has been developed, with the construction of IIBily long span
bridges.
In fact, the feature of tendons outside the concrete has other advantages :
- construction is simplified,
- casting of oeans h> mre easily earned out,
- provided that certain precautions are taken, the tendons can if necessary be
ruplaced during the life of the structure,
- lastly, external prestressing allows the installation of high capacity tendons
[Link] any risk of cracks in the concrete behind the anchor plates.
To the det:rinEnt of external prestressing, one rrrust ~er note a 101\e!' resistance
to flexion at the ult:i!mte limit state, due to the tendon not being bonded to the concrete
(for a detailed study of the rrechanical properties of girders fitted with external tendons,
the reader can refer to the IIBilY articles pertaining to this subject).
The different types on external prestreSsing reinforcarent are as follows :

1. The OOrijl wires or naked strands, galvanized. Such re-inforcaiE!lt has been m;ed on
several bridges ; the guarantee. ,of durability they offer does not sean satisfactory, except
in the case of a very dry clinate (and even then one [Link] check the absence of condensation
inside the box-grrder).
2. The wires or strands, laid [Link] a sheath nade of a SJI)()th steel tube, filled with
a carent grout. This arrangaJalt is preferable, and hcJs been especially chosen for IIBilY
rupairs using auditional prestressing. However, the steel tube, even when painted, is
exposed to cmrosion.
3. 1he wires or stm.~,laicLwithin a high density polyethylen duct (HIFE), filled with
a carent grout. This solution is preferable, as the duct within the box-gtrdcr, is not
exposed to the lN radiations, which could induce ageing and craddng of the polyethylen.
4. The wires or strands, laid [Link] an HIFE duct, filled with a soft naterial, either
grease or was. 1his solution, slightly mre costly, is adrisable, since it allows the
possible :retensioning or repJacaJEilt of a tendon if the need arises.
Cantilever built bridges 33

5. The strand,:individually protected by an HDPE duct, with an :interposition of grease

or wax. This latter solution is the best as it allows an extrere pliability for
re-tensioning or replacing the reinforcerent, as well as an excellent protection against
corrosion.

6. aNllBICN ; A FBI/ <XNSIDERATICN

Bridges constitute a very special kind of structure : their statical schare is clearly
detenrrined and their structural analysis is IIDre simple than IIDSt building franes. [Link],
if the latter can acccm:xlate approx:inative calculations, as far as bridges are concerned,
mistakes are fatal.

The reasons for tk particular sensitivity of engineering structures are [Link] :

- First of all they are [Link] to actual leads closer to nanina1 leads than those
of rrany other structures, and, _:in spite of taking :into account the dynamic coefficient, the
probably tmderest:inated.
- The effects of cl:inatic conditions are also liDderest:inBted :in rrany cases : the
thernal gradient, bebveen the road surface covered [Link] a black bit\.l!rl..noug layer exposed
to the SliD, and the tnlderface of the deck :in the shade, quite often exceeds the prescribed
value of 10° C.
The effects of frost and de-icing salts are also dangerous. For a long t:ine, experts
have been lcdcing for a technique to prevent the pouring of buckets of sodi1ll1 chloride mto
concrete, the sworn elieny of renforc:ing bars and tendons.
- [Link] lastly note an llllfortunate practice, tmder the pretext of presenting a ·
clever alternative, of econarrising as noch as possible on naterials whilst abiding with
the regulations, in order to built as light a structure as possible.
This practice is det:rinEntal to the owners, who nay buy bridges with a far 101\er
durability than expected ; it is also detrimental to the consultants and contractors,whose
gloOOl reputation can be danaged.
A IIDre detailed analysis of the present situation leads us to believe that it results
fran three ir,dependent foctors, which are :
*The developrent of stronger and stronger canpetition, leading to an excessive
dissection of projects, often neans exceeding the operauon's budget, the difficulties :in
carrying out the works according to the chosen liEthod [Link] any saving on naterials.
*The popularization of canputer science, with the appearance of powerful and
effective programs on the nmket, facilitates working out voluminous supporting calculations
at a nxxlerate price ; rrany are misled by this presentatim, forgetting that a III3Ch:ine can
only give back what it has been fed, and that nan remins responsible for conceptim : the
best canputer i.'1 the world, even fitted with the best "C.®'', will only 'WOrk out a second-
rate project 1f cuntrolled by a poor designer.
* Lastly, the third factor is the decreasing teclmical kn.<Mledge of owners, who are
less and less cap:tble of estinating a project's worth, and therefore of ~ out the
hazaruous an<1 dangerous offersA
34 R. Lacroix

Often restricted by over strict rules regarding tenders and canpetition, the owners
tend to award to the lowest bidding consulting engineer, a study which represents only a
slight percentage of the construction cost ; this is whj the studies are saretines slap-dash,
entrusted to a young engineer or a trainee, "cheaper" than an experienced engineer.
In order to restore the qualities of durability to bridges, which used to be theirs,
it is necessary for the owners to be held accountable for their structures,; they [Link]
realize the importance OL the strucutre's global cost. This global cost includes not only
the cost of studies and construction, but also rraintenance, possible repairs and eventual
replac:erent of the structure. The opt:imization of global costs will give priority in each
case to an elaborate preliminary smdy of the structure, which is the only way to prevent
disorders during the structure's life.
Cantilever built bridges 35

COUPE L ONGITUOINALE [Link] A. A

( v)

'
~)

COUP£ B-e

T I

I
I

•,·

~- ___,_
Fig 1 FIRST GENERATICN SBJ1ENI'
t--
I -
__)
I

LONVITUOIP."AL( (L[VATIOH A_ A

COUPE 8. 8

I I

_j
I

~~~~~--~~~~~~~··~~--~~~~~~
i
Fig 2 SEJniD GENERATICN SBJ1ENI'
36 R. Lacroix

---...._1___ _],..;--___,
_Q_

]]_
---------------
-------------

PRINCIPLE OF CONSTRUCTION
Cantilever built bridges 37

'

4
1f
---------
I

~
Jl

STRUCTURAL SCHEMES
38 R. Lacroix

construction
on scaffold

sr=' 'Y'
lest

r
0,5 L I 1 o,sl

I bilateral
bearing

STRUCTURAL DETAILS
Cantilever built bridges 39

lest

'
anchorage
form

working platform

PRINCIPLE OF A MOBILE GANTRY

40 R. Lacroix

!/1 Z~ZZZ ZZ~:::::::::::::::/!/1 11// I'~' Z 1

EXAMPLES OF TRANSVERSE SECTIONS
Cantilever built bridges . 41

on a support at middle of span

VARIABLE HEIGHT - INCLINED WEBS

42 R. Lacroix

I.
ol
II I
I
'I
II
II I I 'I
II I I II
; : I I ::
II I : II
II I lo

~D~ - I

LIAISON DECK - SUPPORT

Cantilever built bridges 43

slenderness
20

19 /
/
/
18 //
/
/
/
17 /
/

16

+ - - - t - - - - + - - - - - - ; . f(m)
50 100 200-

deck height on support

slenderness

50 100 150

height at middle span

RATIO HEIGHr/SPAN
44 R. Lacroix

temporary
anchors

temporary
support

STABILITY DURING CONSTRUCTION

Cantilever built bridges 45

M = Pxd

wind

OPERATING AND CLIMATIC ACTIONS·

46 R. Lacroix

isostatic tendons (M < o)

lhiking tendons (M > o)

~ continuity tendons
Cantilever built bridges 47

WEIGHT OF THE DECK

b = 2,5 t/m3 major. 2 to 3 %

48 R. Lacroix

JUNCTION AT MIDDLE SPAN

Cantilever built bridges 49

CONSTRUCTION OF END SPANS

50 R. Lacroix

ACTION OF PRESTRESSING TENDONS

isostatic
tendons

contin&:Ji_ty
tendons

effect of
prestressing
Cantilever built bridges 51

EFFECT OF PERMANENT LOADS

52 R. Lacroix

EFFECT OF OPERATING LOADS

l
Cantilever built bridges 53

EFFECT OF THERMAL GRADIENT

soc+ G +a } with Ei
10° C + G

& 1
54 R. Lacroix

---------

EFFECT OF CONCRETE CREEP

1
5f-- -
2

. { 1 MPa prefa
. margm
compression
1.5 MPa in situ
 COMPOSITE BRIDGES
NEW DEVELOPMENTS IN EUROPE

P. Dubas
Swiss Federal Institute of Technology, Zurich, Svizzera

ABSTRACT

The present paper deals with new developments for composite bridges, as they have been
opened recently to traffic. Firstly some indications are given for the materials used, i.e.
weathering steel as an example for structural steel, and concrete e.g. lightweight concrete.
The main part of the paper is devoted to the structural arrangement of composite bridges,
both with deck at lower flange level and at upper flange level. For this second arrangement
the relative advantages of structures with open framing or closed framing respectively
(particularly for bridges curved in plane) are described and illustrated by existing com-
posite bridges. Some aspects of the design procedures, especially concerning the web be-
haviour in the postcritical plate buckling range, are considered. The usual methods for the
construction of composite bridges, more precisely for the concreting of the deck (deck slab
cast in situ, precast deck slabs, stage-deck jacking along the steel girders) are mentioned.
The last short section of the paper deals with the possible widening of existing composite
bridges, due to increased traffic.
 56 P. Dubas

1• INTRODUCTION

The paper deals with new developments for composite girder bridges. The range of
spans considered is therefore limited to about 150 m. Composite construction, however, is
also applied for cable stay bridges, for example the Annacis Bridge in the Vancouver area
with a 465 m main span [1 ]. For further information to such structures reference can also
be made to the contribution of R. Walther.
The following paper concerns mainly highway or motorway bridges open~d recently to
traffic. Railway bridges require special considerations, due to heavier loading and to the
reduced deck width.

2. MATERIALS

2.1 Structural steel

In some countries weathering steel is widely used: the reduction in maintenance costs
is mostly higher than the increase in steel price. Fig. 1 shows as an application the ele-
vation of the main span of the Napoltlon Bridge near Brig (outside girder). In the hogging
moment region a high strength weathering steel with a yield stress of 41 0 N/mm2, made in
Italy, has been used, allowing a reduction of the cross-sectional area of the flanges and of
the corresponding welding labour. The webs are relatively thick for the reason explained in
section 4.2.

800 800 800 800

650-40 650-45 650-25

18 16 18

800-60 800-45
stiffeners
83800mm

Fig. 1, Elevation of the main span of the Napol~on Bridge (CH)

During the formation of the protective layer brownish trail may make the concrete
piles dirty. This can be avoided by lateral guidance and water moulding on the bottom
flanges.

2.2 Concrete
For the deck slab a concrete with a 28-day cube strength of 40-45 N/mm2 is gene-
rally used. Lightweight concrete, with an air dry density of 1800 kgtm3 (reinforcement
included) is seldom competitive, due to the higher price of lightweight aggregate. Such a
concrete may be interesting for long spans, e.g. 174 m for the Tay bridge at Friarton
Composite bridges 57

(1977), for which the 30% decrease in self weight of the deck takes a prominent part. For
the bridge at Eptingen (CH) a concrete structure destroyed during construction by a land-
slide had to be rebuilt without one of the two piles. This results in two very different spans
of 71 m and 37 m. The concrete deck of the new composite structure is therefore pre-
stressed by jacking up at the abutment (37 m span).

3. STRUCTURAL ARRANGEMENT

3. 1 Deck at lower flange level

A deck at lower flange level is required when the height as disposal is limited. Fig. 2
shows a simply supported bridge with a 41 m span. The structure, also here in weathering
steel, comprises two tapered main girders and composite cross-girders. Profiled steel
sheeting acts as both formwork and tensile reinforcement. This solution is widely used for
composite floor construction in buildings. For bridges, however, fatigue must be included-
specially for the connection between steel and concrete - so that this solution may be ap-
plied rather for minor roads.
For higher span lengths, e.g. 88 m for the combined railway and highway bridge over
the Rhone shown in Fig. 3, a bowstring (Langer girder) is interesting. The footways are
located outside of the girders so that the span of the cross-girders is reduced. No upper
bracing is provided: the arch boxes, rigidly connected to stiff end cross-girders are wide
enough to insure their lateral stability. The composite cross-girders guarantee the re-
quired vertical stiffness with a limited deck depth. The concrete slab acts also as tie-rod
for a part of the horizontal component of the arch forces.

Fig. 2, Girder bridge over the Sarine Fig. 3, Arch bridge over the Rhone (CH)
river (CH)

For a similar bridge, some kilometers upstream, the structure comprises steel
arches and a deck with concrete lateral beams acting as stiffening system. This hybrid so-
lution, with steel in compression and concrete in tension, may seem odd. The advantage is
the easy erection of the arches, without expensive shuttering.

3. 2 Deck at upper flange level

For bridges with composite main girders the deck is always located at upper flange
level. The structure comprises generally two main girders only, even for a wide deck. In
58 P. Dubas

the beginning, the slab was supported also in transverse direction by closely spaced cross-
girders. Over interior supports this arrangement leads to tensile stresses in concrete due
both to the hQgging moments of the main girders and to the local plate bending. A longitudi-
nal prestressing is then often required.
At the present time the concrete deck spans between the main girders only. This
arrangement is possible also for wide highway bridges, e.g. 20 m for the bridge over the
river Oise shown in Fig.12, with a 250 mm thick slab in the central region. With due re-
gard to the large cantilevers, a transverse prestressing is provided which reduces the de-
flection due to creep and also the crack wide under the dead load; for economic reasons this
solution is generally provided only for decks cantilevering more than 4 m. The cross-gir-
ders act here as cross-frames only, i.e. they prevent a distortion of the bridge sections and
transmit the wind blowing on the lower girder part to the deck acting as wind brace.
For straight bridges an open-framed system is mostly adequate and more economic.
The torques due to a traffic loading on a half wide of the deck are balanced mostly by an anti-
symmetric bending of the two girders, i.e. by warping torsion. For a simply supported
structure with a span r, a bending stiffness Elx for one of the two main girders and a tor-
sional stiffness GK of the deck, the straight line giving the transverse load distribution
derives from the following two ordinates in the axis of the girders, spaced with a distance
2c:

with m2 = (GK/2EixHr /2c)2

This relation applies exactly for a sinusoidal longitudinal loading, but is also appro-
priate for a unifom loading (see for comparison the exact formula in [2]). For a negligible
torsional stiffness GK the distribution follows obviouly from the level principle, with the
ordinates 1.0 and 0 . Fig. 4 shows as an application the distribution line for a continuous
motorway bridge with a 60 m span, i.e. about 50 m for the equivalent simply supported
structure.
1.25 1.25
1· ·1· "I [ml
wLooding1

Loading 2
~ JJJJll I I I i foldi"Jj I I I I I i
1. [Link] .1. 7.45 .1
7.00

83 ~ +
d d

Fig. 4, Transverse load distribution for a composite bridge

Composite bridges 59

The loading corresponds to the new Swiss Code SIA 160/1989. The decrease in girder
bending, in comparison with a distribution according to the level principle, is of the order
of 10% for the sagging moments. The corresponding torsional moments in the deck plate
require generally no increase for the reinforcement.

3.3 Bridges curved in plane with open framing

For curved bridges built some twenty years ago in Switzerland a closed framing was
generally adopted, even for relatively high ratios of the radius of curvature to the span
length. For moderate spans, less than say 100 m, a torsional brace at bottom level was pro-
vided, whereas a proper box section is economic for higher spans only. The increase in la-
bour costs in construction, however, is important, mainly due to the gussets welded to the
bottom flanges which require a well-finished fabrication to avoid fatigue cracking.
For this reason, open framing is adopted at the present time even for low values of the
radius of curvature. The distributed torque (kN·rnlm) M/r generated by the tangential
component of the bending moments M are also here (see section 3.2) balanced by an in-
creased loading on the outside girder and a decreased loading on the inside girder (for sag-
ging moments). When the length of the two concentric main girders is nearly the same, i.e.
when the spacing 2c between the girders is low in comparison with the radius r , the
additional girder load due to the bending moment M0 of the outside girder and the corres-
ponding value Mi for the inside girder is given as

.iq = ± (Mo + Mi)/2cr

The primary bending moments and the secondary moments due to the curvature are
shown in Fig. 5, both for a simply supported structure and for a continuous bridge with
equal spans and a uniform load. The ratio of the additional moments to the primary moments
depends obviously on the non-dimensional parameter f 2t2cr . For the continuous struc-
ture, however, the additional moments are very small because the forces .iq change the
sign at the points of contraflexure. For the limit of application of the approximate analysis
given in Fig. 5 reference can be made to [3].

qt21'-.. q~2 A
-wb">.~~
M~
(for one girder) 8

~ .,. . .- =:~
+720-I"C c==---.... c;;;;:: ' +1..~
? 4

~
M(Aq) ± 384.,.. .
upper sign-outside girder - 8 720·r·c

Fig. 5, Bending moments of open-framed curved bridges

Fifteen years ago a value f 2f2cr s 1 was considered as a limit for a curved open-
framed structure. For the Napoleon bridge shown on Fig. 6, with a main span of 83 m, a
radius of curvature of 400 m and a girder spacing 2c of only 5 m dictated mainly by
geotechnical conditions, the parameter takes the value 3.4. The simplified method given in
Fig. 5 is obviously not applicable here and the structure has been calculated as girder
grillage.
60 P. Dubas

13m

Fig. 6, Napoleon Bridge near Brig

The same model has been adopted for the bridge shown in Fig. 7, with a 32m main
span and an average radius of curvature of 130 m. The axis of the structure is S-shaped,
with skew supports and a junction for a minor road in the main span. The cross-frames are
rolled profiles arranged radially to simplify their connection to the main girders, located
at different levels in each cross-section according to the transverse deck slope.

Fig. 7, Maladeire Bridge (CH) Fig. 8, La Tine Bridge (CH)

 Composite bridges 61

3.4 Bridges curved in plane with closed framing

For another bridge in a mountainous region (see Fig. 8) the main span reaches
47.5 m, with a radius r = 100 m and a girder spacing of 5 m. The torques due to curva- -
ture are therefore relatively high so that a closed-framed structure was required, with
strong bracing diagonals at bottom level and cross braces.
For the Goldswil viaduct near Interlaken (Fig. 9) mainly aesthetic reasons have led to
a "trough" type of box with trapezoidal section and a constant height in spite of the span
lengths varying beetween 81 m over the river Aare, 71 m over the railway station and
45 m for the intermediate spans. During erection a top bracing equilibrates the shear flow
due to the torques M/r mentioned in section 3.3, with a radius r = 680 m. The need for
easy longitudinal placement of the interior part of the shuttering for the slab has led to a
plane of the bracing located at about 0.5 m below the level of the top flanges. For the cor-
responding calculation of the transverse braces reference can be made to [3].

i 1 r
30.65 45 50
r D Railway D

71.10 1
45
j 45
529.75m
! ! !
Aare

81 65 55
1 ?j
4 6

11 OOm

~-='~
jr 2.80 jr
Fig. 9, Goldswil viaduct near Interlaken (CH)

3.5 Bridges with several main girders

An arrangement with more than two girders may be adequate for very large bridges,
e.g. for the bridge over the Limmat river at Zurich (Fig. 10). The new structure is more
or less a duplicate of the old one, a solution required by a public voting. The structure
forms a girder grillage, well adapted for the relatively heavy loading due to the tramway
lines. The erection was made by a transverse launching of the new and the old structure tied
together, so that only a very short traffic interruption was needed.
For the replacement of an old trussed bridge spanning the river Ticino the new com-
posite structure has been built in two stages, each half-bridge consisiting of two steel gir-
ders with the corresponding precast concrete deck.
62 P. Dubas

Fig. 10, Bridge over the Lim mat Fig. 11, Railway bridge near Disantis
river at Zurich

3. 6 Railway bridges
When the height at disposal is very limited, a "trough" box girder with a relatively
large bottom flange plate gives an adequate solution for a railway bridge, in spite of the
higher labour costs. Fig. 11 shows a simply supported structure with a 54 m span during
construction. At the abutments the width of the trapezoidal box is increased to insure the
stability of this curved structure considered as a rigid body. For a "trough" box as a com-
bined railway-higway bridge with large span lengths reference can be made to [4].
The bridge over the Limmat river below Zurich replaces two old trussed bridges [5).
The arrangement with only two girders for a double track railway leads on the one hand to a
reduced steel weight, on the other hand to an increased depth of the deck slab (400 mm).
The structure is open-framed, mainly owing to the fatigue requirements for the bottom
flanges (no gussets required).
In Germany also some composite truss girder bridges have been built for the high
speed new railways lines, due to severe deflection requirements. Such conditions can be
satisfied even by simply supported truss girders, a solution avoiding cracks in the concrete
deck at support. For more details reference can be made to [6].

4. DESIGN PROCEDURES

4. 1 Analysis of structures
In general both the internal force resultants and the section capacities are calculated
by elastic theory. The cross-sections are indeed too slender and allow no plastic hinge ro-
tation, furthemore a negligible stress redistribution only. EUROCODE 4 will give some in-
dications for the modified moment distribution due to concrete cracking in hogging moment
regions.
In sagging moment regions the steel girder is in the tension zone so that no buckling
problem occurs: the cross-sections are therefore compact and the bending capacity may be
determined here by simple plastic theory when the increase of hogging moments due to
fully plastified cross-sections in sagging moment regions is considered in design (7). This
Composite bridges 63

solution is economic because the loading history must not be taken into account, i.e. a
differentiation between loads applied to the steel section prior to the development of com-
posite action and those applied to the composite section is no more needed.

4.2 Postcritical behaviour of thin-webbed plate girders

A design based on the classical plate buckling theory requires longitudinal stiffeners
for a girder depth of say more than 2 m and an adequate web thickness. When the postcriti-
cal plate behaviour is accounted for, on the contrary, Fig. 1 shows that webs with trans-
verse stiffeners only are suitable even for girders up to 4 m (for a bridge in Luxemburg
with a main span of 92 m and a girder depth of 4 m, designed as well according to the Swiss
code SIA 161/1979, see [8]). The web thickness is obviously higher than for a longitudi-
nally stiffened web but the labour costs are largely lower.
The bending capacity of such girders can be calculated by replacing the compression
zone be of the web by an effective width b8 which is inversely proportional to the plate
reference slenderness 1p = ..Jfylacr (fy = yield stress; acr = critical plate bending stress
according linear theory, i.e. acr = ka·0.9·E·(tw/b)2 for a web depth b). This method con-
siders therefore implicitly a shedding of the bending stresses from the web to the compres-
sion flange, so that the flange capacity is the leading factor for the ultimate strength. The
classical buckling theory, on the contrary, evaluates the bending capacity of the whole gir-
der by considering the critical stress acr of the web only.
Similarly the shear capacity of thin webs is given by a tension field model, with the
transverse stiffeners acting as compression posts. For more details concerning the post-
critical plate behaviour reference can be made to ECCS publication no. 44 [9] and to
EUROCODE 3. Composite girders require special considerations; they will be given in the
final draft of EUROCODE 4, not available at the present time.

4. 3 Spacing of transverse stiffeners

The spacing between the transverse stiffeners is very important from an economic
point of view. Both for the Napoleon bridge (Fig. 6) and for the Goldswil viaduct (Fig. 9) it
was intended at the beginning to have relative low aspect ratios, in the order of 1.0, i.e. to
introduce intermediate transverse stiffeners between the cross-braces. Comparative stu-
dies conducted for the two bridges have proved that an arrangement with thicker webs, say
2 mm more on an average to reach the required shear strength, and an aspect ratio of at
least 2 is better: the decrease in the labour costs is namely higher than the increase of ma-
terial supply.

5• CONSTRUCTION OF COMPOSITE BRIDGES

5 . 1 Deck slab cast in situ

The choice of an optimum method for the concreting of the deck is essential for the
economic efficiency of a composite bridge. A conventional wooden scaffolding over the full
length may be interesting for particular conditions. In situ concrete, however, is generally
poured using a movable scaffolding, as shown in Fig 12 for the bridge over the river Oise
already mentioned in section 3.2 (span lengths 45 m). The interior part of the shuttering
will be slided on the flanges of the cross-frames to that these elements must be located at a
suitable level. A deck length of about 15 m can be concreted each week.
When the deck is poured continuously in this manner the problem of concrete
cracking in the hogging moment region needs careful! consideration: the part ot the slab
 64 P. Dubas

cast one or more weeks before is green and has a very small tension strength, so that the
hogging moments due to the concrete weight in the span may cause cracking.

Fig. 12, Bridge over the river Oise (F) Fig. 13, Stage deck jacking for the Sorge
with movable scaffolding bridge (CH)

For the Goldswil viaduct (Fig. 9) two measures have been adopted to avoid cracks as
much as possible and therefore to increase serviceability. On the one hand the connectors
are arranged in groups and the corresponding zones in the slab are not concreted immedi-
ately. All 60 m transverse joints remain open to allow for shrinkage and other movements.
On the other hand the deck is prestressed in the longitudinal direction, especially over the
intermediate supports, with an initial concrete compression in the order of 2-4 N/mm2 .
The transverse joints are poured after as long as possible a time so that the prestressing
decrease due to creep with corresponding hyperstatic moments can be limited.

5. 2 Precast deck slabs

Precast deck slabs are today less utilized than twenty years ago. The advantages are
well-known, especially the concrete weight will surely be supported by the steel girders
only, avoiding therefore cracking during concreting. The presence of many transverse
joints between the slabs (spacing 2-3 m), however, complicates the compliance with the
serviceability requirements in the hogging moment region. This solution remains inter-
esting for special conditions, e.g. for the two bridges in mountainous regions shown in Fig.
7 and 8 (se also (4]). For various laying procedures reference can be made to (1 0].

5.3 Stage-deck jacking along the steel girders

The concreting area is located here generally behind the abutment - for a long bridge
on the span. In the first case the adjacent span must be propped to reduce the slope at the
abutment (Fig. 13); moreover equilibrium conditions for the friction forces require that
the steel structure and the formwork are connected together.
The launching method for the deck combines the advantages of in situ concreting (no
transverse joints) with those of precasting. Obviously all possible hazard scenarios must
be considered. The main problems to be solved are the stability of the top flanges, taking
into account transverse forces during launching, and the effect of possible eccentricities of
the vertical loads. Last but not least the friction behaviour has to be carefully investigated.
On the one hand sliding devices must be provided at the interface, e.g. pads introduced in
 Composite bridges 65

some of the openings for the connectors. A lubrification consisting of colloidal graphite
suspension allows to reduce the friction according to the jack capacity. For bridges with a
longitudinal slope, however, the friction must be sufficient to stabilize the deck. An
accident has shown that the friction coefficient can be as low as 6% (see for example
[ 11]).

5. 4 Erection of the steel girders

The erection procedures for the girders are the same as for steel bridges: the compo-
site action obviously does not play a role in this topic. For the viaduct at Goldswil (see
Fig. 9), as an example, the curved southern part has firstly be launched over the railway
station, without any traffic interruption, and the two adjacent 45 m spans have been
assembled at ground level and erected by crane, each span as a unit (Fig. 14). Finally the
northern part has been also erected by launching, using the end span, curved and skew, as
assembly site.

Fig. 15, Demolishment of the old deck

Fig. 14, Erection of the viaduct at Goldswil

(Figure on the right)

6. WIDENING OF EXISTING BRIDGES

Due to increased traffic bridges must in some circumstances be widened, as this was
required for the Austrian section of the Brenner motorway. Fig. 15 shows the demolish-
ment of the old concrete deck of the Gschnitztal viaduct (span lengths about 80 m) during
winter time, the traffic being then supported by the parallel bridge only. The new deck is
widened on the outside part of the section so that a special lane for heavy vehicles can be
provided. With the use of transverse prestressing its weight is not increased in comparison
to the old one. Only the torsional brace at under flange level has to be strengthened due to
the increased torques resulting from the eccentricity between the new deck and the axis of
the steel structure. The new slab has been concreted using a movable scaffolding as men-
tioned in section 5.1, with the area in the hogging moment region poured in a final stage to
surely avoid cracks due to the self-weight.
66 P. Dubas

FINAL REMARKS

The present paper is a kind of extended summary of the conference presented in French on
15th June 1989 at Udine. Specially the number of figures compared with the slides shown
during the conference has been drastically reduced due to the allowed number of pages. For
some of these cancelled figures, references have been included which give detailed informa-
tion together with illustration. Moreover reference can be made to the publication series
"Bauen in Stahl" edited by "Schweizerische Zentralstelle fOr Stahlbau", CH-8034 ZOrich.
See for example the numbers 511982 (BrOcke Ober den Tessin bei Quartino), 14/1983
(Viadukt Goldswil), 23/1985 (NapoleonsbrOcke Brig), 14/1988 (Pont sur le RhOne a
Saint- Triphon).

REFERENCES

1. Stiemer, S.F., Taylor, P. and D.H.C. Vincent: Full scale dynamic testing of the Annacis
Bridge. IABSE Proceedings P-122188 (1988).
2. Dubas, P.: Developpements suisses recants en matiere de ponts mixtes acier-beton.
Costruzioni metalliche 21 (1969) n. 1, 1-15.
3. Dubas, P.: Some remarks concerning the design of composite bridges. J. Construct.
Steel Research 7 (1987) 233-251.
4. Mason, J.: VerbundbrOcke Ober den Fluss Tocantins in Nordbrasilien. Stahlbau 58
(1989) 129-135.
5 Gut, H. und E. Graf: Neubau Obere Ummatbrocke der SBB. Schweizer lngenieur und
Architekt 107 (1989) 240-244.
6. Keller, N., Kahmann, R. und M. Krlps: Fuldabrucke Kragenhof. Bauingenieur 63
(1988) 443-454.
7. Roik, K., Bode, H. and J. Haensel: Erlauterungen zu den "Richtlinien fOr die Bemessung
und AusfOhrung von StahlverbundbrOcken" - Anwendungsbeispiele". lnstitut fOr
konstruktiven lngenieurbau, Ruhr-Universitat Bochum, Mitt. Nr. 75-11, 1975.
8. Molitor, Ley et Geisen: Viaduc sur le Sernigerbach (Luxembourg). IABSE Structures
C-33/85 (1985) 16-17.
9. European Convention for Constructional Steelwork, TWG 8.3: Behaviour and design of
steel plated structures. Publication no. 44, Edited by P. Dubas and E. Gehri, Zurich,
Swiss Federal Institute of Technology, 1986.
1 o. a a
Dubas, P.: Problemas relatifs Ia conception et Ia realisation des ponts mixtes
acier-beton. Revue Technique Luxembourgeoise 80 (1978) 89-100.
11. Bourquin, G.H.: Composite bridge decking by stage-deck jacking. Journal of the
Structural Division, ASCE 104 (1978) No. ST1, 171-189.
12. Tschemmernegg, F., Passer, H. und 0. Neuber: Verbreiterung und Sanierung von
StahlbrOcken. Stahlbau 58 (1989) 289-298.
 STRUCTURAL SYSTEMS FOR CABLE SUSPENDED BRIDGES

N.J. Gimsing
Technical University of Denmark, Copenhagen, Danimarca

ABSTRACT

The paper describes a number of the features related to the

choice of the structural system for cable suspended bridges,
i.e. cable stayed bridges and suspension bridges.

For the cable stayed bridges most of the text deals with the
conventional three span arrangement found in the major part
of the existing cable suspended bridges, whereas the
problems related to a multispan arrangement are treated for
both the cable stayed and the suspension system.
68 N. J. Gimsing

INTRODUCTION

During its three decade history of evolution the cable

stayed bridge has probably appeared in a larger variety of
forms than any other bridge type. This is due to the large
degree of geometrical freedom associated with the cable
stayed system.

In recent years the many variants have, however, shown a

tendancy to crystalize into a limited number of basic types
that have proven to be superior to many of the early
concepts.

l. THREE SPAN CABLE STAYED BRIDGES

The major part of the existing cable stayed bridges are

of the three span configuration with a large main span
flanked by two smaller side spans, e.g. the same span
arrangement as found in most suspension bridges.

1.1 General Arrangement of Cable Planes

The aim of the cable system is to assist the girder in

transmitting the loads acting on the structure to the
supports.

The loads that have to be considered when choosing the

arrangement of the cable planes are : The vertical loads Pv,
the lateral loads Pl and the torsional moments Mt, as shown
on Figure 1.
Mt

Figure 1. Resulting forces acting on the stiffening

girder.
Cable suspended bridges 69

(a) (b) (c)

I
I
I I
l...._____.l
Figure 2. Bridges with one or two cable planes and
with or without torsional stiffness of the girder.

For road bridges with dual carriageways a single, central

cable plane has frequently been applied (Figure 2(a)). With
this arrangement the cable system only contributes by
carrying the vertical loads Pv globally, whereas the girder
has to transfer the entire lateral load Pl and the torsional
moments Mt. Consequently, the system with a central cable
plane requires a girder with large lateral and torsional
rigidity.

With two vertical cable planes attached along the edges

of the roadway both the vertical loads Pv and the torsional
moments Mt can be transferred by the cable system, so that
only the lateral loads Pl have to be carried by the girder
(Figure 2(b)). Consequently, no torsional stiffness is
required and an open cross section is therefore adequate.

If, however, a double cable plane system is combined with

a torsionally rigid girder (Figure 2(c)) the torsional
moment Mt will be partly transferred by the cable system and
partly by the girder. The vertical and lateral loads will be
transferred in the same way as for the system with no
torsional stiffness of the girder.

.I \
I \
J.:'~-=.---~--M-t-4j.\_"-=5==-- -=·-~f1 lhit
1/2 Pv
~ - ----------~
1/2 Pv
Figure 3. Lateral forces induced by torsional moment
transferred by inclined cable planes.
70 N. J. Gimsing

With two laterally inclined cable planes the transmissio n

of the vertical loads Pv will basically be as for the system
with two vertical cable planes.

For a bridge with no torsional girder stiffness the

lateral loads Pl will have to be carried by the girder, and
the torsional moments Mt by the cable system. However, due
to the inclination of the cable planes the transfer of
torsional moments by the cable system will induce additional
lateral loads in the girder, as illustrated in Figure 3.

Finally, if the laterally inclined cable planes are

combined with a torsionally rigid girder the transmissio n of
the lateral loads Pl and the torsional moments Mt becomes
more complex. Thus, the lateral load Pl will partly be
carried by lateral bending of the girder and partly by the
cable system due to the resistance against rotation offered
by the girder (Figure 4). Similarily, the torsional moments
will also induce latera~ loads.
.
I

.I \.

I ~
-Py0 ______ ___ ~ Py
. b
Mt =Pyx b
Figure 4. Torsional moment induced by transmissio n of
lateral forces to inclined cable planes.

As described above the participati on of the cable system

in transmittin g the loads acting on the girder depends to a
large extent on the arrangement of the cable planes.

This feat~re is further illustrated in Figure 5 showing

systems rank1ng from the pure box girder bridge where all
loads naturally have to be transferred by the girder to a
system with four inclined cable planes allowing all the
loads Pv, Pl and Mt to be carried by the cable system.
So far the cable systems able to transmit the lateral
load Pl has not been used in road bridges but only in a
limited number of pipeline bridges. However, cable systems
Cable suspended bridges 71

adding lateral stiffness could also be attractive for road

bridges with a modest width-to-span ratio .

LOAD TRANSFER BY' I PYLON

TYPE OF STRUCTURE SHAPE
CABLE GIRDER '
SYSTEM (TYPICAL)

rr
PURE BOX GIRDER
Pv, PI, Mt

l
I
i
i
=b Pv PI,Mt

CENTRAL CABLE PLANE

I I
i !I

11
i

1 i
I
Pv, Mt PI

TWO VERTICAL CABLE PLANES

M
I I
\ i
\/
u Pv, PI Mt

TWO INCLINED CABLE PLANES

I I\ I

\ ii .\ i =
I \ i
') Pv, PI, Mt
[,v;]
r ·vv
Y' . t""1.

----'
FOUR INCLINED CABLE PLANES

Figure 5 . Force transfer and cable plane arrangement.

1 . 2 Configuration of the Cable System

In the early cable stayed bridges from the [Link] and

[Link] the cable systems generally contained a very limited
number of stay cables (Figure 6(a)). This led to large
forces in each stay cable and consequently a multi-strand
cross section was required. In the present practice the
preference is for a multi cable system (Figure 6(b))
comprising mono-strand stay cables. This is partly due to
the advantages of a more continuous support of the girder by
the cable system and partly to the easier installation and
replacement of stay cables. However, the · indisputable
advantages gained by introducing the multi cable system has
also been followed by some drawbacks . Thus, the problems
72 N. J. Gimsing

(a)

~ ~I
(b)

~~
Figure 6. Cable systems with few or many stay cables.
II

related to wind induced oscillations of the individual stay

cables have become more pronounced for mono-strand cables
than for multi-strand cables.

The preferred cable configuration is the modified fan

shown in Figure 6(b). Here the cable anchorages in the pylon
are concentrated at the pylon top and spaced as closely as
the installation and force transmission allow. By
concentrating the cable anchorages at the pylon top it is
achieved that the moments in the pylon due to transmission
of unbalanced horizontal forces are kept at ~ minimum
(Figure 7).
Side Span 1 Main Span

PYLON UNBALANCED MOMENT DIAGRAM

HORIZONTAL FORCES

Figure 7. Moments in the pylon induced by unbalanced

horizontal cable components.
Cable suspended bridges 73

To ensure a reasonable tension in the backstay cable, the

length of the side span has to be chosen to somewhat less
than half of the main span length. With a constant spacing
of the cable anchorages in the girder this implies that the
number of cables leading to the side span will be smaller
than the number leading to the corresponding main span half.
To counteract this difference the back stay cable is in most
cases composed of a number of parallel mono-strand cables,
as shown in Figure 8(c). Alternatively, the two fans
radiating from the pylon can be made symmetrical by
extending the cable anchorages into the span adjacent to the
side span, as shown in Figure 8(d). However, with this
system the girder will be subjected to a substantial bending
in the region at the anchor pier.

(c)~~

(~
Figure 8. Asymmetrical and symmetrical fan
arrangement.

The harp system comprising parallel stay cables and

therefore a spreading of the pylon anchorages over the
entire pylon height - might be preferred if the pylon is
made with a large bending stiffness in the longitudinal
direction (Figure 9(e)) or if the girder has intermediate
supports in the side span (Figure 9(f)). As the purpose of
the intermediate supports is to eliminate vertical
deflections of the side span cable anchorages, they must be

(e)~~

Figure 9. Harp shaped cable systems stabilized by

stiff pylons or intermediate supports.
74 N. J. Gimsing

placed so closely that the girder only deflects

insignificantly between the supporting points. This
requirement will be fulfilled if the distance between
intermediate supports is chosen to less than approximately
20 times the girder depth.

In the modern cable stayed bridges the stay cables will

always be fixed both vertically and horizontally to the
pylon whether the cable system is of the modified fan or the
harp configuration.
2. LIMITING SPAN OF CABLE STAYED BRIDGES

Since its introduction in the [Link] the cable-stayed

bridge has gradually increased its free span from 183m in
the Stromsund Bridge to 465m in the Annacis Island Bridge,
and at present several cable-stayed bridges with spans
around 500m are under construction.

Having witnessed this evolution it is natural to ask:

What is the maximum span that can be reached with the cable-
stayed bridge?

To answer this question it is necessary to consider each

structural element and determine which ones will ultimately
determine the maximum length of the free span.
2.1. The Stay cable

The stay cable is the structural element that primarily

characterizes the cable stayed bridge and it is therefore
reasonable to first consider how this element behaves with
increasing span length.

First of all, the load-carrying capacity of a stay cable

is influenced by the sag - illustrated to the left in Figure
10 - and as the relative sag increases with the length of
the stay cable the load carrying capacity will decrease.

This is illustrated to the right in Figure 10 where the

efficiency ratio (defined as the ratio between the force To
in a straight cable and the force T in the real, sagging
cable) is plotted against the horizontal cable projection.

The ultimate length of the stay cable is obviously the

length where the efficiency ratio becomes zero, which
corresponds to a case where the cable tangent at the bottom
is horizontal.

Based on realistic assumptions regarding design stress

level and density of the cable material the ultimate lengths
Cable suspended bridges 75

To/T (Efficiency Ratio)

3000m 4500m

Figure 10. Load carrying capacity of a stay cable as

expressed by the efficiency ratio To/T.

will be approximately 3000m for a chord inclination of 0.25

and 4500m for a chord inclination of 0.5.

In real structures it will probably not be acceptable ·to

have cables with an efficiency ratio of less than 0.5 as the
stay cable naturally also has to carry other loads than its
own weight. But even for an efficiency ratio of 0.5 the
maximum horizontal stay cable length becomes l500-2400m,
which corresponds to main span lengths of 3000-4800m. Thus,
the load carrying capacity of the stay cables allows a most
significant extension of the present cable stayed bridge
spans.
The sag also results in a reduction of the axial
stiffness of the stay cable as it can be expressed by the
ratio between the equivalent modulus Eeq (Ref.3) and the
real modulus of elasticity E (Figure 11).
E..IE

a
650m

Figure 11. Stiffness of a stay cable expressed by the

ratio Eeq/E.
76 N. J. Gimsing

It is in general difficult to specify the allowable

reduction in axial stiffness for a given case, as the static
and dy~amic behaviour of the entire structure is influenced
by the flexibility of the cables.

If the limiting ratio between Eeq and E is stipulated at

0.8 the maximum horizontal stay cable length becomes
approximately 650m corresponding to a main span length of
1300m. Thus, the stiffness requirements can easily be more
restrictive than the strength requirements.

The stay cable length might also be limited by

considerations regarding dynamic behaviour, and already in
some of the existing cable-stayed bridges the stay cable
oscillations have presented a problem. However, in the span
range found to-day the undesirable dynamic actions have been
eliminated by introducing relatively simple stabilizing or
damping devices.

Serious dynamic effects would undoubtedly occur if the

natural frequencies of the lower modes of the individual
stay cable oscillations coincided with the natural frequency
of the total system characterized by vibrations of the
entire stiffening girder and the towers.

As the natural frequency of the individual stay cable

decreases with the span length, ressonance will inevitably
occur for a certain cable length. With typical values the
frequencies of the first individual mode and the global mode
will coincide for cable lengths between 600m and 1200m
(Figure 12). Consequently, also the dynamic action might
easily be more decisive than the load-carrying capacity of
the stay cable .

NATURAL FREQUENCY
FIRST MODE:
FIRST MODE
SECOND MODE

TYPICAL RANGE OF
LOWER FREQUENCIES
FOR MAIN SYSTEM
SECOND MODE:

L
300m 600m
-
Figure 12. Ressonance between oscillation of the
individual stays and the global system.
Cable suspended bridges 77

An overall improvement of the system can therefore be

achieved by adding secondary trajectory cables, as shown in
Figure 13. Such secondary cables will significantly reduce
the sag variations of the primary stay cables under varying
axial forces and they will restrain free oscillations of the
single cable. Consequently, both the axial stiffness and the
dynamic behaviour of the single cable will benefit from an
addition of the trajectory cables.
TRAJECTORY CABLES

IMPROVES: REDUCES:

STIFFNESS OF STAY CABLES OSCILLATION OF INDIVIDUAL STAYS

LOAD CARRING CAPACITY OF STAYS

Figure 13. Application of secondary trajectory
cables.
To be efficient the trajectory cables have to pull down
the stay cables which involves a reduction of inclination of
the stay cable tangent at the bottom and an increased
tangent inclination at the top. This will unfavourably
influence the load-carrying capacity of the stay cable, but
this effect is of minor importance due to the fact that this
parameter is less restrictive than the other two.

2.2. The Stiffening Girder

In the cable stayed bridge the stiffening girder

contributes to the global load transfer by transmitting the
horizontal stay cable components as axial forces, and by
carrying the lateral wind load by bending to the supports.

The maximum axial force in the stiffening girder

obviously increases with the span length so that for larger
spans the cross sectional area As of the girder at the pylon
have to be chosen larger than the area Ao in the midspan
region.

If it is assumed (more or less arbitrarily) that it is

acceptable to have a 2,5 times larger girder cross section
at the pylon than at midspan, then the limiting length of
78 N. J. Gimsing

the half span (=horizontal stay cable length) will be

approximately 370m for a concrete bridge and 940m for a
steel bridge (Figure 14) - again based on realistic values
for densities and stress levels.
Ao/Ao

I
I
2.5 ·- ·- ·- ·- -,t-·- ·- ·- ·- ·- ·- ·- ·
/i i
CONCRETE 7 j i
I . I
/ ! .
/
/ .I I.
/
/ I.

J~ .
1.0
I
i
!
a
L ---- . !L _-
370m 940m

Figure 14. Increase in the cross sectional area of

the girder to allow transfer of axial forces.
Consequently, the axial strength of the stiffening girder
might be the limiting factor for a concrete bridge, whereas
it is less likely to be decisive for a steel bridge.

The lateral stability of the stiffening girder will

generally limit the free length of the cantilever arm during
erection (Figure 15). Thus, the length-to-width ratio of the
cantilever must be limited to a value between 15 and 25 to
avoid unacceptable dynamic actions . For a bridge with a

MAX alb - RATIO BETWEEN 15 AND 25

Figure 15. Lateral stability of the stiffening

girder.
Cable suspended bridges 79

roadway width of 12m this requirement might limit the span

length to as little as 360m.

The limiting length-to-width ratio mentioned above can

not be stipulated at a fixed value as it will depend on a
number of factors such as the local wind climate, the
efficiency of the temporary fixation to the pylon pier, the
moment of inertia of the stiffening girder and its degree of
streamlining.

To overcome the problem related to the lateral stability

of the stiffening girder a twin box arrangement as shown in
Figure 16 might be introduced. Here the total deck width b
is split in two parts each with a width of b/2 and joined by
transverse beams and horizontal cable diagonals. Thus, in
the lateral direction the stiffening girder will act as a
truss with the two boxes forming the chords.

Figure 16. Twin box arrangement to improve the

lateral stability.

The twin box concept was originally proposed by

Richardson (Ref.4) as a mean to improve the aerodynamic
stability of long span suspension bridges but it might very
well be even more efficient in improving the lateral
stability of of a cable stayed bridge where the cable system
(at least in the self anchored case) does not render any
lateral support to the girder.
80 N. J. Gimsing

3. STRUCTURAL SYSTEMS FOR MULTISPAN SUSPENDED BRIDGES.

The majority of all cable suspended bridges built so far

are made as three-span structures with a large main span
flanked by two shorter side spans.

In the cases where more than three cable suspended spans

have been required, it has often been chosen to apply two
(or more) conventional three-span structures in a row. This
was the case for the West Bay Crossing of the San Francisco-
Oakland Bay Bridge, and the same principle has recently been
used also in two of the Honshu-Shikoku Bridges: The Bisan
Seto Suspension Bridges and the Hitsuishijima and
Iwagurojima Cable Stayed Bridges.

The main reason for using three-span structures as a

basic element also for bridges with a larger number of spans
is undoubtedly that special problems arise if a true
multispan cable suspended bridge shall be designed.

A true multispan bridge will comprise a number of

consecutive main spans of equal size and shorter side spans
at the extreme ends only, as illustrated at the bottom of
Figure 17.

I::::1?--: -::::1?=::1

Figure 17. Three span and multi span cable suspended

bridges.

From a structural viewpoint the larges~ problems relate

to the lack of efficient longitudinal support of the inner
pylon tops, as they are not restrained by an anchor cable
leading to a point where the girder is supported vertically.

Generally, it can be stated that in multispan cable

suspended bridges the main effort must be concentrated at
achieving sufficient stiffness, as adequate strength is
easier obtained. However, the flexibility that will result
if the multispan structure is not carefully designed might
also have some side effects that influence the safety. Thus,
a flexible structure is more vulnerable to aerodynamic
actions and might easier be subjected to fatigue due to
larger stress variations under moving traffic loads.
Cable suspended bridges 81

3.1. Multispan Cable Stayed Bridges.

To improve the deformational characteristics of a

multispan cable stayed bridge two basic approaches can be
followed. One is to increase the bending stiffness of the
girder or the pylons, and the other to modify the cable
system.

If it is chosen to increase the bending stiffness of the

girder, significantly larger depth-to-span ratios than found
in the normal three-span bridges have to be used (Figure
18(b)), typically approx. three times larger (1/40 intead of
less than 1/100). This inevitably leads to a larger dead
load influencing unfavourably the dimensions of the stay
cables, the pylons and the substructure. Consequently, to
choose a larger stiffness of the girder often constitutes
the most expensive way to limit deflections of a multispan
cable stayed bridge.

li<J??=:
(a)

(b)

(c)
>

(d)

Figure 18. Multispan cable stayed bridges stabilized

by increased stiffness of the girder or the pylons.
For systems with pylons fixed to the piers the required
overall stiffness can be achieved by increasing the bending
stiffness of the inner pylons. This offers the advantage
that the girder depth can be kept unchanged, but as shown in
Ref.(l) the width of the pylons have to be increased by a
factor of approx. 5 (Figure 18(c)).
As this leads to disharmonic proportions it will o1fen be
preferred to apply triangular pylon structures, as shown in
Figure 18(d).
Note that for a self-anchored system, as generally used
in cable stayed bridges, the inner pylons should have their
82 N. J. Gimsing

legs inclined only above the girder, whereas the legs below
should be vertical. This is in contrast to the earth
anchored case where the pylon legs should be inclined all
the. way down to the top of the bridge pier.

Modifications of the cable system to give a more

efficient restraint of the inner pylon tops can be made as
shown in Figure 19(e)-(g).

Figure 19. Modified cable systems to stabilize

multispan cable stayed bridges.
System (e) is characterized by long anchor cables leading
from the tops of the inner pylons to the girder at the two
adjacent pylons, i.e. to points with efficient vertical
support.

Alternatively, the inner pylons might be restrained by a

horizontal tie cable connecting all tops, as indicated for
system (f).

Finally, the necessary stiffness might be achieved by

overlapping the fans at midspan, as shown in system (g).
This solution requires, however, that the number of main
spans is limited, probably to less than three.

If the systems (e)-(g) are made as self-anchored systems,

they all have to be made with a continous stiffening girder
from one end to the other. This is due to the fact that
there are no regions between the two end supports where the
axial force in the girder is zero. Consequently, the
distance between expansion joints might end up to be
unattractively large.

If partially anchored systems are used, both systems (f)

and (g) will allow expansion joints to be inserted in the
Cable suspended bridges 83

intermediate spans. Thus, if the continuous tie cable of

system (f) is earth anchored at the ends of the side spans,
the stiffening girder will in the dead load condition have
zero axial force at the center of all main spans.

Among the systems shown in Figures 18 and 19, system (d)

appears to be the easiest to erect as the partially erected
superstructure will be efficiently supported by the double
columns of the pylon structure (Figure 20). For the systems
(e), (f) and (g) a (costly) temporary fixation of the
superstructure to the pier is required as the cables
stabilizing the inner pylon tops can not be erected until
the entire stiffening girder is in place.

Figure 20. Erection of a multispan cable stayed

bridge with triangular pylon structures.
3.2 Traditional Single-cable Suspension Systems.

It is a wellknown fact that an earth anchored cable

system possesses an ability to deform into a configuration

(A) _______ H--- LL~--~_ _..u._____

t- L/2 -t---- L ---t-Lt2----+

<s>-- ~ r 4=-:: ~ ~
--~--L L L
e-

Figure 21. Structural systems for multispan

suspension bridges.
84 N. J. Gimsing

that will allow it to carry any applied load - as long as

all active cable elements are in tension. For a bridge with
a large number of consecutive spans a solution as shown
undeF (B) in Figure 21 would ,therefore, at first glance
seem appropriate. In this system, all pylons are of the
traditional column-type rendering only vertical support to
the cable system.

However, the deformations required to create equilibrium

in a loading case with traffic load in one main span only
are unrealistically large. For a bridge with a latge number
of spans the sag of the traffic loaded span will increase to
F(p+g)/g, where F is the sag in the dead load condition, p
the traffic load per unit length, and g the dead load per
unit length. Consequently, the deflection under maximum
traffic load will be (p/g)F, and this value becomes
unacceptably large with realistic sag ratios, e.g. with
p=0.2g and F=lOO meters the deflection would be 20 meters!

This feature is illustrated in Figure 22 showing the

midspan deflection as a function of the sag ratio for the
system (B) and for the reference system (A). It is seen that
in the actual case the sag ratio of system (B) should be
chosen to approx. 0.03 or 1/33 to arrive at the same midspan
deflection as for system (A) with the traditional sag ratio
of 1/10. Such a reduction of the sag ratio would, however,
for a bridge with spans of 1000 meters imply an increase of
the cable steel quantity per unit length by a factor of
approx. 4.5, which must be regarded as unacceptable due to
the associated cost increase.
Relative Midspan Deflection
I
t--system (B)

:::1------./-,-/-/ ~_-:---- -
System (A)

Sag Ratio
0.05 0.10 0.15 0.20
Figure 22. Relative flexibility of systems (A) and
(B) in figure 21.
Considerations along these lines probably have been-
decisive for the additional anchor pier in the central span
of the existing bridges mentioned in the introduction. For a
multispan bridge the principle of adding intermediate anchor
piers is illustrated in system (C) of Figure 21.
Cable suspended bridges 85

Elimination of the intermediate anchor piers can be

achieved - as in the cable stayed case - by stabilizing the
inner pylon tops in the longitudinal direction of the bridge
by a horizontal tie cable, as shown in system (D). However,
this solution is only applicable in bridges with moderate
spans as the horizontal cable otherwise will be
characterized by a too small equivalent modulus of
elasticity.
For large multispan suspension bridges with a single-
cable system the only solution left seems to be to apply
triangular pylon structures comprising two inclined legs, as
shown in system (E). Provided the main dimensions of the
pylon structures are chosen to give adequate rigidity, this
system can lead to deformational characteristics that are
even more favourable than those of the reference system (A).

The fixation of the pylon tops in the longitudinal

direction of the bridge significantly increases the
frequency of the first symmetrical mode of vibration that is
often critical for bridges with long side spans. Actually,
by fixing the pylon tops the first antisymmetrical mode will
often turn out to be the most critical. Thus, it might also
prove advantageous to take measures to increase the
frequency of the first antisymmetrical mode, such as
clamping the main cable to the girder at midspan and
restraining the stiffening girder at the pylons by hydraulic
dampers (shock absorbers). These devices would allow slow
movements due to temperature change but exclude fast
movements due to wind excited oscillations or moving traffic
loads.
The main disadvantage of system (E) is the increase in the
quantities to go into the pylon structures. Under maximum
horizontal force, occuring for one-sided traffic load, the
resulting force from the cable system will have to be
transferred almost entirely by the legs leaning towards the
loaded span, due to the inclination of the resulting force
from the cable system. It is ,therefore, to be expected that
the quantities to go into the pylon structure will be almost
doubled when changing from a vertical column-type pylon to a
triangular pylon structure.

Despite the disadvantages regarding the quantities of the

pylons, system (E) must be regarded as the most promising
among the traditional single-cable suspension systems.
3.3 Untraditional Double-cable Suspension Systems.

To improve the ability of the suspension system to

transfer nonuniform loading, it has in some cases been
proposed to use systems where each cable plane contains two
(or more) main cables with different geometrical
configurations.
86 N.J. Gimsing

A system as shown in Figure 23(J) has earlier been

proposed. The main attraction of this system should be that
each of the two continuous cables can be anchored to every
second pylon at the level of the bridge deck (point B in
the figure). However, this advantage must be paid by a
considerable increase in the quantity of cable steel
compared to the traditional system (I).

In the dead load condition there will be no forces acting

between the cable and the pylon at the anchor point B as the
horizontal force in the cable is constant and there is no
break in the cable curve at the pylon. Therefore, for dead
load the cable A-B-C will be equivalent to a free cable A-B
with a length that is twice the span of the bridge! At the
same time the sag ratio of this cable (F/2L) is only half
the sag ratio of the traditional cable (F/L). These features
make the system extremely unfavourable to carry the very
decisive dead load.

(I)

<J>
+--- -----r~;----
:t:===-~-=-=--f-biS):-'
1
------1-
=-==-=-=::·---==--t---=---11-
____ L L ------+

CM> ±s:;: ·~I I lfP1:

Figure 23. Double-cable suspension systems.
For a bridge with spans of 1000 meters each with a sag of
100 meters (=vertical distance between pylon top and bridge
deck), the cable stress from the self weight of the cables
Cable suspended bridges 87

will for system (J) amount to approx. 50% of the allowable

cable stress, leaving only the other 50% for carrying the
dead load of the stiffening girder and the traffic load. For
the traditional system, the cable stress from the self
weight will be equal to approx. 12% of the allowable cable
stress - a much more acceptable value.

The improved stiffness under traffic load that can be

observed for system (J) is mainly due to the reduced sag
ratios of the cables. Thus, under traffic load, where the
cables can be regarded as supported also at point B, the sag
of the cable A-B (and B-C) is F/4, e.g. a fourth of the sag
ratio of the traditional system (I). In this connection it
is interesting to note that if system (B) in Figure 21 is
made with a sag ratio of 1/40 then this system will also be
sufficiently stiff, and at the same time be characterized by
simpler and cheaper pylons than those of system (J).

Double-cables each with a cable curve coinciding with the

funicular curve of an asymetrical loading (as shown under
(K)) have been proposed- and in a few cases even built
primarily to limit the deflections under traffic load acting
in only one half of the span.

In the case of a multi-span bridge, the system does not

immediately indicate any special advantages in relation to
the critical loading case with traffic load confined to one
span. But with triangular pylons, as those of Figure 21(E),
the system could show some deformational benefits. This
could especially be the case if the two main cables were
clamped to the stiffening girder at midspan, as indicated
on Figure 24. Here the central cable clamp is at the top
of the stiffening girder and the adjacent portions of the
girder supported directly through brackets onto the lower
cable. To assure that the lowest point of the cable is above
the bottom of the stiffening girder (which is essential from
an aesthetical point of view) the girder depth has to be
larger than approx. 0.03 times the cable sag, i.e. around 3
meters for a 100 meter sag.

Figure 24. Double-cable system with triangular pylon

structures.
The double-cable system of Figure 24 will be
characterized by having favourable dynamic properties. Thus,
the configuration of the main cables increases the frequency
88 N. J. Gimsing

of the first asymmetrical mode and the fixation of the pylon

tops has a similar influence on the first symmetrical mode.
Also, the system renders an efficient longitudinal restraint
to the stiffening girder so that it is unnecessary to apply
any special centering devices.

In relation to the-double cable system of Figure 24, it

shall be emphasized that the required cable geometry can be
achieved automatically during erection by suspending each
half of the stiffening girder from its own cable.

The evaluation of the double-cable suspension systems

shall be concluded by the system shown in Figure 23(L). In
this system, shown in more detail in Figure 25, the two
symmetrical main cables are arranged with different sags.
This makes the system able to carry loads of different
intensities in the individual spans without changing the
total horizontal force. Thus, in the dead load condition the
top cable will be subjected to its maximum tension, whereas
the bottom cable will have its largest force in the
dead+traffic load condition. In other words the single span
can adjust to different loadings by changing the
distribution of forces between the two cables.

~II II~
Figure 25. Double-cable system with two
interconnected cables with different sags.
For this reason, system (M) does not require pylons with
a large rigidity in the longitudinal direction of the
bridge, so that conventional column-type pylons can be
applied. On the other hand, the pylons have to be higher to
assure a reasonable sag of the top cable and a reasonable
difference in sag between the two cables - both features
being of importance for the cable steel quantity. For a 1000
meter span, a sag of 50-70 meters for the top cable and 150-
200 meters for the bottom cable seem realistic.

In a comparative investigation between a conventional

three-span bridge with a sag of 150 meters and spans of 750,
1500, 750 meters, and a double-cable suspension bridge in
accordance with system (L), with an infinite number of spans
each 1500 meters long, and cable sags of 225 meters and 75
meters, respectively, it has been shown that the maximum
midspan deflection for the two systems is practically the
same (less than 2.5% difference). This result is based on
the assumption that the pylons are completely flexible in
the longitudinal direction of the bridge. In the actual
comparative investigation, it was furthermore found that the
Cable suspended bridges 89

quantity of cable steel per unit length would be approx. 20%

higher for system (L) than for the three-span bridge.

A comparison between system (E) of Figure 21 and system

(L) indicates that the latter will lead to a saving of
approx. 15% in the quantities of the pylons, but an increase
of approx. 20% in the quantity of cable steel. With these
values, the superstructure of system (L) will ,undoubtedly,
be more costly than that of system (E). Therefore, the
possible savings in the substructure have to be taken into
account to arrive at a conclusion on which system is to be
preferred.

REFERENCES

Ref. 1 : Gimsing, N.J.: "Multispan Stayed Girder Bridges",

Journal of the Structural Division ASCE, October 1976

Ref. 2 : Gimsing, N.J. and J. Gimsing: "Analysis of

Erection Procedures for Bridges with Combined Cable Systems
- Cable Net Bridge Concept", Techn. Univ. of Denmark,
Dept. of Structural Engineering, Report No. R 128, 1980.

Ref. 3 : Gimsing, N.J.: "Cable Supported Bridges- Concept

and Design", Wiley, Chichester 1983.

Ref. 4 : Richardson, J .R.: "Aerodynamic Stability of Twin

Suspension Bridge Concept", IABSE 12th Congress, Final
Report 1984.
 SOME BASIC PROBLEMS IN THE DESIGN
OF LONG SPAN CABLE STAYED BRIDGES

F. de Miranda
Politecnico di Milano, Milano

summary:

Design criteria for increasing the stiffness of long span

cable stayed bridges are recalled and discussed.
Some new technical innovative solutions are proposed and
analized.

1. Introduction

If one pauses for a moment to consider certain modern typo-

logies of bridges over the past 40 years, such as cable
stayed bridges (Fig.l), the attentive critic will be struck
by an evolution that is so rich and variegated that it might
almost be considered a revolution. This revolutionary pro-
cess, with the help of new construction and assembly techni-
ques, and new technologies for the use of materials, has
brought results that are objectively of very great interest
in the field of long span bridges, even from a formal point
of view. And this is just the case of cable stayed bridges.
92 F. de Miranda

Fig.l Messina Bridge Design of "Gruppo Larnbertini" (1969)

L

r~~~ B, I I

a) Three span cable stayed bridge - fan system

b) Single large span cable stayed bridges - fan

system.

c) Three span cable stayed bridge - harp system

d) Cable stayed bridge-harp system with cables

anchored to fixed points in the side spans.

Fig. 2 : Structural systems of cable stayed bridges

Long span cable stayed bridges 93

The construction of bridges has gone on to face structural

problems of growing importance so that the bridge designer
has been able to express and realize his ideas in terms of
statics. So in bridge design the analytical type of problem
has assumed a dominant role, with the result that for some
time now most of the technical and scientific publications
in this field have concentrated on the development of new
calculation methods. However, during the last 30 years
enormous progress in automatic processing has largely chan-
ged the situation, so much so that more thorough structural
analyses for bridge design can now be developed through pro-
grams dealing with the static and dynamic behaviour of three
dimensional structures.
So the overcoming of the analytical type of problem with the
introduction of the computer, should also lead to a revalua-
tion of the importance of synthesis in the process of desi-
gning long span bridges. In fact, much of the quality of a
bridge depends on the first design phase, which is really
the creative moment.
The conception of a load-bearing system is a creative act
which is based only in part on scientific data. It also de-
pends on a sensibility to statics, which remains, like the
sensibility to aesthetics, a capacity to understand and as-
similate physical laws needed for any good design.
The constructional aspects, together with the more strictly
theoretical and analytical side, is of fundamental importan-
ce for the design of long span bridges.

2. Cable stayed bridges

The cable stayed system (Fig. 2) contains straight cables

connecting the stiffening girder to the pylons. In the fan
system all stay cables radiate from the pylon top, whereas
parallel stay cables are used in the harp system.
Besides the two basic cable stayed systems (the fan and the
harp system), intermediate systems can also be found; thus,
in the modified fan system the cable anchor points as the
pylon top are spread sufficientily to separate each cable
anchorage.
94 F. de Miranda

For cable stayed bridges the trend has been to move from sy-
stems with relatively few heavy stay cables to multi-cable
systems with a large number of stay cables supporting the
stiffening girder more continuously.
For the concept and design of cable stayed bridges too,
aspects related to erection have a very strong influence, as
is the case for any structure of considerable size. Thus,
the structural systems and materials, as well as the design
of details, must be chosen with due regard to the erection
procedure.
A straightforward solution is to erect the entire stiffe-
ning girder on temporary supports before adding the cables,
as illustrated in Fig.3 for a fan cable stayed bridge with a
Earth Anchored Cable system, where four main stages are in-
dicated.
This erection procedure offers the advantage that the girder
can be erected continuously by cantilevering from one end to
the other, allowing the trasportation of men, equipment, and
material on the completed part of the deck (Fig.4).
Also, the procedure leads to an efficient control of the
geometry and cable tension. The disadvantage is related to
the temporary supports that must be used. In many cases
clearance requirements during the construction period, or
deep water under the main span, will exclude the installa-
tion of the necessary number of temporary supports, and the
procedure will not be feasible. Temporary supports can be
completely avoided if the bridge is being erected by the
free cantilever method, as illustrated in Fig.5, where four
main stages are involved.
With this procedure it is essential to have a very efficient
fixity of the superstructure to the main piers throughout
the construction period, as the entire stability depends on
this fixity until the end pier is reached. Also, the lateral
bending stiffness of the girder must be sufficient to ensure
the stability of the cantilever arm with a length correspon-
ding to half of the main span.
Thus, the procedure is especially advantageous in bridges
with a large width-to span ratio of the girder (Fig.6).
The cantilevering of a cable stayed bridge requires that all
girder joints are closed as soon as the girder units are in
place, to allow the transmission of the axial forces and the
bending moments induced in the deck during the subsequent
tensioning of the stay cables.
Long span cable stayed bridges 95

Fq~. 3 :

]
S0lf-anchored cable stay~d
bridge erected on temporary
!"'".upport s.

Stage 1: ConHtruction of .n. I .n.

JL----c+ ]-.,
the main piers and temporary
su pport s.

Stage ?: Erection of the I I I Jl.

sti ffening girder.

Stage 3: Insta llation of

stay caoles, mod e rately
tensioned.
.42ra1 ~r
Stage 4 : Removing of the
temporary r;upports and sc 1 f
tensioning of the cable.

Fig.4a: Indiano Bridge (Florence) Fig.4b: Indiano Bridge in the stage

during deck erection by of tensionig of cables.
longitudinal launching~
96 F. de Miranda

1 1
Construction of piers and pylons- The girder
St~ge _ l:
units above the main piers are erected (temporarily)
.ll

pixed to the piers.

Stage 2: Erection of deck and cables by balanced free

cantilevering.

Stage_ 3: Erection of central part of the bridge

Stag_e_ 4_: Closing of main span central additional dead

loads from wearing surface, etc • • • are applied.
Fig. 5 : Self'-anchored cable stayed bridge erected by
double-sided free cantilevering from the pylons.

Fig.6: Zarate Brazo Largo Bridges dur.i~ erection.

Long span cable stayed bridges 97

3. Structural Analysis of Cable Stayed Bridges

Reference is to a continuous three span stayed bridge with

the fan system and with a continuous distribution of cables
along the girder, which is the most suitable system for long
spans. It is assumed that construction, by regulating the
tension in the cable, will ensure a practically straight fi-
nal configuration with no bending moments in the deck. The
structural analysis due to permanent loads can be carried
~ut starting from the equilibrium equations of a beam ele-
ment in the left half of the bridge (Fig.7):

~~ + ii cos a = 0 (1)

g- ii sen a= 0

in which z is the abscissa of the cross-section considered,

N is the axial force in the deck, N~ the axial force in the
cable. With reference to the [Link] abscissa f = z 1 H 1
and considering that:

tg a =-}- ;
) cos a
Rt
=

then the second equation of system (1) gives:

ii = [Link];F (2)

which, when substituted in the first of the equations

(1), gives:

dN =-gHf .
d!'
98 F. de Miranda

From this, taking into account th~t at the centre ( r=L/2H),

because of the erection procedure,· it can be assumed that
N = o, on finally we obtain:

( 3)

The action of the live loads modifies the initial equili-

brium corresponding to the permanent loads, and sets up an
additional stress-strain state. The stress state already
existing before the action of the liv,e loads is made up of
tensile stresses in the stayes and of compression stresses
in the pylons and in the [Link], the action of
the dead loads fixes the equilibrium configuration of the
cables, and thus controls their reactions through the value
of Dischinger's virtual modulus of elasticity (Fig.8):

( 4)
E•=--E__
')'2~[
where: 1 + 120'3
~=specific weight of steel in the cable
lo= horizontal projection of the lenght of the cable
E= modulus of longitudinal elasticity of the cable
a = tension of the cable.
The additional strain in the bridge is further indentified
(Fig.9) by the vertical displacements v<Zl of the girder, by
the rigid horizontal displacement w of the girder itself,
and by the elastic horizontal displacement u of the tops of
the pylons.
Structural analysis for evaluating static effects due to li-
ve loads, after the first simple applications in the 1950's,
has been studied theoretically by various authors with ap-
proximation methods that are very interesting, but tend to
be rather laborious and complex.
At the same time, however, the complete and exact structural
analysis of the stress-strain state has been carried out sy-
stematically by means of a discrete model which also takes
into account the effects of non linearity due to strains in
the cables and variations in the geometry of the structure.
The discrete model can be obtained either by the analysis
with a computer of a system with a given number of cables,
or by solving with a numerical method (e.g. the finite ele-
ment method) the equations of the continuous model.
Long span cable stayed bridges 99

Fig.7: Typical structural system of wide span cable stayed bridges.

rsoo
1"400 t.
1"200
moo
800
600 E
400 Y2 t!E
1+---
200 12 6"J
o.___,___.._..~.--L........._..~.-__ ,
0 ~o Cm}

Fig.B: Virtual modulus of elasticity of the cable.

Fig.9: Distorted system under live loads on half central span

100 F. de Miranda

The results obtained in this way ~re very precise and make
it possible to state that the earlier analytical obstacles,
considerable though they were, have now been overcome, ob-
stacles that were responsible for the initial distrust in
the cable stayed bridge and its late development.
In fact, computers have made it possible also to conduct in-
vestigations and reasearch on stiffening systems of cable
stayed bridges for application to long spans, or to very
heavy live loaded (railway double track bridges), as will be
illustrated later.

4. The Stiffening of Cable Stayed Bridges

A three span cable stayed bridge with a fan system of cables

of the type shown in Fig.7 is already in itself one of the
stiffest structural systems for bridges.
But,first of all, we have to outline how to the optimum va-
lues of the H/L ratio (concerning the weight of the steel
required for the cables) also an average optimum case of
rigidity is corresponding; this means that, in relation to
such values, are minimized the differences between rigidity
(in the same plane of the structure) of the zones near the
towers and of the zones near the centerline, as well as the
absolute rigidity values.
The elastic deflections in the central zones of the span due
to live loads on the central span are also influenced, and
quite considerably, by the value of the horizontal longitu-
dinal displacements of the tops of the towers, and by the
angle formed by the generic cable with the axis of the
deck.
Referring to Fig .10, for the generic stay ( i) loaded like
shown in Fig .11, the vertical displacement Ov 111 is:

q~ H
seii'(i; . Se'ii'Q(;
6u [;A;
6v(il = sen at= --~s""'"e-'-n.:!.a-~-- =
Long span cable stayed bridges 101

Furtherly, because it is:

A.- N
·--=Uai

the value of becomes:

0 _ q.:iH 1 Ua; Ua;H 1

v(i)- [ ; - sen3a; q,:i =L -s-e'-n2_a_;
sen a;
The function Ov = f ( d; ) is represented in Fig .12, where
clearly appears that for~= go• is:

while, with the same value of Oa , for ~ 30• the value of

becomes 40nm
Then, for low values of H/L , wihle the vertical displace-
ments o.
of cables near to the towers are smaller, contem-
porarely the values of o. for the cables near the center of
the bridge will increase very much, because increases itself
the geometric coefficient:

7"
a3.
serr2amin = ( 1+ 3 + ••••
)-2
Practically, in order to not differenziate too much the lo-
cal flexibility of the bridge around the center of the span,
it is necessary to limit the value of dmin
For the type of structure considered, if we adopt for the
ratio H/L, values between 0,21 and 0,24 , we can consider to
give a high stiffness to the structure itself.
102 F. de Miranda

I
I
I
H: I
I
+-·6-.I ______;;....__..:..L...-~~
r.v .llL
senoc
-t-- ----
[Link]

fv

H
Ci;-y

so• oc•

Fig.12
Long span cable stayed bridges 103

However, for long spans this stiffness reduces considerably

in the central area of the span for a ·length of about (0.3
+ 0.4) L across the mid-span. This is due, above all, to the
reduction in the axial stiffness of the longest cables due
to their dead load:

EA
E. A I (5)
-~-=

where:
A= cross section of cable
1 = length of cable
10 = horizontal projection of 1,
a = stress in the cable

In order to overcome this handicap, and so to introduce the

cable stayed bridge with all its technical-economic advanta-
ges into the field of larger spans, certain innovations are
needed in their design construction and erection. Let us
consider some of these possible innovations.
A first solution to counter the Dischinger effect may be ob-
tained by fitting a series of counter-stays AB', AB", AB'",
AB"", etc ••• , stiffly anchored in A (Fig.l3); in this case,
and in the hypothesis of inextensibili ty of the counter-
stays, the function E*A/1 becomes the straight line (shown
by the dotted line in Fig.l4) of equation:

E.A EA
-~-=-~-

An intermediate behaviour takes place in the real hypothesis

of more or less extensible counter-stays (Fig.l4 curves d ,
, ~, ~ , etc ••• ) • Such counter-stays permit to contain the
Dischinger effect and to increase the cross section of the
stayed-cable without changing the force Ng•
Only in this case low-resistance cables ([Link]
concrete cables) could be optimally used even for spans
greater than 400 m.
A second solution, proposed by the author in 1969, provides
for counter-stays to be fitted as per Fig.l5, i.e. so as to
connect the counter-stays themselves in correspondence of
104 F. de Miranda

the deck points, and perm! ts to obtain a considerable de-

crease of the Dischinqer effect, considerinq. also the fact
that the axial riqidi ty of the counter-stays is very hiqh
because the values 10 of the horizontal projections of their
lenqth are qenerally very low.
Generally speakinq, a stiffeninq achieved accordinq to the
arranqement indicated in the first solution involves the use
of a quantity of steel for the· counter-stays qreater than
for the second solution.
Let us now investiqate and analize the efficiency of the
counter-stays accordinq to the scheme of Fiq.15.
I t a sinqle stay rod is placed accordinq to Fiq.16 and ten-
sioned with a force F, the increase f of the deflection of
the cable mid-point is linked to force F by the relation
(Fiq.17).

L4Nf = 2NsentJR = F

In fact, at the 1st order, there is no increase of the axial

force N in the cable because of the applied force F.

In fact we have:

senfJ =_f_ ,
.1.
2
and therefore:
f = FL
4N

If we now qo back to the scheme as per Fiq.16, the point A -

where the stayed-cable is connected to the truss- may be su-
bjected to a vertical saq, as point B; on the contrary,
point C, the pier head, may, as a first approximation, be
considered fixed.
Long span cable stayed bridges 105

The vertical desplacement vA of point A then increases the

axial force in the cable

AN= E.A AL
L .L.l '

where ~L is the component of vA ·along the cable axis; but in

this case, because of the stay rod, the reduced modulus E*
is to be calculated with reference to a cable with length =
L/2: in fact the mid-point of the stayed-cable is linked to
the stay rod and therefore the free span of the cable is
halved. This is true if the increase of the deflection Af
due to the force increase ~N (Fig.18) is of the same order
of magnitude as ~L, i.e. if the stay rod has sufficient ri-
gidity.
As regards the validity of the stay rod as the restrain of
the cable mid-point, we finally have to check that the de-
flection increas df due to VA and v6 is small, that is, in
the same order of magnitude as AL. In fact, because of AL,
we have an increase A f given by:

~f =

and since:

f =FL ,
4N
we obtain:

Af F E. A ( __E_ E.A .
~L = N KtL+N N KtL

Therefore, if At and Lt are respectively the cross-sectional

area and the length of the stay rod (counter-stay) , we ob-
tain:

~f F E.A _ F Lt A _ F 1 A
AL(N EA,L-NL""A."-N2At
Lt
If we assume F/N :::: 0.01, then it is also: Ar_/A _ 0.01, and
therefore A.f/ AL is in the order of the units.
106 F. de Miranda

As

Fii13 Fig.14

Fig.15

/'
/ \
' \
\

Fig.16

~~~
r
N l!L
--P~---
~
2

N F
l·tg. 4.16.
~F

Fig.17 Fig.18
Long span cable stayed bridges 107

We may conclude that the stay rods may stiffen the cables so
long as the initial force F is at least in the order of
[Link]. As a first approximation, therefore, the effect of
the stay rods is to interrupt the span of the cables as re-
gards the calculation of the Dischinger module.
However, in order to limit the possibility of deformation of
the stayed-cable bridge, it is paramount that every cable
has to be sized so as to offer maximum axial rigidity E*A,
where E* is the virtual elastic modulus (Dischinger) estima-
ted in the presence of permanent loads only, and A is the
cross-sectional area of the cable.
In sizing the stayed-cable, two factors should be conside-
red:
a) The axial force in it, in the configuration subjected
to dead loads only, is determined, and its value is
(Fig.19)

b) If c5s = N /A is an "assigned stress" in the stayed-

cable underg permanent loads, the cross-sectional area
value is:
A: N11 •
Ug

The value of c59 is to be established taking into account the

maximum increase C1p due to live loads and the value of the
allowable fatigue stress, and also the need to guarantee
sufficient axial rigidity to the cable.
The value of stress dg that ensures the maximum value for
axial rigidity E* A is then estimated; the result is:

(6)

The function:
108 F. de Miranda

runs as shown in Fig.20, where the rigidity of the cable

appears, given the values Nq, '}' , 1 0 ;it shows a maximum in cor-
respondence of a suitable value of ~ determined by the con-
dition:
d~ = 0 '
dCJ9

that, for the optimal value ( 6g )opt , provides the expres-

sion:
( 7)
(uu)opt = ~ 1'2 ~ E
Therefore (6), through (7), gives the maximum value that may
be obtained for the axial rigidity E* A, the value of which
is:

-3 Nu E ~6
(E.A)opt -_2 2 2 • ( 8)
I' ~ E

The values of ( a9 )opt and (E*A)~pt are represe9ted ~n Fig.21

for steel cables ( ~ = 7,85 tjm ; E = 2.1 x 10 t/m )
If in ( 6) we let :

Ko:

the same may be written as follows:

This function is represented in Fig.22 for certain values of

K0 as Ng varies.
If we multiply everything by A/1 (with 1 =constant), we ha-
ve:

_E._A_ =--,---=E:___ _ (9)

I I(U\ + KoA
2)

The curve E*A/1 (Fig.23) for a certain value of K0 increases

Long span cable stayed bridges 109

Fig.19

Fig.20

![Link]. 105(~l
0 9opt. ( tj m2 l
NE I
35000 14
30000 ![Link]. 12

25000 10

20000 8

15000 6

10000 4

5000 2

100 200 300 400 l!ml

0 L(m)
0 200 400 600 800

Fig.21
110 F. de Miranda

E"
2"000000 L, =100m
II ,,, =7.85 t/mc
Ka=466 m'-Nv= 1300t
1"600000 K~'=128 .n'-Nv=2000f
K'a=
" 58 ni'-Nv=2600t
1"200000

80(!'000

400"000

0 5 10 15 2S 30 35 40 A
(tmj
1------t 500 cmq

Fig.22

_E~_A (Kg/em) [Ol•1="-=[y

llo.100 m

220~=00~1--------~~~~~~~~
I ::.120
m
E=2·10• Kgjcmq

167000

111"1

5 10 15 20 25 30 35 40 A(cmq)
1-----i 500 cmq

Fig.23
Long span cable stayed bridges 111

in an almost linear manner for very small values of A (very

high tensions 09 = Nq/A, at first exceeding any limit), rea-
ching then a maximum for A = A obtained from the equation:

d [ -~-
E.A} = O
dA '
decreasing to zero for A = oo ( i. e. CT =0)
This result indicates that, generally speaking, it is possi-
ble to obtain an axial stiffening of the stayed-cables by
increasing the area of the resisting cross-section up to the
limit of A.
In order to increase the flexural rigidity of the structure
through an increase of the axial rigidity of the stayed-
cables, we may also use a cable with a larger cross-section,
obtained, starting from the original stayed-cable, by injec-
tion of mortar in the sheath, and subsequent prestressing by
means of an additional cable with cross-section A11
(Fig. 24).
If load Nq is entrusted completely to the section A1 of the
main cablGs (cables I), we have:

uJ 11 = Ng
AI
<= Uam) •

After the inj action and cure of the mortar, the secondary
cables (with section Arr> are pre-tensioned to a value
0.9 CTamm' and at the same time the main cables(!) are
relieved from stress, reaching a value:

(I)A
Ug :::::: 0,9 Uam ,

while concrete is prestressed with a stress:

ulol=- [Link] (::::::140+180kg cm- 2 ) .

Ac+nA.
The subsequent intervention of live overloads brings the
stress in the steel cables (main and secondary) from
o. 9 (1' mm to CTg +Up :::oamm, making the phenomenon of fatigue in
the steel cables negligible, since 4 o-::: o, 1 o- ~mm, while con-
crete is relieved from stress to a value: lminjCT[Link]-0,10~ol.
112 F. de Miranda

Fig.24a

Fig.24b

\ ~ Ill.'""!' ""

·~- '""'"<L~

! UTA

Fig.24c
Long span cable stayed bridges 113

The axial rigidity of the stayed-cable thus treated is the-

refore modified in the expression E* A, for the following
reasons:
the area A of the cable has been increased, which un-
doubtedly is a positive fact;
the elastic modulus E* is instead reduced because the
mortar adds weight to the cable.
In order to check the effect of the reduction of E* modulus
we should remember the original expression of this quantity.
From an analysis of the behaviour of a stayed-cable with
cross-section A and with weight per unit of length g = I A
subject to an axial force Ng , we obtain the following ex-
pression for modulus E*:

If we assume the stayed-cable to be rigid with respect to

the flexural deformability of the truss, the overall stress
in the stayed-cable remains fixed after prestressing.
If we homogenize the areas of concrete in steel, the overall
value of area At is:

The elasticity modulus of the prestressed stayed-cable , Ep*

becomes:

E: = E
1 + 9f I~ EAt
12N:
where gt is the weight per unit of length of the prestressed
stayed-cable:

9t = 'Y ( A1 + Au) +'Yc Ac

114 F. de Miranda

The modulus E~may then be written as:

E: = ___E_ _ __ (10)
1+
'Y,2 I! E
12 0"93
where:

The ratio between the axial rigidity of the prestressed

stayed-cable and the axial rigidity of the steel stayed-
cable, may be considered as an indication of the effective-
ness of prestressing: its value is:

'Y2~ E
1+
K: E: A, =~ 12o:
E. A A )12~[
1+
120i

This ratio may theoretically be greater than, equal to, or

lower than 1.
Relation (10) supplies for rigidity Ep* A of the prestressed
stayed-cable an expression similar to (6), i.e. for the pre-
stressed stayed-cable, too, as U9 varies, the axial rigiditx
runs according to Fig. 20, and therefore a maximum value (E•At)opt
that, taking into account (8), is:

(11)

Relation (11) shows that the maximum axial rigidity value of

a prestressed stayed-cable only depends on the average spe-
cific weight }'t , 1 0 and Ng being egual.
Long span cable stayed bridges 115

The value of the ratio between the maximum optimum axial ri-
gidity of the steel stayed-cable and the maximum optimum
axial rigidity of a prestressed stayed-cable, is therefore:

And since

_.!!_ = 'Y(A• +AukYAc _ 'Y{A.+Au)+'YcAc

'Y 'Y (A• +Au +Ayn)- }'( A1 +Au) X Ac
n

as 'Yc is always > y ;n, therefore: 'Yt 1 y >1.

In conclusion, the maximum optimum axial rigidity of the ca-
bles is obtained with steel-only cables.

But the simplest and most effective solution for the axial
stiffening of the stayed-cables is based on a criterion
that, with particular construction methods, provides for the
static interaction of high-tensile steel stranded cable and
the sheath when this is formed by a tubular steel sleeve:
1) The cables formed only by the strands support, by them-
selves, all the dead load of the deck, reaching a
stress value CT9 , value that, added to tension C1p in-
duced by live loads, does not exceed the allowable sta-
tic value for the type of steel used, i.e. the value
apart from the fatigue strength of the strands.
2) The steel sheath, only after the application of all
dead loads to the deck, is solidly fixed to the cable,
rigidly connecting the sheath to the stay heads by
means of welding.
With this process the cross-section A* of the strands is in-
creased by the value A~ corresponding to the cross-section
of the steel sheath only for absorbing the stresses arising
in the stayed-cables because of the live overloads.
The injection of mortar in the steel sheathing shall be done
after the cable heads of the high-tensile steel strands have
been solidly fixed to the sheathing formed by the low-
116 F. de Miranda

tensile steel tube. This operation is carried out, if pos-

sible, with the aid of a movable load placed on the deck in
correspondence of the section in which a pair of stayed-
cables are connected to the deck. This load shall be pre-
sent during the injection operations and until the (quick)
cure of the mortar.
In this case the tensional layers in the stayed-cables are
identified by the following values.
During a first stage a stress arises in the strands:

,..-(11_,... _ Nu
ve - vg - -
A.
due to the dead loads.
During a later stage, because of the live loads, the tension
in the strands reaches the value:

max a. =O"u + _N_P'--

A.+A<P
During this stage the stress reaches the following value in
the steel sheathing:

The maximum stress in the strands and the maximum stress in

the sheathing shall not exceed a*~~~ and CT"Jr;),.,, respectively, of
the fatigue strength of the strands and of the steel sheat-
hing.
The effects on the static behaviour of the bridge are cleary
positive, as it is easily possible, by fully using the
strength characteristics of the materials used, to almost
halve the maximum flexural sag that would occur when forming
the stayed-cables by using only high-tensile steel strands.
We shall presently show an application of this interesting
solution to a double-track railway bridge designed by my
office for crossing a 350 m wide navigable channel (Fig.25).
The cross sectional area of cables, formed by strands and
steel sheating in combination, in line with the full utili-
zation of the strenght characteristics of the materials
used, permitted to reduce of about 60% the deflections of
the bridge related to those corresponding to use cables for-
med only by strands of high tensile steel.
Long span cable stayed bridges 117
118 F. de Miranda

In order to project the cable stayed bridges, with all its

techni'cal and economical advantages into the field of the
maximum spans attainable with the suspension bridges, a high
number of innovations is needed in their design and erec-
tion.
Let us consider some of these possible innovations.
To start with, notice that for cables in general and parti-
cularly for longer cables, if 1 0 could be halved for the
same u ; then, with the live load acting on the c*entral
span, 'there would be a small percentage redution of E com-
pared to E, and so reduced deflections of the deck.
If, as a new step, the longitudinal displacement u of the
top of the towers could also be reduced, then the deflec-
tions of the deck would be furtherly reduced too. In addi-
tion, if the values of H could also be reduced without at
the same time reducing the lower values of ~ , then at the
points where the cables join the deck, there would be no in-
crease in'the deflection of the deck itself.
These goals can be reached by adopting the bridge system
shown in fig.26.
Note the following details:
1) The introduction of a special box cross-section
counter-stay hom-shaped in steel plate (braced tran-
sversally), which makes it possible to halve the values
of 1 for longer cables, and also to reduce the heigh H
of t~e towers by (10-15)% without reducing the values
of [Link].
2) The use of reinforced concrete towers of height
H'=(0.85-0.90)H,A-shaped in the longitudinal plane.
In order to check the efficacy of the structural system of
fig.26 compared to the cable-stayed bridge system of fig.27,
some comparisons were made with reference to a cable-stayed
bridge of three main spans (600+1800+600)m designed by the
present writer and others in the Lambertini Group for a road
and rail crossing over the Strait of Messina (see fig.l).
For the same spans and loads, this ~ew system tending to re-
duce the influence of parameters E , u and OG; for longer
cables gave the results shown in fig.26 and 27 for the fle-
xibility of the bridge. Note the reduction in the maximum
values of the elastic deflections in the central area of the
deck. They are about 34% lower, and with an elastic curve
distinguished by smaller values of the maximum inclinations
of the deck itself; values of N in the deck will be reduced
too.
Long span cable stayed bridges 119

I 600 I 600 600 1

! !I!I!I!IIIII!I!III!III!IIIII!IIIII J J J J J JJJJJJJJJJI JJj JIIIIJJ,~~;I:~~~II!:I~~~III!I iCi
1

li1 = 2 , 8 5 q2 2 =roadway load

1
~~~~
J:1so I
.150. 600 1800 600

Fig.26: Cable stayed bridge system with horn-shaped counter-stay

I 600 I 600 600 t

q,llllllll!llllllll!lllllllllllliiiii!JIIJJJJJIIIJJJJJJJJJJIJJJIJIJJIIouuu!moo•uouuuuuu! i q2

~~~~1

Fig.27: Typical cable stayed bridge system for long spans

120 F. de Miranda

The horn shaped counter-stay makes it possible to erect the

structures of the central span in two stages. In the first
the horn and the two upper half-stays are erected at the sa-
me time as the cables of the side span. In the second stage
the erection of the central span (deck and lower half-stays)
is completed.

The A-shaped tower, still in the longitudinal plane, would

also reduce the L/1 ratio, which for the classical cable-
stayed bridge was fixed at 3.00 to ensure that the anchorage
cables had an adequate safety coefficient for stability
against unloading when the mobile overload is present omly
on the side spans.
In this way the stiff A-shaped reinforced concrete tower
would make it confortably possible to increase each side
span to 700m, thus permitting the two anchorage piers to
reach the land on either side of the strait, at a distance
of 3200m in the point where the sea-bed is less deep.

The planned continuity of the deck beyond anchorage piers

with two small self-supporting 150m spans, and the addition
of two (tensile) counter-stays connecting at the two inter-
mediate points the anchorage cables to the lower parts of
the towers,offer a further contribution to the reduction of
the flexural deformations of the deck for a system with such
ratios between the spans.

BIBLIOGRAPHY
-De Miranda F. ,Sylos Labini F.:Condizioni di ottimalizza-
zione del peso strutturale del ponte
strallato con schema a ventaglio
-costruzioni Metalliche, n•l/1972.
De Miranda F., Como M., Grimaldi A., Maceri F.: Basic
Problems in Long Span Cable Stayed Brid-
ges - [Link] Structures, University of
Calabria, Report n•25, 1979.

De Miranda F. Ponti strallati di grande luce

[Link] - Roma 1980.

Gimsing N.J. Cable supported Bridges - Concept and

Design John Wiley & Sons - 1982.
De Miranda F. Design - Long Span Bridges -
International Symposium on Steel Bridges
London 25-26 Feb.l988 - Institution of
Civil Engineers.
 ON THE DURABILITY OF REINFORCED
AND PRESTRESSED CONTRETE STRUCTURES

M. Mele
University "La Sapienza", Rome, Italy

E. Siviero
Istituto Universitario di Architettura, Venice, Italy

ccln infrastructures, which more and more need to be

designed as national European systems, a uniform
functionality in the territory must be sought overriding
the lack of coordination in investment and maintenance activity ...

Carlo Azeglio Ciampi

Governor of the Banca d'ltalia
Final comments
May 1990

1. Introduction

Vitruvio, in Treatise on Architecture, indicated as requisites for buildings: firmitas, utilitas

and venustas. Safety ( firmitas) therefore always had to accompany functionality ( utilitas) and
the rules of appearance (venustas). It may be assumed, in a more general acceptance, that
firmitas also contains the concept of continuation in time. In that sense, however, a new
requisite, vetustas, must be explicitly added. In fact, when referring to natural stones, he
recommended extraction of samples from the quarry and leaving them exposed to the
weather for two years, checking their condition before, and the lstituto Universitario
di Architettura of Venice has adopted these so much so as to insert them in its own coat of
arms (fig. 1).

Fig. 1. Vitruvio: Firmitas, utilitas, venustas

lstituto Universitario di Architettura of Venice.
122 M. Mele, E. Siviero

The idea of durability, considered as an awareness of deterioration for disrepair phenomena,

became very apparent from the second half of the XVIIIth century, in parallel with the advent
of building technology based on the use of new materials (steel, concrete) as alternatives to
the traditional materials (timber, brickwork) where the time-behavioural characteristics were
better known. In fact at that time the first phenomena of embrittlement of the first iron bridges,
phenomena of cracking of concrete and corrosion of steels, and later - after the first
application of prestressed reinforced concrete, - phenomena of corrosion under the
prestressing steel forces and unsuccessful injection in post-tension cable sheathing, became
apparent.
Subsequently, as from the '50s, new materials were used without full knowledge, based
on an experimentation support, owing to compelling needs for the short term reconstruction
of what had been destroyed by the events of war.
In the last ten years, progress in the methods of structural analysis, together with an
increased architectural creativity and an at times excessive use of prefabrication, have led
to the frequent use of reduced thickness with ever smaller dimensions without adequate care
in studying the construction details.
Added to this a very limited maintenance so that structures designed with a correct
structural plan have deteriorated over the period of 20-25 years with serious functional
difficulties and heavy repair costs. In addition, difficulties of access to the faulty parts have
frequently caused further delays to inspections and performance of the appropriate action.
As from the '80s the improvement in the quality [Link] has led to a reduction in the
quantities used while still achieving the required strengths. However this also increased the
porosity ofthe concrete which, as is known, is one ofthe major causes of corrosion by external
agents.
Even from the aspect of increase in mechanical properties, the at times frantic search to
achieve the greatest strength in a short time, in order to satisfy technical standards, has
resulted in a flattening of the time curve with an apparently reduced reserve of strength as
compared whit those previous years (fig. 2).
I
I

'::.\._ ---------]
I

:
I , , -- J

I
,k:::'____ ----------------- !
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I

21 . . •••
Fig. 2. Stress variation with time for concretes in the past and today.
Durability 123

It has therefore become necessary to develop a constructional awareness in the

professional community and a sense of responsibility of the various operators in the light of
sufficient time response of the structure to its original design and its intended use, particularly
in view of the relative financial implications. Accordingly, the concept of durability and
maintenance have now become a decisive part in the cultural-technical background of Public
Administrations. In particular, a thorough analysis of these problems is contained in a ANAS
document, published in 1988, on the «durability of road structures», which examines various
concurrent causes including in particular:
- the design and building of ever more slender structures using a systematic and frantic
search for structural lightness achieved by progress in structural analysis and the
development of constructional procedures;
- less careful study of the constructional details and working details as a consequence, inter
alia, of the post-eighties separation between the roles of the designer and the worksite
manager;
- the ever wider use of prefabrication techniques, not matched by a use of reliable materials
and frastenings, and careful checks particularly at the time of testing.
All this has been partially faced by the introduction of semi-probabilistic rules for approach
to the limit states whereby control of the work is required not only with regard to safety against
collapse but also in respect of the working conditions which represent, although incompletely,
a check on certain durability problems.
In a definition given to the mixed FIP/CEB committee, durability is expressed as the
suitability of a structure to withstand attacks by various corrosive agents, whilst essentially
maintaining the functional features unaltered. This definition, which refers to the behaviour
of the material in environmental and operating conditions, is equivalent to expressing
durability as a longterm property, one which is therefore linked to the loss of performance as
a function of time.
The concept of durability must therefore be extended from the intrinsic material to the
structural items, on to the structures overall. More properly the term of «efficiency .. should be
used for structures in the sense of matching between the functional duration and the life of
the structure itself.
It is therefore possible to define durability as the suitability of a structure to retain its original
functioning use in normal working and maintenance conditions during the entire anticipated
life period. In that way other parameters, closely linked to the definition of durability, are
involved.
This trend towards allocating greater scientific strictness to the concept of durability is
confirmed in the definition given by ASTM in the document «Standard practice for developing
acceleration tests to aid prediction of the service life of building components and materials»
of 1981: durability is the capacity to maintain the service life of a product, a component, a unit
or the entire structure over an allocated time.
Service condition and therefore capability of use have been seen as the capacity of these
units to perform the functions for which they have been designed and constructed. These
service characteristics are affected by externals of atmospheric, ecological and degradation
or deterioration. That definition of durability, as given above, is combined with the more
modern definition of service life, considered as that period, following the construction of a
structure, during which all the remain at the minimal acceptable values in ordinary working
conditions. Therefore durability and service life are on design requirements to be satisfied for
a pre-established period of time.
124 M. Mele, E. Siviero

This relationship between durability and service life was adopted and developed from the
technical aspect and also from the economic and functional aspect during a CEB-RILEM
workshop held in October 1986 at Bologna relating to the durability of concrete structures,
noting how service life can be identified from the technical, functional and economic aspects
defined as follows:
- Technical Service Life (TSL: period of time beyond which an unacceptable service
condition is reached. This aspect, within the approach to the limit conditions, is linked to
the concept of safety as final limit state, to the concept of operation as limit service state
and concept of external quality of the structure from the appearance aspect;
- Functional Service Life (FSL): period of time beyond which the structure becomes
obsolete from the functional aspect. The functional duration is clearly influenced by any
modifications over the course of time as compared with the operating requirements
provided for at the outset;
- Economical Service Life (ESL): period of time beyond which a total substitution of the
structure is economically more advantageous than its continuation in service with
recourse to maintenance or repairs.
Economic life therefore depends on a cost-benefits balance sheet for the use of a
structure with repair and maintenance costs or total rebuilding costs.
The technical service life is however linked to the maintenance of a minimum level of
efficiency which must indeed be considered as relating to safety, usability and external
characteristics.
In the light of the foregoing clarifications it may be asserted that the correct design of a
lasting structure or, as is said, «design for durability» must materialise in a technical duration
of the structure which is greater than or at least equal to the other two, therefore introducing
a relationship of dependency of the life of the structures on functional or economic factors.
It is therefore possible to identify a service life which depends on the one hand on the
properties of the materials and structural characteristic and, on the other hand, on the
environmental conditions on functional and economic requirements and also the maintenance
carried out.
In brief, the effective life of structure will consist of the minimum value between TSL, FSL
and ESL, whilst the objective of the design from the aspect of durability will be achieved when
it is guarantee that the TSL is not less than the FLS or ESL.
The time factor therefore enters in the philosophy of the structural design, it being
necessary to guarantee the retention of the margin of initial safety throughout the entire
service life, which entails the maintenance of its integrity over the course of time. This aspect
induces us to consider maintenance in a wider acceptance than usual which includes the
need for checks, repairs, substitutions as activities for which the owner of the structure is
responsible and which are not strictly connected to the guarantee of durability of the structure.
Furthermore, whilst structural safety, considered apart from thorough knowledge of the
actions and characteristics of materials, constitutive laws and structural behaviour, the same
cannot be said when the valuation includes the time factor. In fact, from this aspect,
knowledge is still lacking and accordingly precise standards are also lacking. Therefore at
present the sole indications are limited to a statement of principles of general characteristics
such as:
-use of the structure;
- performance criteria;
- the environment;
Durability 125

- the materials used;

- the form of the structural elements and the constructional details;
-the quality of work and control level;
- the protective measures;
- maintenance.
It is in fact necessary for durability to be considered as a fundamental prerequisite in the
phases of:
-planning and design;
- realisation;
- use period.
In this context, in order to achieve the pre-established objective, strict cooperation
between owner, designer, builder and user becomes absolutely necessary and each of these
has to perform his own role in conjunction with the others.
The design documents therefore cannot be limited to drawings accompanied by technical
specifications for realisation, but must also include the plans for control and maintenance of
the structure.
According terms of reference to the following aspects, becomes offundamental importance:
- for the design, in order to reduce the probability of the occurrence of human errors;
- for the realisation in order to identify and correct insufficient results in the productive
process;
- for maintenance, as a control that the safety levels remain unchanged over time.
The consequences of failure to control may indeed become dramatic as unfortunately too
many examples show.
In that sense, a useful reference may consist of what is known as the cclaw of the five .. (De
Sitter) which identifies the following four phases in the life of a structure
A. design and realisation;
B. ordinary maintenance;
C. special maintenance with local damages;
D. general repairs with structural restoration;
forming a brief evaluation criterion for the costs of action during the various phases based on
the adoption of a multiplier 5 in change in between one phase and another in the instance of
progression of negligence.
So considering the extra cost in the instance of good design and correct performance
during phase A as being equal to 1, failure to implement this induces an extra cost:
-in phase B equal to 5;
-in phase C equal to 25;
-in phaseD equal to 125.
It follows that it is necessary to act correctly both during phase A with good design, correct
realisation and adequate quality control and during phase B with inspection, control,
maintenance, systematic operations, thus achieving an extension of the service life of the
structure from its there aspects mentioned above: technical, economic and functional.
In that sense, the design choice is therefore effected by reducing the overall cost function,
defined as follows, to a minimum:
 126 M. Mele, E. Siviero

where:

c:
C overall cost;
initial cost;
C; cost of repair action during the life of the structure;
P; the probability that the event which requires the action of repair will be required.
In brief, from a general point of view, the durability design criteria, particularly in
accordance with the CEB FIP Model Code 1990, may be identified as follows:
a) choice of the structural shape suitable for the environment, accessibility, inspection and
maintenance;
b)good quality of the concrete mix and sufficient cover of the bars;
c) correct design of reinforcing details;
d) control of the nominal spacing;
e) provision if necessary of protection of the steel and/or for the concrete surface;
f) specific technical details for the materials to be used with recommendation for use and
indications for maintenance of time of live.
However, to tackle the matter of durability from a theoretical aspect has proved to be a
fairly difficult problem. In that sense, an attempt has been made by a study group of CTE
coordinated by prof. Toniolo in the Italian CEB Group on the subject «durability.. of CEB-FIP
Model Code 1990. These concepts were then adopted by other scholars in three ANAS
seminars by prof. Radogna on 301h October 1989 at Rome, by prof. Migliacci on 131h
November 1989 at Milan and Mr. Macari, engineer, on February 131h 1990 at Padua.
In view of the substantial interest which they assume and the subsequent developments
which have become apparent, these concepts have been adopted in their entirety in
paragraph 6 of this note appearing below.

2. Deterioration and protection factors

The required life for a structure and the inspection and maintenance programme call for
protection levels linked to environmental conditions and also to the chemical and physical
attack mechanisms.
As indicated by Eurocode 2, environmental conditions may be classified in 5 exposure
classes (table 1).
In general, chemical attack can be originated by:
-use of the structure;
- corrosive environment;
-contact with acid solutions or gases or sulphate salts;
-chlorides;
- alkaline-aggregate reactions.
Physical attack results usually from:
-abrasion;
-freezing/unfreezing action;
-penetration of water.
In any event, sufficient corrosion of the metal reinforcing is required. In normal conditions,
this objective is achieved by an adequate cover consisting of good-quality concrete
characterised by low permeability.
Further protective measures must be implemented in particularly corrosive environments
both for the concrete surfaces and the reinforcing itself.
Durability 127

Tab. 1. Exposure classes related to environmental conditions (Eurocode 2).

Exposure classes Examples of environmental conditions ·

1
dry environment interior of buildings for normal habitation or offices(1)

a - interior of buildings where humidity is high

without (e.g. laundries)
frost - exterior components
2 - components in non-aggressive soil and/or water
humid environment

b - exterior components exposed to frost

with frost - components in non-aggressive soil and/or water
and exposed to frost
- interior components when the humidity is high
and exposed to frost

3
humid environment interior and exterior components exposed to frost and de-icing agents
with frost and de-icing
salts

a - components completely or partially submerged in sea water,

without or in the splash zone
4 frost - components in saturated salt air (coastal area)
seawater --
environ-
ment b - components partially submerged in sea water or in the splash zone
and exposed to frost
with
frost - components in saturated salt air and exposed to frost

The following classes may occur alone or in combination with the above classes:

a - slightly aggressive chemical environment (gas, liquid or solid)

- aggressive industrial atmosphere
5
aggressive
--
chemical b - moderately aggressive chemical environment (gas, liquid or solid)
environ-
ment(2)
--
c - highly aggressive chemical environment (gas, liquid or solid)

(1) This exposure class is valid only as long as during construction the structure or some of its components
is not exposed to more severe conditions over a prolonged period of time.
(2) Chemically aggressive environments are classified in ISO/OP 9690.
128 M. Mele, E. Siviero

Tab. 2. Minimum cover requirements for normal weight concrete (1) (Eurocode 2).

Exposure classes, according to Table 1

1 2a 2b 3 4a 4b Sa Sb Sc(3)

Minimum(2) Reinforce- 15 20 25 40 40 40 25 30 40
cover(mm) ment

Prestressing
steel 25 30 35 50 50 50 35 40 50

Notes
(1) In order to satisfy the provisions of EC 2. These minimum values for cover should be associated with
particular concrete qualities, to be determined from table 3 in ENV 206.
(2) For slab elements, a reduction of 5 mm may be made for exposure classes 2-5. A reduction of 5 mm may
be made where concrete of strength class C40150 and above is used for reinforced concrete in exposure
classes 2a-5b, and for prestressed concrete in exposure classes 1-5b. However, the minimum cover
should never be less than that for Exposure Class 1 .
(3) For exposure class Sc, the use of a protective barrier, to prevent direct contact with the aggressive
media, should be provided.

EXPOSED CONCRETE SURFACE

tlI••
mm
-::~:;M
11=-
~
25
25
CfC +
~~~ ~
12
20

501 501 501

100 100

g mg
sol

Fig. 3. Detail of surface reinforcement in aggressive environment with high concrete cover.
Durability 129

-Jct--
[Link]

I
JA ! 0·01 A ct. ext
s
I
I _]
I I

Fig. 4. Surface reinforcement as proposed by Eurocode 2.

,...-

r-- r- r--
r-- .....-
,_ ·- r-- r- r- f- 1- 1- -t- r- _40

r-

r--

EC2 CEB A
- - - D'" r- s GB
1- - - _15

P'IP
=-=
Dlt r '='
NL I
USA SU(CNR) -
I ( D.K. 27.07.85)

Fig. 5. Concrete cover (mm) as suggested in national and international standards.

130 M. Mele, E. Siviero

crack width crack width .... o.5 mm

~~~~~~~~~~n

Fig. 6. Penetration of chemical attack on a concrete surface after one year for a crack width
of 0,2 mm (6a) and of 0,5 mm (6b).

L011gitudinal uc lion of axially load~d sp•cim•n

Primary crack [Link] fore• on bar

-Jwl- Fore~ compontnfs on bar
(dut to tltdg• action and
dtformation of tuthJ

Fig. 7. Detail of crack near a reinforcing bar and local slip.

Durability 131

The protection of reinforcing against corrosion depends, as is known, on the continuous

presence of a surrounding alkaline environment which is guarantee by the concrete coating.
The thickness of the required coating depends both on the conditions of exposure and the
quality of the concrete itself (table 2).
In certain cases, for high concrete cover, it may be necessary to provide a skin reinforcing
both to control and to guarantee adequate resistance against detachment by parts of the
concrete (figures 3 and 4).
In fact, by looking at the concrete covers required in the various standards (figure 5) it
is clear that, with values in excess of 40-50 mm, the use of skin reinforcement is required.
This is justified by the conditions of chemical attack over the course of time in the
presence of cracking, as schematically shown in figure 6 which demonstrates the extent of
carbonation after a year of exposure for a crack of a size of 0,2 mm (figure Sa) and 0,5 mm
(figure 6b) respectively, whilst figure 7 gives the details of the crack by the reinforcing bar
together with the local slip. The danger of steel corrosion is therefore evident.
Accordingly it will be necessary to limit the size of the cracks and protect the structure from
corrosive environments.
It goes without saying that a further protection of the steel is not only appropriate but at
times even necessary.
This objective can be achieved by the use of bars protected by epoxy resins or galvanised
or in stainless steel or with recourse to cathodic protection.

3. Materials

3.1 Concrete

Knowledge of deterioration factors and the possibility of use of materials and resources
to overcome them nowadays makes it possible to design a concrete which is not only lasting
but strong.
Furthermore it is necessary to modify the habit of the mind of the designer who will
therefore not only use the strength characteristics of an ideal concrete to the maximum, but
also act so that the desired properties are maintained unchanged for the useful life of the
designed structure.
In fact, although concrete is the material which is most widely used in building, knowledge
of its properties and in particular of its behaviour over the course of time is very limited. An
adequate «Culture" for a material, the microstructure of which should also be considered, is
also lacking.
But in the field of design it can also be said that sufficient awareness in consideration of
all the properties of concrete is also lacking. A structure is in fact normally described by
geometrical details (section, height and dimensions in general), accompanied by efficiency
indications (strength or admissible stresses), all of these completely abstract as compared
with the complex reality of the material with the structure is produced. Even when, for
structures of a certain size, attention is paid to the characteristics of the material, considering
for example shrinkage, creep and the coefficient of thermal expansion of the concrete, at
times this is carried out with formal calculations without going into the physical reality or, if
necessary, the chemical reality of the material.
Indeed concrete demonstrates a substantial capacity for adaptation, generally
132 M. Mele, E. Siviero

corresponding quite well to expectations, but when exposed to unexpected stresses or

corrosion, at times it is subject to early deterioration.
The two fundamental parameters to obtain a lasting concrete are, as is known, the W/C
ratio and the strength class.
There is no doubt whatsoever that concrete must contain a minimum amount of cement
so much so that in the concreting regulations of the principal countries, including European
ENV 206, the minimum proportion of cement is prescribed together with the maximum W/C
ratio, according to the degree of corrosion of the environment and the maximum dimensions
of the aggregate (table 3).
The W/C ratio and degree of hydration combination determines, for the hardened cement
paste determines the degree of porosity which is linked to its strength and permeability.
At normal temperatures some of the constituents of Portland cement begin to hydrate
immediately in contact with water, but when the reaction products affect the anydrous cement
granules the reactions slow down substantially. At this point suitable curing assumes
fundamental importance and this consists in keeping the cement paste as saturated as
possible with water at least until the space filled by it has been substituted by hydration
products in other words until its capillary porosity is completely blocked. The hydration
process in fact can take place only when the vapour tension in the pores is some 80% of the
saturation value. In the vicinity of that value the hydration speed is at the maximum.
The suggestion of a practical nature to devote particular care to curing the poured
concrete during the initial period is therefore fundamental, not only in order to obtain optimum
development of strength but to avoid dangerous cracking from shrinking owing to drying and
in particular to make it impermeable if it is also to be lasting. It is therefore necessary to keep
the poured concrete as damp as possible until it has reached sufficient mechanical strength
properties, which can be evaluated from time to time in relation to the product and its use. Of
particular importance is the study of the cracking phenomenon, particularly in relation to the
time of appearance of the cracks (figure 8).
Before of during setting, cracks can be:
a) of an intrinsic nature owing to the phenomenon of cracking and plastic settlement (figure 9);
b) of an extrinsic nature owing to frost, heat variations, the giving way of formwork (figure 1O);
After setting, cracking can be:
a) of a physical nature; porous inert, shrinking, cracking;
b) of a chemical nature: corrosion, alkali-silicon, carbonation;
c) of a thermal nature: thermal cycles, hydration heat, accelerated maturing;
d) of a structural nature: external loads (figure 11 ), deferred phenomena (figure 12), dif-
ferential settlement (figure 13), adhesion (figure 14), errors in fitting reinforcing (figure 15).
A table of cracking which shows the various aspects mentioned above appears in figure
16, whilst figure 17 shows the details of cracking possible in structures.
It is therefore necessary to stress the snare of cracking which can occur during setting.
In fact, during the hardening, the concrete gradually acquires a tensile strength, although
limited, whilst the presence of the reinforcing prevents it from becoming strained. Cracking
commences after a few hours from the poring when the tension generated by the prevented
straining overcomes the tensile strength (figure 18) with simultaneous net reduction in
deformability (figure 19).
From what has been stated above it is therefore essential to design carefully the concrete
mix having recourse, where necessary, to use of appropriate superplasticizing additives to
reduce the W/C ratio.
Durability 133

Tab. 3. Durability requirements related to environmental exposure (ENV 206).

Requirement 1 2 3 4 5 6 7 8 9
Exposure class according to table 1
1 2a 2b 3 4a 4b Sa 5b Sc(1)

max w/c ratio for (2):

- plain concrete
- reinforced concrete
-
0.65
0.70
0.60 0.55 0.50 0.55 0.50 0.55 0.50 0.45
- prestressed concrete 0.60 0.60

min. cement content (2)

inkg/m3for:
- plain concrete 150 200 200 200
- reinforced concrete 260 280 280 300 300 300 280 300 300
- prestressed concrete 300 300 300 300

min. air content of fresh

concrete in %for nomi-
nal max aggregate size
of(3l: (4) (4)
- - - 4 - -- - -
--
32mm 4 4
-- 16mm - 5 5 - 5 - --
Bmm - 6 6 - 6 - -
frost resistant
aggregates <6> - - yes yes - yes - - -
impermeable concrete - - yes yes yes yes yes yes yes

sulphate resistant ce-

types of cement
mert (5) for sulphate
for plain and reinforced
contents:
concrete according to
> 500 mgAcg in water
EN 197 > 3000 rnglkg in soU

The above values of w/c ratio and cement content are basad on cement where
there is long time experience in many countries. However, at the time of
drafting this pre-standard experience with soma of the cements standardized
in EN 197 is limited to local climatic conditions in soma countries. Therefore
during the lila of this pre-standard, particularly for exposure classes 2b, 3, 4b
the choice ol the type of cement and its composition should follow the
regulations valid in the place of usa of the concrete. Alternatively the
suitability for the usa of the cements may ba proved by tasting the concrete
under the intended conditlonl of usa. Additionally, cement CEI may ba used
generally for prestressed concrete. Other types of cement may ba applied if
experience with lhasa types is available and the application is allowed by the
regulations valid in the place ol uaa of the concrete.

--ndio
(1) In addition, lha concra1e 111811 ba pn~IICIId against dinlct wllh lhe aggrllllive mecla by coatings unless lor
particular - such prOI8CtiDn is -..iderad u - . . y .
(2) Pozzotana or lal8nl hydraulc addllionl may ba taken inlo _ , lor lha minimum ~~ oontant and lha
maximum only if permillad by~ oflha na1ionlll sllndMII and raguiiiMons valid in lha
plllce ol usa.
(3) Wl1h aSI*ing faclllr oflha entrand air void syatlm > 0.20 mm .-urwct on lhe hMianad concnte.
(4) In cues wharalhe degree of aaturalion is high far piQiongad perioda ol time.
Other values or - may apply if lha ~ is •lid and doc:umenled ID haw adequate lrostl'lllis-
aooonling ID lhe ragulalionl Vlllid in 1he plaOII ol usa oflhe ooncnta.
(5) The sulllh* riiSisla.- of lha oement 1111111 ba judged on lha baais of national standards and ragulatione in lhe
place wh.w l h e - is Ul8d.
(6) Asleaed against 11ut ragulalionl valid in lhe plllce of usa oflhe - ·
134 M. Mele, E. Siviero

CAUSE Dl FESSURAZIONE
CAUSE OF CRACKING

CARICHI ESTERN!
LOADING SERVICE CONDITIONS -
REAZIONE ALCALI-SILICE
ALKALI-SILICA REACTION ---
---
......... ...-
CORROSION£
CORROSION
... __ ---
.._ ......
RITIRO OA ESSICCA11ENTO
DRYING SHRINKAGE - r----
CONTRAZIONE TERI11CA
EARLY THERMAL CONTRACTION --
-- ,_
,_ r- !...

RITIRO PLASTICO
PLASTIC SHRINKAGE - ,--- ........
ASSESTAI1ENTO PLASTICO 1
PLASTIC SETTLE11ENT -~- ....
ORA GIORIIO SETT. HfSf ANNO SO ANN!
HOUR DAY WffiC IIONTH YEAR SO YfARS

Fig. 8. Time of appearance of cracks.

sfinups

Fig. 9. Cracks due to plastic settlement.

Durability 135

formwor

formwork

Fig. 10. Inadequate formwork.

( Pure flexure

--f ) \ ( ~+-- Pure tension

Shear

)
Torsion

HeficaJ crncl<ing

~------} Concentrated load

Fig. 11. Load induced cracks.

136 M. Mele, E. Siviero

FESSURE !CRACKS!

~ FESSURE ICRACKSI
I~

Fig. 12. Cracking due to coupling of different Fig. 13. Cracking due to differential settlements.
age concrete in a wall.

crack theorotic•l

FleXUtal cradc
Bond

~---ts=-+.~.~-r"'l

Fig. 14. Cracks due to lack of bond Fig. 15. Errors of placing of reinforcement
and anchorage of bars. in cantilever.
Durability 137

Fig. 16. Examples of cracks in hypothetical concrete structure.

Fig. 17. Alignment of cracking pattern relative to reinforcement on top of a slab.

138 M. Mele, E. Siviero

Classification of cracks as in figure 16.

Type of Letter Subdivision Most common Primary, Secondary Ramedy(- rmec1

cracking (see location cause causes suming basic appearance
figure) (excludng factolll redesign is
rasllainl) Impossible)
in all cases
reduce
rastraint

A Oierraft. Deep
forcement sections
- Excess Rapid early Reduce Ten minutes
Plastic B An:hing Topel blaedng drying bleeding to thrae hours
setllement columns conditions (air entrain-
- men!) or
ravibrate
c Change Trough and
of depth wallleslabs

0 Diagonal Roads
and slabs
- Rapid early
Plastic Reinlon:ed
shrinkage E Raxlam concrete Law rate Improve early Thirty minutes
slabs ofbleedng curing to six hoUill
-
Oierrain- Reinlon:ed Dillo plus
F larcement concrete steel,_,.
slabs su"-

G External Thick walls EXC888heal

rasllaint generation

Eally - Rapid cooling Reduce heal

andlor
One clay to
twoorthrae
thermal lnlllmlll Excess insulate weeks
contraction H restraint Thick slabs temparaiUra
gndlenls

Long-term Excess Reduce water Several

drying Thin slabs Inefficient shrinkage content weeks
shrinkage (and walls) jointa Inefficient Improve or months
curing curing

J AgU!st Fair faced Impermeable

fllnnwork concrate lonnwor1l Richmi- Improve One to seven
Crazing
- curing
a finishing
days
sometimes
Floalad Slabl 0.. Poor curing much!-
K concreta mwllng

L Nllllnl ~and Ladl

[Link] of cover
Conasion
of rainfarce- - Poor quality
concrate
Elininala Mora than IWO
c:aus. listed years
mant Calcium Pr8csst Excess
M chloride concreta calcium
chlorida

[Link]- (Oinp Rnclive aggregal8 plua Elininate Mora lhan five

gregaae N locations) high-alkali csment c:[Link] years
reaction
Durability 139

~
c:
"...
.
.........
"...
.s"
."
..
"...

hardening time in hours

Fig. 18. Evaluation of strength and restraint stresses in young concrete (schematic).

- •
....
~
0,4

- o.z
.s O,J
..
a

"'
"'"'t:l
·-"'e
";
0,1

0
1 z 4 6 B 10 h 1 7 ZB days
ag~ of concr~t~ (I ogarithmic J

Fig. 19. Ultimate tensile strain of concrete depending on age.

Marked progress in the preparation of high-durability concretes has recently taken place
with use of special additives having a pozzolan and thixotropic action which in addition to the
densifying effect arising from the reduction of water owing to the presence of plasticizers,
produces a further increase in the microstructural density to fill the interstitial voids between
the granules of cement with particles of amorphous silicon, much smaller than the cement
granules (figure 20).
 140 M. Mele, E. Siviero

Fig. 20. Additive with high pozzolanic and tixotropic action.

In that way, an extremely dense and compact concrete is obtained, capable of physically
resting the penetration of corrosive agents and also incapable of chemical attack owing to the
absence of lime.

3.2 Steel

As indicated in paragraph 2, the protection of the reinforcing against corrosion depends

on the continuous presence of an alkaline environment in which the metal bar is immersed.
In that case a microscopic layer of passivated ferric oxide is formed on the surface,
preventing anodic dissolution.
When the pH falls below value 9 owing to carbonation of the concrete or attack of the
chlorides, chemical attack, which spreads speedily to the final condition, begins.
A schematization of the process is given in figure 21.
In those conditions, it appears necessary to have recourse to protective measurers for the
bars to prevent deterioration.
To that it is possible to use:
-bars covered with epoxy resins;
- galvanised bars;
- stainless steel bars.
Alternatively what are known as «corrosion inhibitors» can be used.
With regard to bars protected by epoxy resins, their initial use goes back to 1973 with
applications in the United States. The results obtained seem to be quite good although the
risk- not to be neglected- of damage to the protective film which can take place for various
reasons, particularly during the transport and use phase, persists.
This aspect calls for particular care in the protective coating quality control process. In fact,
with this technology, the protection provided to the bars against corrosion consists essentially
Durability 141

....
c:
....
.!:! ....z
E
..... "'[Link]
0
~ 0
Gl

c
Gl
....0

I< initiation >1</ropagatioolj Time

INIZIO ROPAGAZI E TEMPO

Fig. 21. Scheme of reinforcement corrosion.

of a physical barrier, with generally no chemical interaction between the protective film with
its interface with the aggregate. Furthermore, one aspect which has not yet been completely
solved is that of aging of the resin and its reaction of the course of time with organic
substances which may be present in the concrete.
With regard to galvanised bars, these have become widely used over recent years.
With dip-galvanising at 450-460 degrees C, a zinc thickness of approximately 70-100
microns, sufficient to protect the steel from corrosive agents, is generally obtained.
This protection acts at physical level, as a true and real barrier and at galvanic level, with
the zinc and steel immersed in the concrete which acts as an electrolyte where, owing to the
difference in potential, the zinc behaves as an anode in relation to the steel, preventing its
rusting {figure 22).
Furthermore, in current circumstances, there is no general consensus on the corrosion
behaviour of galvanised reinforcing bars in time terms in relation to the level of alcalinity of
cement, the use of additives, the variability of the W/C ratio, partial or total exposure to the
corrosive environment.
It seems however that one can state the following:
- galvanised steel tollerates a higher concentration of chlorides than ordinary steel;
- in exposed structures in a marine atmosphere the deterioration of the steel coating is
greatly delayed and even reduced when zinc bars are used;
- with highly alkaline concretes a loss of protection of galvanised bars is possible. However
this phenomenon can be adequately controlled with the use of chrome additives.
In any event, using galvanised bars, it seems to be effectively possible to reduce the
coating of concrete, at least in normal type environmental conditions.
 142 M. Mele, E. Siviero

CALCESTRUZZO CALCESTRUZZO
BAGNATO WET CONCRETE INOURITO HARDENED CONCRETE

SAL
insolub. Zn
LEGiiE Zn-Fe UNSOLUBLE SALtEGHE Zn -F e
ALLOY ALLOY

Fig. 22. Protective action of the zinc in reinforcing bars.

The use of stainless steel bars which, in current circumstances, represent the best offer
on the market, but less so from the technical aspect, certainly appears to be effective.
Although the use of high bond stainless steel bars as reinforcing for concrete is quite
recent, knowledge of their behaviour under chemical attack is well known and depends
mainly on their chemical composltion.
Its somewhat high initial cost justifies its general use in the entire structure only in very
important cases in particularly corrosive environments, for example: structures subjected to
freezing and unfreezing cycles, tunnels, marine structures, off-shore structures.
A reasonable compromise between the economy aspect and the technical result can be
achieved by the use of stainless steel as surface skin reinforcement where high cover is
required.
Finally, some notes from professor Pedeferri on cathodic protection, recently introduced
in Italy.
ccCathodic protection takes place by making the potential of the structure to be protected
more negative in relation to the corrosive environment so as to eliminate the corrosion
reaction or to reduce it to negligible values. The polarization effect is obtained by circulation
of a weak direct current from the environment to the structure which therefore functions as
a cathode, with the use of a second electrode functioning as an anode.
In practise the cathodic protection is achieved in two different ways:
with ccsacrificial anodes», that is by coupling with a reactive metal, such as zinc, aluminium
or magnesium, which gives off electrons to the steel by means of a short circuit-current
and, simultaneously, sends positive ions in solution, becoming corroded;
- with an ccimposed current.. or by means of an external electrical circuit in which an
electromotive force is inserted acting as a pump which sends electrons to the earth to be
protected cathodically, taking them from an inert metal immersed in the solution which
functions as an insoluble anode.
Durability 143

The cathodic protection of the reinforcing in concrete applied in the past to prestressed
piping and underground tanks, such as structures at sea, has been becoming widespread in
recent years for the protection of bridge decks, in particular when these .have been
contamined by antifreeze salts.
In electrochemical terms the effect of this protection is that of causing the cathodic
polarization of the reinforcing and maintaining the potential of a steel at a range in which the
corrosion reaction stops or proceeds at a negligible rate. The reduction in the potential entails
the stoppage of corrosion even in the present of chlorides.
In the design and realisation of cathodic protection imposed current installations, the most
complex aspect is that of design of the anode structure in relation to the high resistivity of the
concrete and the limited volume of electrolight (concrete) through which the current is
distributed.
The first imposed current installations used individual anodes in cast iron, silicon iron or
graphite, supported by the slabs and covered by a layer of conductive asphalt as secondary
anode, in order to distribute the current from the primary anode to the whole of the surface
to be protected and to limit acid attack at the anode/concrete interface. Installations produced
in accordance with this concept operate but require maintenance of the conductive asphalt
coating.
Subsequently anodic structures consisting of titanium or platinum-niobium wirte with a
backfill of conductive material were produced, housed in a series of parallel channels in the
metal covering, section 2x2 em approx. and length 10-12 m, spaced by approximately 30 em.
In order to achieve the channeling, avoiding an effect on the reinforcing with the resulting short
circuiting between the reinforcing and the anodic structure, an adequate metal cover
thickness is necessary. Currently the development of anodic structures is directed towards
grid systems, possibly available in the form of plastic or metal based networks, to be located
on the bridge to be protected, and to be anchored with a layer of concrete to which a
waterproofing layer can eventually be applied».

4. Design aspects

The concept of durability must from the outset lie behind the design choices. As mentioned
in the introduction, the immediate economic aspects which, in the more recent past, have led
towards incorrect choices, can no longer constitute the sole factor for evaluation.
Systematic resource to prefabrication as a technology considered to be in the forefront,
has proved to be disappointing. Particularly in relation to the lack of attention in the realisation
phase, it has created more problems than it has solved.
Work site industrialisation and part prefabrication with pouring during the completion work
can certainly find ample space provided that the operating methods are carefully evaluated
by carefully studying the constructional details and providing the most suitable devices to
facilitate inspection and maintenance over the course of time.
The design and its realisation, musttherefore be closely integrated, particularly in the case
of important structures.
Realisation in several phases enormously conditions the state of stress, not only in the
transitory phases but also in the final utilisation states.
The structure has a memory of the phases passed through and the structural calculation
must take due account of this.
144 M. Mele, E. Siviero

In general terms, the predominant line is now, in primis, the elimination where possible
of transversal joints which have to be designed with perfect impermeability. It will therefore
be appropriate to have recourse to hyperstatic schemes of the continuous type or, if they are
isostatic, those which at least make the slab continuous.
Gerber hinges must be eliminated or, where they appear, designed with ample space for
inspection and maintenance (figure 23).

! I ...
I \
--- !
I

!
I I
I
i ~-- !
I ~~ I

I I
I I

Fig. 23. Detail of a Gerber hinge.

A suitable and robust transition slab between the upset and the shoulder must be provided
to eliminate the ccstep.. generated by the settlement of the upset and thus avoid early
damages to the structure. Slenderness of structural parts must be limited both globally in the
height/span ratio of beams, and locally in the thickness of slabs, webs, spans of cantilevers
etc. In that way strains and stresses are contained. The greater thickness available also allow
for a safer and more easy positioning of the reinforcing and possibility of guaranteeing more
reliable concrete pouring (figure 24).

• • • •• •• • • •
• •••••••

CONSIGLIA TA
RECOHHENOEO

: I •• ••
: I : : ••••••••
Fig. 24. Bar arrangement in cross sections.
Durability 145

In that sense a limitation of the minimum thickness of slabs and webs appears to be
opportune in particular, assuming:
- for slabs a lower limit of 20 em;
- for webs a lower limit of 14 em, never going beyond 1/1 Oth of the height in the case ofT
or doubleT beams or below 1/12th of the greatest dimension in the case of the walls of
box beams.
Particular attention should be paid to the maximum stresses on the materials, bearing in
mind the fact that the values admitted by the Standards are frequently found to be excessive
in relation to the phenomena of cracking and fatigue which have become apparent after a few
years of service.
Another important aspect is the geometry of the transversal section of the structure. This
must be of compact form, reducing the surface-volume ratio to minimum (figure 25).

c ___ CONSIGLIA TA
RECOMMENDED
SCONSIGLIA TA
NOT RECOMMENDED

Fig. 25. Geometry of cross sections.

In such a way, in addition to a reduction of exposure to corrosive agents, the advantage

-not to be neglected- of the convenient placing of the reinforcing is achieved, avoiding
dangerous congestions and thus facilitate the concrete pouring.
With regard to the reinforcing, the correct design should be aimed at reducing cracks and
ensuring good anchorage in the concrete. Anchorages should be offset, avoiding the
simultaneous ending of a group of bars. Insufficient anchorage may in fact induce dangerous
cracking, local chipping as far as loss of the balance of the beam (figure 26, 27, 28).
Off-load thrusts which are caused by the need to balance the system of transversal forces
originating from the anchorage should also be avoided. Adequate coverings and suitable
bracing should therefore be provide (figure 29, 30).
In any event, it should be recalled that the layout of the reinforcing should be directed,
where possible, in the direction of the isostatic tensile lines (figure 31, 32).
But there are further aspects to be considered with the utmost attention in the design
phase. For example, the elimination of water/concrete contact is of fundamental importance
tor the life of the structure.
146 M. Mele, E. Siviero

Nz

/FESSURA POSSIBILE
POSSIBLE CRACKING

Fig. 26. Anchorage of a bar: Fig. 27. Anchorage of a bar:

stresses in concrete. necessity of transverse reinforcement.

TRAIETTORIE OJ COMPRESSJONE
COMPRES~ION [Link]

FESSURE
'Nz

ROTTURA LOCALE
LOCAL BREAKING

Fig. 28. Detail in the support.

Durability 147

INTERRUZIONE INTERRUPTION
.6Ny
VARIAZIONE 01 TENSIONE STRESS VARIATION
... +1-,...;-..,.--l·t..,._d...,_._ _+
~ 0 •• •

Nz:6Nz

SPINTf • ·vuoTo· SPJNT! • ·vuoTO"

UN![Link] T!D THRUST UN!~UilJIAAT!O THRUST
AISCHJO OJ fESSURA
IIISX OP CRACKIN&

Fig. 29. Unequilibrated thrust Fig. 30. Unequilibrated thrust

due to anchorage of a bar. due to stress variation.

,)11]
. ·~.
.6 11 21
I
) I
I

SEZ. 1-1 SEZ. 2-Z SEZ. 3-3 -f-!!+

~~ J~
az 'Czy az "'zy
d liZ l't'zy•OI
FASE 1
STAGE 1

SEZ. 1-1 SEZ. Z-2 SEZ. 3-3

~~~ JJV }:-(

ar ar az e12 111 az [Link]

Fig. 31. Stress distribution in a simply supported beam uniformly loaded (stage 1).

a'

FASE 2

STAGE Z

Fig. 32. Stress distribution in a simply supported beam uniformly loaded (stage II).
148 M. Mele, E. Siviero

Washing (figure 33) and dripping (figures 34, 35) should therefore be avoided by an
adequate design of water run-off (figures 36, 37, 38, 39) and of the water-seal joints (figure
40), with the possible replacement of deteriorating items (figures 41, 42) and, as mentioned,
providing the possibility of inspecting the various non-visible parts.

CONSIGLIA TO
RECOHHENOEO

SCONSIGLIA TO
NOT RECOHHENOED

Fig. 33. Washing.

:czzzz~

Fig. 34. Dripping. Fig. 35. Dripping.

Fig. 36. Lack of water discharge.

Durability 149

i
~ecidentall water filled
blow-up tf lt.g. luky dr1inl
w1tor froozu

outlet nttdtd

Fig. 37. Lack of water discharge. Fig. 38. Lack of water discharge.

8 }··~"
dificult to

+ II
II
il ~ 0111nttin
and

0
Orton pipe
r1pair
:I
II usy to

Fig. 39. Water discharge in a pile.

JOINT

CONCRETE CROSS
-+- HEAD BEAM

COLUMN

Fig. 40. Impermeable joint.

 150 M. Mele, E. Siviero

CALCESTRUZZO OEGRAOATO SURFACE CONCRETE CORRODED

IN SUPERF,ru--«~~~~
1

Fig. 41. Surfaces subject to corrosion. Fig. 42. Replaceability.

Finally one comment on construction techniques. The «disappointment» at the poor

results obtained with integral prefabrication, particularly in view of the limited care devoted
to the design and realisation of the joins, nowadays recommends that prefabrication should
be limited to the longitudinal beams and that the slab should be cast in situ.
In case of prestressed reinforced concrete, the need to ensure reliable adhesion between
the prestressed reinforcing and the concrete should be stressed, not only for static reasons
but also in order to protect the cables.
In many instances the tendency to use pretensioning bars rather than post-tensioning
cables can be seen. The more obvious reason is that since there is no absolute guarantee
of perfect success of the injection, any corrosion of the cable is decidedly less dangerous in
terms of safety.
Although this is certainly true, it nevertheless is likewise true that in order to avoid
unnecessary penalties, the problem should be solved upstream with the necessary controls
in order to guarantee quality.
An a priori disapproval of post-tensioning cable beams does not appear to be justifiable
both in view of their greater rationality from the static aspect and efficiency of reinforcing, and
because the problem of injection does not present particularly important difficulties when
appropriate materials and techniques are used.
When, however, the technique of prestressing with external cables is adopted, this being
particularly frequent in restoration work, it is preferable for the protection of the cables to be
ensured by adopting reinforcing of the bonded type (bars precoated with resin or strands
lubricated and coated with PVC) rather than recourse to a sprayed concrete coating (which
tends to crack easily) or galvanisation (which brings in problems of fatigue strength and life).
In addition the cables have to be fixed to the beams by suitable devices capable of taking up
the offload thrusts exerted by the reinforcing where the reinforcing changes direction; finally
it is appropriate for these devices to be provided with rather limited spacing (not more than
ten metres) both in order to attenuate cable vibration and to constrain the cables to follow the
deformation of the beams as much as possible.
Finally, again from the point of view of constructional realisation, it should be pointed out
that progress in concrete technology currently makes it possible to carry out in situ and in any
environmental condition, concrete pouring which is just as reliable as when effected in the
prefabrication works, at the same time accelerating production cycles. This explains the
Durability 151

current widespread tendency towards structural solutions which are cast in situ which, no
longer penalised by the quality of the concretes and construction times, present all the
advantages of monolithic structures and uniformity in the rheological properties of the
concrete.
These results are achieved by designing the concrete with the use of plasticizing, super
plasticizing and hyperplasticizing additives by means of which it is possible to increase the
mechanical strength even at low temperature, the impermeability, the dimensional stability
and durability in the presence of high thermal variations.
The use of super-plasticizing additives furthermore may constitute a valid alternative to
the heat treatment of concrete (steam curing). It is possible in fact to substitute a complete
cycle (14 to 22 hours at 70°) with:
- simple addition of superplasticizer in non-winter months;
- addition of a superplasticizer integrated with a light thermal cycle (2-3 hours of steam) in
the winter months, achieving in each case after 18-20 hours, the same mechanical
strengths as can be achieved with a complete steam cycle.
Furthermore, the elimination of this latter treatment makes it possible to achieve further
advantages such as:
- the elimination or reduced use of equipment necessary from the application of heat cycles;
- the increase in mechanical strength on long curing;
- the reduction of the time dependent phenomena (shrinkage and creep) and therefore
better behaviour of the structures in service.
It is therefore possible to obtain reliably, in any climatic condition, concretes which have
after 18-20 hours of curing, a strength equal to some 70% of the required characteristic, this
latter characteristic being reached after three days of curing. It is obvious that this cannot fail
to limit interest in the prefabricated structural solutions and should indeed direct the attention
of the designer and builder towards solutions which are overall more correct provided that the
rationality and constructive economy are based on industrialisation of the worksite and
reliability of the materials used there, rather than on simple saving of time obtained by
reducing work in situ to pure and simple assembly operations.

5. The quality assurance

In accordance the quality of concrete structures is currently entrusted, partly to the

designer who has the responsibility for the design of the structures and partly to the site
manager who will comply with the provisions of Law, checking compliance between the work
and the design and also observance of the regulations relating to the implementation of the
design and the quality of the materials used. In general, all this takes place with the
supervision of the work and the carrying out of some tests on the concrete and on the steel
and, furthermore, a static test in accordance with procedures largely laid down by Law.
Load testing is frequently carried out on completed structures. Obviously a control
procedure of this type finished up by being somewhat reductive. Much more significant is the
result obtained when the test is carried out during the course of the work. The concept of
testing during the course of the work is adopted with ever greater frequency in order to give
better guarantee for the Administrations, giving wide terms of reference to the technician who
has to express himself with an official report and operate with real possibilities of action.
 152 M. Mele, E. Siviero

Control must therefore relate to the designing, realisation of the work but also the activity
of the site Manager on behalf of the customer.
What is known as the final acceptance must provide for complete description and a
geometrical survey of the work with an indication of any discrepancies from the design, errors
in subverticality of the piles, cracks existing at the time of construction for reasons of various
types, checking the state of deformation of the structural elements owing to the effect of
permanent loads. In addition to the tests on the material, the test certificate must contain
details relating to the tightening of the cables, the recorded loss, the injected operations.
This document is intended to assume the function of a real «record of birth» of the
structure to which constant reference may be made in the inspection of the structure for
further comparison with the initial situation when an anomalous fact Is found during the life
of the structure.
All this procedure appears to be somewhat limitative and, in fact, nowadays the concept
of a quality of construction obtained by means of organised procedures which involve all
persons participating in the realisation of the work.
According to that concept, quality is not considered as an abstract concept but a group
of measurable parameters and not from a deterministic point of view but rather a probabilistic
aspect. Quality assurance is a mothodology which, according to systematic and planned
actions, is capable of supplying an adequate degree of reliability that a structure and its parts
will provide the required performance an the envisaged operating conditions. This is, as
already mentioned, in relation to the following aspects:
- for the design, in order to reduce the probability of occurrence of human errors;
- in the realisation, in order to identify and correct insufficient results in the productive
process;
- in maintenance, as a control that the safety measures remain unchanged over the course
of time.
That methodology, which must involve the entire constructional process (promotion
design, materials, realisation and use) is described in various manuals, e.g. «Quality
Assurance Systems for Concrete Construction, published by the American Concrete Institute»
(ACI, Journal7/85) and also with regard to structures in reinforced concrete and prestressed
reinforced concrete, in the more recent Eurocode 2, 1989 and CEB-FIP Model Code 1990.
A first aspect relates to the measures for the prevention of human errors which must be
implemented.
In all the technical procedures of the construction process, e.g. adequate structural
analysis; drawings, complete and easily readable lists and documents; application of
systems not very sensitive to errors; procedure for construction of a simple type; structure
easily accessible for inspection and maintenance etc.
In the organisation and constructional process management sector, for example: clear
definition of the task and responsibilities of the employees and their co-operation.
In the field which influence human behaviour, e.g.: measures aimed at increasing the
capacity and motivation of the employees and reducing the influence of intentional or
unintentional disturbance factors; continuous construction and training etc.
According to the CEB-FIP Model Code 1990, a plan generated for quality assurance must
concern the following aspects at least:
1. Organisation of the personnel at the various action levels (responsibility authority, inter-
relationships) etc;
2. The way to achieve the required objectives;
Durability 153

3. Project control including procedures for variants during the course of the work;
4. Control of documents both in view of their reliability and the capability of tracing them at the
appropriate places and within the required times;
5. Supplies relating to services, products acquired, sub-supplies;
6. Production and construction;
7. Data recording.
It should be pointed out that this latter aspect is of extreme importance for the purposes
of future inspections for the programme relating to maintenance over the course of time.
Measures to identify and correct insufficient results are implemented by quality control which
is one of the aspect of quality guarantee.
The quality control consists of:
- acquiring information;
- formation of opinions based on information;
- decisions based on opinions.
This is explained by the following phases:
- individual checks;
- internal control;
- external control;
- compliance control.
The control is effected for various phases of the constructional process:
- the design;
- the materials and the components, the production and the construction;
- the complete structure;
- the realisation.
With regard to the design it is necessary to carry out:
- check of compliance between the specifications laid down and those effectively adopted;
- verification that the calculation models and the numerical models are correct;
- verification that the drawings and other design documents are in accordance with the
calculations carried out and the required specification.
The calculation controls can be effected using various techniques:
- direct verification following the design step by step;
- parallel verification carrying out the calculations independently of the original design;
- cross verification when the results of the intermediate and final calculations are used at
input data to demonstrate validity.
The design controls must check that:
- the calculation results have been correctly transferred;
- the documents are consistent one with the other;
- the drawings are clearly interpretable without ambiguity.
The control materials and the components provide for two phases:
- production control (production process);
- conformity control (identification, testing, compliance).
The production and construction control relate not only to the materials and components
but also to:
- connections and assemblies;
- construction and erection procedures;
- structure geometry.
Table 4 summarizes the various aspects dealt with.
154 M. Mele, E. Siviero

Control during the operation of the structure must provide for the following checks:
- condition of any protective systems;
modifications in materials the originating and/or progression of faults;
the local and general deformative condition.
- the deterioration of the materials and/or components which may reduce structural safety.
From this aspect, the responsibility for control is clearly that of the owner who must
therefore acquired and record all information suitable for defining the programme for the
necessary maintenance and servicing.

Tab. 4. Objects of material and component production and construction control (Eurocode 2).

Object Control of materials Control of construction

and production and wor1<manship

Concrete Constituent materials Transport, placing, compacting

Composition, Curing
production Surface finish
fresh concrete
hardened concrete

Formwor1< and falsewor1< Materials Robustness, erection,

removal, cambering,
deflections, ground supports,
tightness, internal surface

Reinforcement Specified material properties Handling and storage, cutting,

Surface conditions assembling, fixing, laps and
joints, welding, placing
Cover to reinforcement

Prestressing steel and devices Specified material properties Handling and storage
Surface condition Cutting
Prestressing devices Placing
Straightness of tendons Prestressing devices
Grout Tensioning
Grouting

Structural members; Dimensional deviations

Precast Units Camber, deflection
Compliance with the order
Durability 155

6. Theoretical measurement of durability

According to the proposals of the above mentioned working group of GTE for the MODEL
CODE 1990, durability of a structure is verified when the required service life T. does not exceed
the conventional design life Tc. This is defined by means of a standard formula representing
the various indexes as partial factors in multiplication terms:

where:

To the conventional limit of life assumes for example equal to 100 years;
<X; partial factors which take account of the various aspects involved as follows:
a, depending on the W/C ratio (table 5);
~ depending on the workability of the mix (slump value) and the technique for the
production of the concrete (table 6);
a3 depending on the class strength of the concrete (table 7);
a4 depending on the curing and environmental conditions (table 8);
a5 depending on the condition and type of surface finish (table 9);
a6 depending on the cover thickness (table 10);
~ depending on the class of exposure of the environment not taken account elsewhere
(table 11 );
a 8 depending on special situations relating to the structural typology (table 12).

Tab. 5. Water content factor a,, referred to water/cement ratio of concrete mix.

W/C s; 0.45 0.46-0.50 0.51-0.55 0.56-0.60 0.61

;:>:

a1 1.00 0.95 0.90 0.80 0.70

Tab. 6. Workability factor~· related to the consistence of fresh concrete (slump test in em) and to
special techniques of production.

humid plastic semifl. fluid extruded

slump S5 6+10 11+15 <':16
a2 0.80 0.90 0.95 1.00 1.00

Tab. 7. Resistance factor a 3 , related to the strength class of concrete.

class 16 20 25 30 <':35
CX3 0.75 0.85 0.90 0.95 1.00
 156 M. Mele, E. Siviero

Tab. 8. Curing factor a 4•• related to the curing provisions and to the ambient conditions.

ambient
treatment sheltered medium severe

open 0.95 0.85 0.70

protected 1.00 0.95 0.85
heated 0.95 0.95 0.95

Tab. 9. Structure factor lls· referring to the service state of reinforced concrete elements and to their
surface finishing.

service state surface finishing

free-face current special

Exposure classes 1 and 2a (covered construction)

cracked 0.85 0.95 1.00
uncracked 0.95 1.00 1.00
compressed 1.00 1.00 1.00
Exposure classes 2a (uncovered) and 2b (covered)

cracked 0.70 0.85 0.95

uncracked 0.35 0.95 1.00
compressed 0.95 1.00 1.00
Exposure classes 2b (uncovered), 3, 4 and 5
cracked 0.55 0.75 0.90
uncracked 0.75 0.90 1.00
compressed 0.90 1.00 1.00

Tab. 10. Cover factor a,, referred to the increased design value c over the minimum concrete cover
c adopted for reinforcement detailing.

c-c(nm) 0 5 ~10 plain concrete

06 0.90 0.95 1.00 1,00

Tab. 11. Exposure factor~· due to environmental conditions effects not included in the other factors.

1 2 3 4 Sa Sb 5c
1.00 0.95 0.90 0.85 0.80 0.70 0.60
Durability 157

Tab. 12. Special factor Cla· due to particular cases of structures:

A- «any other case";
B- prestressed concrete elements with unbonded post-tensioned internal tendons with
grease protection;
C - prestressed concrete elements with bonded post-tensioned tendons in grouted
internal ducts;
D- prestressed concrete with bonded tendons having sheathed unbonded parts;
E - elements with enclosed metallic devices protected by simple surface grouting.

situation exposure classes

1, 2a (covered) 2a (uncovered), 2b (covered) 2b (uncovered), 3 - 4 - 5
A 1.00 1.00 1.00
B 0.90 0.85 0.75
c 0.95 0.90 0.80
D 0.95 0.90 0.80
E 1.00 0.95 0.85

A further step in this direction can be taken by considering two additional partial factors
in relation to the quality assurance process and control activity during the lifetime:
exg depending on the quality control during the design and realisation;
a.10 depending on the inspection and maintenance programme.
This is a first attempt at tackling this matter in an organised manner. Furthermore, beyond
the numerical indications contained in the tables and which should in any event encounter
adequate experimental reflection, a fundamental merit of the proposed method is the
indication of a possible path to be followed.

7. Conclusion

To conclude we would like to cite several comments which Christian Menn elaborated at
the XI congress of the FIP held at Hamburg in June 1990 on the topic« The place of durability
in Bridge design concepts»: «In the construction of bridges it is not particularly difficult to
increase the durability- Bridges, like every other structure are designed, in questions of detail
too, according to certain basic fundamentals and such must be executed with care using the
ideal materials. The damages that have considerably reduced the lifespan of the structure are
in great part consequence of errors in construction, inappropriate use of materials, and lack
of maintenance. These errors are now well known and it is the task of engineers to avoid them.
Durability factors cannot be attained by employing new untried technologies that absolve the
designer and the owner of their responsibility of a quality project control. Old errors must not
be repeated for the sake of a reduction of construction costs and completion time ... Also to
mention is that «the engineers must also realise that an effective strategy to increase the
durability must be based on analysis of detail and a rational application of material technology.
We must finally free ourselves from exhaustive ccexact.. analysis that takes up a lot of energy
and distracts us from the real problems. Durability can be guarantee with a minimum of
calculation ...
 158 M. Mele, E. Siviero

To confirm the above we can, without going back too far in time, observe the well thought
out bridges of Robert Maillart, jobs well carried out, and for the main part in a decent state of
repair almost fifty years on. When the project is well conceived and execution is correct,
suffice a few protective measures combined with satisfactory checking procedure, consideration
of the climate, effective inspection and maintenance to guarantee a lifespan respecting the
expectations of those concerned.

References

1. Vitruvio: Trattato deii'Architettura.

2. Comportement en service, entretien et reparation, CEB Bulletin d'information n. 138 AoOt
1980.
3 Durability of concrete structures, State-of-the art report, CEB Bulletin d'information n. 148
Fevrier 1982.
4. CEB-RILEM International workshop "Durability of concrete structures" Copenhagen,
May 1983, Workshop Report, CEB Bulletin d'information n. 152, April1984.
5. Comportement en service entretien et reparation, CEB Bulletin d'information n. 163,
Mars 1984.
6. Draft to CEB-GUIDE to Durable concrete structures, CEB Bulletin d'information n. 166,
May 1985.
7. Collegio degli lngegneri di Milano: Le opere in calcestruzzo, Durabilita Protezione
Ripristino, CLUP Milano 1986.
8. Giampiero Del Piero (a cura di): Manutenzione, Riparazione e Durabilita delle strutture
in cementa armata, CISM Udine 1987, Collana di lngegneria Strutturale n. 5.
9. Pietro Pedeferr (a cura di}: Corrosione e Protezione delle strutture metalliche e in
cemento armato negli ambienti naturali, CLUP Milano 1987.
10. M. Macori, G. Scaramuzzi, V. Alunno Rossetti, M. Mele, G. Fontanieri: Durabilita delle
opere d'arte stradali, ANAS, Roma febbraio 1988.
11. Filippo Navarra, Le problematiche sui ripristino delle costruzioni monumentali, civili e
industriali, Padova 4 marzo 1988.
12. Franco Massazza: Durabilita delle strutture in cemento armato, II material e. Padova 27
maggio 1988.
13. Enzo Siviero: Durabilita delle strutture in cemento arm ato -II progetto, Pad ova 27 maggio
1988.
14. Peter Schiess! (a cura di): Corrosion of steel in concrete, Report of the Technical
Committe 60 CSC RILEM, Chapman and Hall 1988.
15. Giampiero Tognon: lndirizzi tecnologici per Ia realizzazione di calcestruzzi e strutture
durevoli, Cluva Editrice, Venezia 1989.
16. Elio Giangreco, Edoardo Cosenza: Durabilita dei materiali e delle strutture. Corso sui
consolidamento statico delle strutture in c.a. e c.a.p., Treviso 6-10 marzo 1989.
17. Durable Concrete Structures, CEB Design Guide, CEB Bulletin d'information, n. 1B2
June 1989.
18. Giandomenico Toniolo (a curadi): Propostadel Gruppo di Lavoro CTE al Gruppo Italiano
CEB sulla durabilita delle strutture in c.a. nell'ambito del CEB-FIP Model Code 1990,
Milano 1989.
Durability 159

19. Commission of the European Communities: Industrial Processes Building and Civil
Engineering. Eurocode n. 2, Design of Concrete Structures. Part. 1: General Rules and
Rules for Buildings, October 1989.
20. Emanuele Filiberto Radogna: Seminario ANAS, Calcestruzzi e malte ad alta durabilita,
Roma 30 ottobre 1989.
21, Antonio Migliacci: Seminario ANAS, Milano 13 novembre 1989.
22. Medardo Macori, Durabilita delle opere d'arte stradali: aspetti teorici, Padova 13 febbraio
1990.
23. CEB-FIP Model Code 1990, First Draft. CEB Bulletin d'information, n. 195-196, March
1990.
24. [Link].: Corrosione delle armature del calcestruzzo, INARCOS lngegneri Architetti
Costruttori, Bologna n. 508, aprile 1990.
25. G. Menditto, R. Capozucca, N.N. Cerri, L. Marini: Protective treatment of steel bars in
reinforced concrete structures. Methods and experimental results. Steel '90, Advanced
Technology Protective Treatment and Aesthetics, Genova 7-11 maggio 1990.
26. Luca Sanpaolesi: Protection of concrete reinforcements, the state of the art- Steel '90
Advanced Technology Protective Treatment and Aesthetics, Genova 7-11 maggio 1990.
27. Enzo Siviero: Durability of reinforced concrete structures and use of stainless steel, Steel
'90 Advanced Technology Protective Treatment and Aesthetics, Genova 7-11 maggio
1990.
28. Federation lnternationale de Ia Precontrainte. Proceedings XI Congfess, FIP '90,
Hamburg, June 4-9 1990.
29. De Luisa A., Cimolino M.: Analisi tecnico-economica delle opere d'arte stradali, esame
di una direttrice tipo. Tesi di laurea IUAV, luglio 1990.
 CRACKING DUE TO THERMAL EFFECTS
ON BRIDGES

H. Falkner
Technical University of Braunschweig, Braunschweig, Germany

The thermal effects on bridges, with regard to the crack development and
imposed deformations can, if not treated carefully by either computational
and/or constructional means, result in considerable damage to the
structure. Some bridges constructed in the last two or three decades show
examples of this undesired effect, but nevertheless the few cases published
are only the peak of an iceberg of numerous cases of damage, sometimes in
connection with other effects, which can be linked to the thermal influence
on bridges. See Fig. 1, which shows the damaged web of the Jagst Bridge in
Germany.

= ~// '"
"' -~. . . . . . . . . .:rt;:: ·~
:---23.0 m , 32.2 m 1

.... ...... r,~ =~

r:~J . <e~VLI~ ~:::::~

Elevation of the Jagst Bridge

Crack pattern on the damaged web

1 - - - - - - - - - - - - - - 23.0 m -------------~

u~'H''¥*"'''''[Link]!~'"2'''m2*'''??Z'"'':11
- ~

Fig. 1: Crack pattern of the damaged Jagst Bridge

 162 H. Falkner

A typical example for the crack development in prestressed concrete bridges

is the cracking in the internal span(s} of three or more span bridge, or
the crack development near the internal support of a two span bridge. In
transverse direction the cracking in a box girder bridge due to thermal
effects normally takes place in the web or the deck, especially when thin
members are monolithically connected with thick construction members, see
Fig. 2.

!\web

'
AT I
AS -+-~
A

SECTION A-A

CRACKS WITHIN THIN BOTTOM SLAB HORIZONTAL CRAO<S IN WEBS

Fig. 2: Typical crack development in a box-girder bridge

Not only the temperature effects on the internal stresses of the whole
structure have to be regarded. The total elongation of the structure caused
by a constant temperature rise can cause the same damaging effects, e. g.
roller bearings rolling from their base plates or additional imposed
deformations by insufficiently dimensioned expansion joints.

In order to study the therma 1 effects on bridges, one has to define the
many parameters affecting the temperature distribution within the[ cross
section, such as:
 Cracking due to thennal effects on bridges 163

- air temperature
- solar radiation
re-radiation
- convection
Material
density Properties
conductivity

~"" -"" ~E
- specific heat
- time
- wind speed. OUTSIDE TEMPERATURE

From this enumeration it becomes clear, that it is hardy possible to take

all these factors specifically into account in a bridge design. The wind
speed or the daily cycle of shade/air temperatures changes the temperature
distribution within the cross-section, but for a bridge design it is
essential to find a description for the maximum expected temperature
gradient and the change of the average temperature with respect to the
bridge temperature at the time of erection.

Different countries have developed more or less sophisticated approaches to

this problem. These vary from a simple temperature gradient over the depth
of the cross-section and a fixed value of the average temperature increase
or decrease to a more sophisticated assessment of the temperature
differential taking the depth of the cross-section and the thickness of the
bridge deck, the blacktop-cover and even the colour of the bridge into
account. The form of this temperature differential is either described by
some form of a parabula or by polygonal lines. In some contries even the
in- and decrease of the average temperature to be applied in the design can
depend on the actual location of the structure. A comparison between
different temperature differentials according to various sources is given
in Fig. 3.
 164 H. Falkner

TEMPERATURE - °C SURFACING

Priestley Emerson BS 5400,1972

(~W ZEALAt>ll I ISTEEL1 CONCRETE AND
COMPOSITE CONCRETE l

1-5 °C 2-5°C

Fig. 3: Comparison of different temperature differentials for a 1500 mm

depth of section according to various sources

These non-linear temperature differentials have been developed on the basis

of extensive temperature measurements in different countries, and, if
compared with each other, show more or less the same shape and temperature
values. But one has to bear in mind, tha~ the application of these non-
linear temperature differentials are in contrast with the normal design
assumption that the section remains plane after deformation. On account of
this then only a part of the temperature field can be responsible for the
sections deformation and it must be linear {see Fig. 4). The difference
between the strains which could be caused by neglecting the influence of
the sections fibres upon one another and the strains of a plane section due
to temperature would have to be taken up by self-equilibrating stresses (
see Fig. 4.). Therefore it is customary to determine a plane temperature
field from these temperatur differentials for the computation of the
restraining forces, whereas the self-equilibrating stresses will not
produce any forces and moments within the cross-section.
 Cracking due to thennal effects on bridges 165

To

0 G
TE!yl
Tm +8
:r(y)

Tu T
TEMPERATURE PLANE TEMP. NONE-LINEAR
FIELD T (y) FIELD TEMP. - FIELD

STRESSES DUE TO saF- EQIJILI-

BORDER CONDITIONS BRATING STRESSES

Fig. 4: Sectional stresses due to temperature; section remains plane

In this context the influence of the heat of hydration has to be mentioned.

If during the hydration process the warm concrete is cooled down at the
surface, it can lead to tensile stresses which in most cases exceed the
tensile strength of the young concrete. In order-to avoid these effects the
contro 1 of the heat of hydration by chos i ng a 1ow heat cement, proper
detailing of the construction members and a curing concept is essential. If
these precautions are taken, it is normally possible to reach a balance
between the tensile stresses in the building member and the development of
the tensile strength of the concrete in order to avoid this cause of damage
(see Fig. 5).
 166 H. Falkner

°C temperature of concrete
60
40 ~~_--...._......-P-....._-temperature by G =0
I

20~----~:--~~----~
I
I
I time
G tension ( N/mm2)
4+-----~--~~~~~

2~---;~~~--~
o~~--~--~~~~--~--
2 hardening

G compression ( N/mm2)

Fig. 5: Temperature and stress development in a cross-section during the

hydration process

The result of a structural analysis will be as close to reality as one

succeeds in taking the actual structural behaviour into account. If a
simple linear analysis of the structural behaviour is carried out, an
additional increase of the temperature differential will result in a
proportional increase of the structural response. If the effect of creep is
taken into account, one can already assume a slight decrease of the
structural response, but some investigations show that the assumption of a
reduction of approximately 10% seems to be realistic. The most realistic
results will be obtained if the non-linear behaviour of a reinforced or
prestressed concrete structure is taken into account. Here, especially if
the ultimate limit state of the structure is approached, an increase in the
temperature differential will result only in a minor increase of the
structural response, see Fig. 6.
 Cracking due to thennal effects on bridges 167

R
R R
I
Ru_ RuL -----7!I I
I

I' RUL -----;f

I I II
I I I I
RSL I
RsL ---~ I
Rue II I
II I
Rsc RUNL
I I I
Rsrt.

I AT
·I~
MT
T
I· AT ·IAATI- T
I· AT ·I~
AAT
T

LINEAR BEHAVIOUR LINEAR BEHAVIOUR NON LINEAR BEHAVIOUR

INCL. CREEP

Fig. 6: Safety considerations for an additional temperature rise AT with

regard to the structural response

This non-linear behaviour has to be considered within the scope of the

structural safety considerations. Therefore the superposition of restraint-
and load effects for the service limit and ultimate limit state is normally
carried out with different part i a1 safety factors, which, for the service
limit state take account of the mostly still linear behaviour of the
structure and the normally non-linear behaviour under ultimate load
conditions, where, as mentioned alre~dy, an additional temperature gradient
results only in a disproportional increase of the total structural
response, as can be seen in Fig. 7.
 168 H. Falkner

effect by t:.T
effect by Yxt:.T

Fig. 7: Decrease of MAT under load effects above service conditions for a
ductile structure

This loss of stiffness, in a concrete structure due to the development of

cracks until a stabilized crack pattern is reached (see Fig. 8), on one
hand reduces the restraint forces on the other hand appears visible on the
surface of the structure. Therefore the crack development, especially with
regard to the allowable crack width and distance has to be carefully

tl!
li
1,0 ....--------------- r--------,

stabilized
crock pattern

Ec•Esred 1
~----......! Plow d =p

~1·--- crock ·- ~1
formation - -......

Fig. 8: loss of stiffness due to cracking

 Cracking due to thennal effects on bridges 169

controlled by the amount, placement and spacing of reinforcement in the

endangered zones of the structure. This is especially [Link] for
reinforced and partially prestressed structures, where the development of
cracks is already an explicit part of the design assumption . Moreover, it
is not advisible to try to control the crack development in a fully
prestressed structure by increasing the prestressing· force , but it has
proved to be more effective to control the crack development in the
dendangered zones by the placement of carefully designed constructive
reinforcement . This is mainly due to the poor bond characteristics of
grouted prestressing cables and the low ductility of the prestressing
steel. Therefore a structure with a certain amount of reinforcement steel
behaves, as far as the duct i 1i ty and the crack deve 1opment is concerned,
far better than a fully prestressed structure, where the rupture of one
prestressing cable may lead to a progressive failure of the other
prestressing cables and can endanger the whole structure.

The horizontal displacements of the superstructure, due to a constant

temperature rise, are normally defined by the distance from a resting point
or the fixing point of a bridge (see Fig. 9). All these displacements have
VIADUCT

POl NT CJ= REST

SUPPORT CONDITIONS :

'- SLIDIII() -' HINGED '- FIXED -' HINGED '- SLIDING --'

WIDE SPAN RIVER BRIDGE

SLIDING ------' HINGED SLIDING

Fig . 9: Possible statical systems for different types of bridges

 170 H. Falkner

normally to be absorbed at the end of the structure by appropriately

dimensioned expansion joints, and therefore the arrangement of the bearings
and the fixing point has to be carefully considered in order to minimize
these movements. Sometimes, depending on the slenderness of the supporting
columns, slide bearings can be omitted in the middle of a bridge near the
resting point, if the-restraint actions due to imposed deformations on the
columns remain sufficiently small, which is normally not the case for
systems with short, compact piers (see Fig. 9). Even the orientation of a
bridge, i. e. the heating of only one webs depending on the height of the
sun can induce horizontal deflections which have to be accounted for. For
smaller bridge systems these deformations are normally negligible, but
should at least be considered by constructional means for wide-span bridge
systems.

An example for extensive temperature measurements carried out on newly a

erected bridge is the Rhine Bridge Cologne-Deutz (see Fig. 10). Here, over
a period of 2 1/2 years the bridge temperatures, deflections (vertical and
horizontal} and movements of the bearings were measured. Examples of these
measurements are given in Fig. 11 and 12 .

. •·_·-.J ' •"-;'lll

- ·:.~·.' ·~

Fig. 10: View of the Rhine Bridge Cologne-Deutz

 Cracking due to thennal effects on bridges 171

MEASURING PROFILE 1- ABUTEMENT "DEUTZ"

1, 2 31 4, 51
- --- - I
I I 11
•6 !13 ll
I 19
6, 7, a,
CROSS -SECTION

MEAN

- 2

- 4

1980 1981 1982 1983

- 6

Fig. 11: Rhine Bridge Cologne-Deutz: Temperature difference between deck

and bottom plate
20 T [°C] .:
~AVERAGE TEMPERATURE IN MID-SPAN
I !AVERAGE TEMPERATURE NEAR PIER "DEUTZ"
rr--...;;......-::-......::::::;::::~

5
.·6 7·.· 8 9 10 11 Days
2 3 4

2 3 4 5 6 7 8 9 10 11 Days

Fig. 12: Rhine Bridge Cologne-Deutz: Daily cycle of air temperature,

average bridge temperature and movements of the bearing on
abutement "Deutz"
 172 H. Falkner

Another example for a bridge, where, due to the actual shape of the bridge,
the influence of the temperature had to be taken very exp 1i cit 1y into
account is the Ganter Bridge on the Simplon Pass (see Fig. 13).

Fig. 13: The Ganter Bridge on the Simplon Pass road against the sunlight

Furthermore it has to be mentioned, that for very long multiple span bridge
systems the arrangement of expansion joints at both ends, but with fairly
large movements, or the arrangement of additional expansion joints over the
intermediate supports with lower movements, i. e. for prefabricated bridge
systems, can be considered. But as these additional expansion joints always
provide weak points within the structural system, a policy of minimizing
the possible expansion joints should be followed.

The development and introduction of structures with external prestress,

either for new bridges or for the repair of insufficiently designed or
dimensioned older structures, leads to one final point which should be
mentioned. So far the temperature for the prestressing cables can be
assumed to be the same than the temperature of the surrounding concrete.
But with the arrangement of the prestressing cables on the outside of the
webs, between the webs of a T-beam or inside a box-girder cross-section,
temperature differences between the prestressing cables and the cross-
 Cracldng due to !henna! effects on bridges 173

section can occur (see Fig. 14). For these cases one has to consider if
these differences can lead to significant restraint actions and _how they
have to be taken into account in the actual design.

ADDITIONAL CONSIDERATIONS FOR EXTERNAL (OR INTERNAL I

PRESTRESSING CABLES

POLYGONAL ARRANGEMENT WI TH
MULTIPLE DEVIATORS

~-
AT BETWEEN EXTERNAL CABLES A~ CROSS- Ll~AR ARRANGEMENT WITH SINGLE
SECTION D...E TO DIFFERENT HEATING DEVIATION BLOCKS
BEHAVIOUR

Fig. 14: Temperature considerations for bridges with external prestressing

cables
 STRUCTURAL EFFECTS OF TIME DEPENDENT
BEHAVIOUR OF CONCRETE

F. Mola
Politechnic of Milan, Milan, Italy

ABSTRACT

The procedures for the linear viscoelastic analysis of R.C. and P.C.
structures with particular emphasis to bridge-structures are presented.
After a brief discussion of the fundamental properties of the
constitutive linear viscoelastic law, based on the superposition
integral, the two simplified models, namely the classical model and the
Dischinger model are presented and widely discussed, stating their
stress-strain laws of differential type. The numerical algorithms and the
approximate techniques allowing to express in a convenient way for the
applications the integral constitutive law are then introduced and a
particular discussion is devoted to the algebraic approximate procedures.
The homogeneous structures are then examinated and the basic theorems of
linear viscoelasticity are deduced, showing the results deriving from an
exact analysis when loads or imposed deformations or additional
restraints are applied to the structure. As regards the non homogeneous
structures the very important class related to the homogeneous
viscoelastic structures with elastic restraints is studied. With
reference to bridge structures this class is very important as it
includes the cable-stayed bridges or the bridge beams with P.C. or steel-
concrete transverse sections. The problem of the evaluation the state of
stress in cable stayed bridges or in P.C. or steel concrete sections is
approached by means of the unified procedure, stated by the author and
called Reduced Relaxation Function Method, which is here applied in its
direct or inverse form.
176 [Link]

The method is explained in its fundamental aspects and approximate

solutions of particular simplicity are suggested, defining the upper and
lower bounds for the exact solution. In a wide number of actual cases
these bounds are very close so that the approximate solutions can be
recommended for practical purposes. Three numerical examples, related to
cable stayed bridges and to a steel-concrete section are then discussed
showing the marked effects of creep on the long term state of stress of
this kind of bridge structures.

1 . INTRODUCTION

Concrete exhibits time dependent deformations of two kinds: the

first referred as shrinkage deformation is independent from the state of
stress and the second called creep deformation is produced by the applied
sustained stresses. These deformations have generally a not negligible
influence on the behaviour of R.C. and P.C. structures and in many cases
their effects can be very marked so .that they have interested in a large
extent structural engineers, researchers and builders, as it is pointed
out by the very high number of theoretical and experimental works
published in the last three decades.
The shrinkage deformation, more exactly called drying shrinkage,
produces a time-developing volumetric shrinking of the hardened concrete
and is essentially a consequence of the loss of water from the concrete
pores. As it is independent from the applied stress it can be considered
as an imposed isotropic deformation having a prescribed time-law and can
be easily taken into account in the structural analysis.
As regards R.C. and P.C. statically determinate structures composed
by unidimensional elements as beams and columns, shrinkage produces self-
equilibrated stresses in the transverse sections as it is elastically
restrained by the reinforcement, while in redundant structures further
stresses arise as a consequence of the reactions of the redundant
restraints. In order to proceed to a correct structural analysis a
reliable description of the shrinkage deformation, based on experimental
data has to be performed. The results of several experimental works, [1],
[2], [3], [4], show that the final values reached by the shrinkage
deformation are quite large, approximately similar to the elastic ones
produced by the service loads, so that internal or external restraints
can generate high tensile stresses in the concrete and in R.C. structures
the development of cracks which can cause an advanced structural
degrading. In P.C. structures shrinkage reduces the prestressing force
and the structural safety as regards the cracking limit state. When the
reinforcement is not symmetrically disposed in the transverse sections
or the shrinkage deformation depends in a non linear way from the
geometrical coordinates, the restraining stresses produce flexural
deformations which can generate not negligible transverse displacement
as in the case of composite steel-concrete sections or P.C. beams
Time dependent behaviour 177

connected to R.C. slabs. The delayed deformations produced by applied

sustained stresses, referred as creep, have structural effects quite
different from those related to shrinkage as they modify the constitutive
strain -stress law of the material. A reliable evaluation of the creep
deformations can be made only by means of a wide experimental research
from which one can define theoretical models able to correctly predict
the structural behaviour and its evolution in time. Referring to the
structural service phase, where the state of stress is not too high,
about CJc ~ 0,4 fc, the experimental results, [5], [6], [7], [8], can be
fitted by means of the theory of linear viscoelasticity, based on the
principle of superposition and show that the final values of the creep
deformations are of about (2+5) times larger than the elastic ones.
According to linear theory of viscoelasticity the general formulation of
the uniaxial strain-strees law of concrete drives to an integral form
which is very complex from the computational point of view, so that
simplified models or approximate formulations of the integral form are
of great interest, in order to obtain approximate results sufficiently
refined for practical purposes without performing cumbersome
calculations. The creep structural effects are more important when the
structure exhibits marked inhomogeneity or the sequence of the applied
stresses or imposed deformations is complex. Generally creep produces
variations of the state of stress in non homogeneous redundant
structures, stress redistributions in the transverse sections of R.C. or
P.C. members, reduction of the state of stress produced by imposed
deformations, particularly the prestressing force, the ar1s1ng of
reactions in the restraints imposed after the application of the
sustained actions. From these facts we can conclude that the creep
effects are particularly important in the structures built in several
phases, subjected to actions applied at different times and to the
imposition of additional restrain~ during the construction process.
These aspects are typical of the present bridge structures for
which the building procedure is of great importance in order to obtain
efficient and economical constructions. The building procedures of modern
bridge structures require the presence of parts made of concretes of
different rheological properties, as it takes place in the P.C. beams
connected to R.C. slabs or the presence of homogeneous parts subjected
to loads and restraints applied at different times as in the segmental
bridges or the presence of homogeneous viscoelastic parts and elastic
parts as in the cable-stayed bridges.
At the present time a reliable analysis of a bridge structure
requires the evaluation of the effects of the time dependent deformations
of concrete, so that the Codes [9], [10], [11], [12] give the basic
clauses for the time-dependent structural analysis, however, a
comprehensive formulation of the problem and the explanation of the most
convenient algorithms, either exact or approximate is not easily
available in the technical literature, so that in the present chapter,
after a brief introduction to the theory of linear viscoelasticity, the
basic methods for the creep structural analysis will be discussed and a
particular algorithm for the structural analysis of cable-stayed bridges
178 F. Mola

and P.C. or steel-conc rete composite sections, called the Reduced

Relaxation Function Method will be presented with some numerical
applicatio ns.

2 . FUNDAMENTALS OF THE THEORY OF LINEAR VISCOELASTICITY

Under a uniaxial stress a(t), applied from t 0 and in presence of an

imposed deformatio n e(t), the concrete strain-stre ss law, according to
the linear theory of viscoelast icity can be written in the subsequent
form

J do ( t
t

e ( t) = 1) J ( t, t 1) + i ( t) (1)
0

where J(t,t') is the creep function given by the expression

J(t, t 1 ) = - -1-,- (1 + cp (t, t 1)) (2)
E( t )

The creep function represents the strain produced by a unit stress

applied from t' where E(t'), ~(t,t') are respective ly the elastic modulus
and the creep coefficien t. The time-develo pment of the creep function is
reported in fig. 1 and the following mathematic al conditions hold:
1
J(t 11 t 1) I cp ( tf 1 t 1) : 0
E( t 1 )

iJJ ( tl t 1) ::!: 0
at (3)

oJ <tl t'> s 0
at'

The integral in eq. (1) is the mathematic al formulatio n of Me Henry

J(t,t')

J(tlt )
0

t t' t t
0

Fig.l Creep function

Time dependent behaviour 179

Principle of Superposition, [13], [14], [15], and has to be considered

as Stieltjes integral.
Denoting by ~ai(ti) the instantaneous stress increments arising at the
times ti, eq. (1), in its explicit form becomes

f
t
e(t) = a(t 0 ) J(t, t ) + ~ Aai(ti)J(t,tJ + da(t 1 ) J(t, t 1) + e(t) (4)
to

represented in the particular case of step varying stress and e (t)= 0

in fig. 2.
Eq. (1) allows to calculate the strain when the stress is known,
while if the strain is known, it becomes a Volterra integral equation in
the unknown stress a(t). In order to solve this equation, denoting by
R (t,t') the solving kernel of eq.(l), given by the solution of the

a Lla 2

6al

a(t l
0

t t t t
0 1 2

t t
0 1
a (t )
0

E(t )
0
Fig.2 Application of the Superposition
Principle for a step stress path
180 [Link]

integral equation:
(5)

the applicatio n of the Principle of superposit ion to eq. (1) gives:

Jd
t
(6)
a ( t) = (e ( t 1 ) -i ( t 1) ) R ( t, t 1 )
0

The function R(t, t 1 ) is called relaxation function and it satisfies to

the conditions
R ( t 1 t 1)
I = 1 = E ( t 1)
J( t 1 , t 1)
aR <t, t'> ~ 0 (7)
at
aR(t, t:'l l!: 0
at'

The time-develo pment of the relaxation function is shown in fig. 3 and

in fig. 4 is graphicall y represented the integral operation expressed by
eq. (5).
E(t I)
R(t,t 1 )

(t,t 1 )

~--~--+---~--------~~~~~~~t
t
0 1 (t,tl)
R(t,t )
0

Fig.3 Relaxation function

The experiment al data show that the analytical expression of the creep
function J(t, t 1 ) can be put in the form of sum or product of two
functions depending only on the variable t or on the argument (t-t 1 1: ) .

According to the product form, adopted in [16], we have:

J(t, t 1) =- 1-,- [1 + f(t 1) g(t-t 1)] (B)

E( t )
Time dependent behaviour 181

J,R J(t,t')

E (t') t'

Fig.4 Geometrical representation

of integral equation (5)

while the sum form, adopted in (17] gives

J ( t, t 1 ) = - 1-,- + fl ( t- t 1) + [g( t) - g( t 1) l (9)

E( t )

Expressions (8), (9) assume that the creep deformation can be obtained
combining the creep behaviour of two limiting models, namely the
classical viscoelastic model and the rate of creep or Dischinger model.
In the classical viscoelastic model we have J(t, t') = J(t - t') while
the Dischinger model assumes J(t, t 1 ) =1 I E(t 1 ) + [q>(t) -q>(t 1)] I E0 •

These two models allow to highly simplify the analysis of viscoelastic

structures as they reduce the integral law (1) to differential laws which
are much more simple and do not require significant computational
efforts.

3. THE CLASSICAL VISCOELASTIC MODEL

From the rheological point of view this model derives from the assembly
of elementary models, [18], [19], [20]. These models, sketched in fig.
5 are the elastic model, also called Hooke model and the viscous fluid
model, named Newton model. The uniaxial constitutive law for the Hooke
model is

€ ( t) ~ (10)
E

while for the Newton model we have

€. (t) =
,
~ (11)

The elastic modulus E and the viscosity coefficient ry are kept constant
182 F. Mola

(J

(J a

(1) (1)
E E

(J a

Fig.S Hooke elastic model

cr
and Newton viscous model
Fig.6 Classical viscoe-
lastic standard model

1n time and the dot represents the derivative with respect the time t
(e.g. E: = de(t) I dt).
Assembling these two elementary models we can build the composite model
of fig. 6 named standard creep model, which is governed by the subsequent
equilibrium and compatibility equations
ae =a
a~l + a~l =a (12)

eH + e~l + e =e
The constitutive laws of the elementary models allow to write

(13)

• (1l a~l
eN =- -
TI (1l

so that combining eqs. (12) and (13) we obtain the following strain-
stress law for the composite model
(14)

where for the parameters •*, E, ~-the subsequent expressions hold

Time dependent behaviour 183

retardation time (15)

Eq. (14) is a first order differential equation with constant

(16)
!p. = final value of the creep coefficient

Eq. (14) is a first order differential equation with constant

coefficients and it is equivalent to the integral law (1). As a
particular case for e
= 0, a= 1 applied in t' from eq. (14) we obtain
the deformation e(t) representing the creep function J(cl(t,t'), which is
the solution of the differential equation
e + ..!.. = 1:.
't* E
(1 + mT•'/
) I 't* (17)

with the initial condition

e ( t 1) = 1
E
(18)

From eqs. (17), (18) we easily obtain

e(t) = Jlcl (t, t 1) = ~ [ 1 + tp.

(
1 -e-.-.-
-lt-t 1J )] (19)

In an analogous way for e = 0, e = 1 applied in t' the stress a(t)

coinciding with the relaxation function is obtained from eq. (14) solving
the differential equation
..!..
't*
= aE + !!..E (1 + m
T•
l/ 't 0 (20)

with the initial condition

E = o ( t 1) (21)

From eqs. (20) and (21) we have:

o(t) =Riel (t, t 1) = _E_

[1 + tp. • e
- (l••->
•"
(t-t'>] (22)
1 + !p.
184 F. Mola

The expressions (19), (22) show that the creep function Jlcl(t,t') and the
relaxation function R(cl(t,t') depend from the argument (t-t'), so that,
as it is pointed out in fig. 7, the Jlc>and Riel curves do not modify their
configuration when t' varies, as they are simply rigidly translated along
the t axis.

(c) (t,t')

1/E

to t1
tR(c)(t,t') E

,-~-
H ~0 1
Fig.7 Creep and relaxat~on unct~ons
. f .

of the Classical standard model

From eq. (19), for the creep coefficient cp<c>(t,t') we derive the
expression:
- ( t- t 1)
q> (c) (
(
t, t 1) = q>. 1 - e •'
)
(23)

and eliminating the exponential in eqs. (23) and (22) we derive:

R (c) (
E - 1 +
t, t') = - + [ !p. ( 1 - lfl (c) ( t, t 'J )(1••->J (24)
1 q>. ·-

so that we can state that the relaxation function R(cJ and the creep
coefficient cp<cJ are connected by a functional relation expressed in a
finite form.
From eqs. (19), (23), (24) we derive the following asymptotic values for
J(c), cp<cl, R(c)
(1+q>.)
lim Jlcl ( t, t 1)
t-- E
v t'
lim
t--
q>(c) (t,t 1) = q>. v t' (25)

lim t, E
t-•
R (c) ( t 1)
( 1 + IJI.)
v t 1
Time dependent behaviour 185

Eqs. (25) allow to proof a basic property of the classical model. Let us
suppose that the model is subjected to a generic stress path, only
satisfying the asymptotic condition
lim o ( t) = o (oo) < oo (26)
e~•

In this case, evaluating the limit for t ~ oo of eq. (l) and remembering
the first of eqs. (25) we obtain

e (oo) = 0 ( t.,l
(l+cp.)
+
(l+cp.)
E J do ( t
t
1) = o (oo) (l+cp.) + e (oo) (27)
E E
t.

From eq. (27) we can observe that for t ~ oo the constitutive law of the
classical model coincides with that of an elastic model with a modified
elastic modulus E' given by the expression
E' = E (28)
(l+cp.)

so that inserting eq. (28) 1n eq. (27) we obtain the following strain-
stress elastic law
e(oo) = o(oo) +e(oo) (29)
E'

4. THE DISCHINGER MODEL

This model also derives from the assembly of elementary models for
which the parameters are supposed varying in time. For the Dischinger
model, sketched in fig. 8 and discussed in [21], [22], [23] the
equilibrium and compatibility equations become

a
Fig.8 Dischinger model
186 F. Mola

oN=o (30)
eH+eN+e=e

and the constitutive laws for the elementary models give

(31)

Combining eqs. (30) and (31) we obtain the following constitutive law for
the Dischinger model
• • = E(at)
e - e +
a
Tiltf (32)

From eq. (32) for e = 0, a = 1 applied in t' we can reach the creep
function J< 0 l(t, t') by solving the differential equation
• 1
e = (33)
Tiltf

with the initial condition

e (t 1l 1-
= -- (34)
E ( t 1)

Integrating eq. (33) we have

e (t) = J(D) (t, t') = __1_ + lp(D) (t) -cp(D) (tl) (35)
E (t 1) E2s

where
tp<Dl (t) = J __
1J ( t'>
1_ dt'. Ea (36)

and E28 is the elastic modulus calculated fort' = 28 days. The choice of
E28 is merely conventional, as we can refer the creep coefficient
expressed by eq. (36) to the actual modulus E(t') or to any other value
of the elastic modulus. However the modulus E28 is universally accepted
as it was for the first time adopted by Dischinger.
From eq. (32), putting € = 0, c = 1 applied in t' we obtain the
relaxation function R< 0 l(t,t') by solving the differential equation
• (
(J t) ( )
+~=0 (37)
E ( t) 11 ( t)
Time dependent behaviour 187

with the initial condition

(38)
o {t 1) = E ( t 1)

From eq. (37), (38) we obtain

-It
I
E(f>
'1 (t)
dt (39)
R ( t) = R (D) { t, t 1) = E( t 1) e <

As a particular case, if we assume a constant elastic modulus E(t') = E28 ,

from eq. (39), remembering eqs. (36), (35) we derive
(40)

so that, as for the classical viscoelastic model we can state a finite

functional relation between the relaxation function and the creep
coefficient. The analysis of the curves J< 0 >(t,t') and R< 0 >(t,t'), reported
in fig. 9 and fig. 10 shows that the creep function curves are parallel

(D)
J (t,t')

1/E(t )
0
t' t
t t
0 1
Fig.9 Creep function of the Dischinger
model
E(t')

Fig.1U Relaxation function of the Dischin-

ger model
188 [Link]

OJ(D) ( t, t 1)
when t' varies as the derivative does not depend from t'
at
while this property is no longer valid for the relaxation curves as it
can be easily argued considering eq. (39).

5. ACTUAL VISCOELASTIC MODELS

The experimental results show that the creep behaviour of concrete

lies at the interior of the two limiting behaviours related to the
classic viscoelastic model and the Dischinger model. This fact is clearly
pointed out by the results that we obtain by applying the stress path
reported in fig. 11, coinciding, according to the principle of
superposition, with the two superimposed stress paths of fig. 12.
Applying eqs. (19) and (35), assuming that for the two limiting models

0
0

t t' t
0
[Link] Loading-unloading process

0
0

;-----~----------.-----------~t
t t'
0
-o
0
Fig.12 Application of Superposition
Principle
e:

t t' t
0

Fig.l3 Deformations according to Clas-

sical Model(C) ,Actual Model(A) and Di-
schinger Model(D)
Time dependent behaviour 189

the final values of the creep coefficient are the same and the elastic
modulus of the classical model coincides with the initial elastic modulus
E(t 0 ) of the Dischinger model, we obtain the strain paths [Link]
respectively by (C) and (D).
As reported in fig. 13, these paths are the same for t 0 < t ~ t' but are
quite different when the stress is removed.
According to eq. (35) for the Dishinger model the creep deformation
developed for t < t' is kept constant while the classical model releases
this deformation and returns to its initial state, when t ~ oo, as stated
by eq. (19).
The actual behaviour is intermediate, so that, as indicated by line (A)
the creep deformation is only partially released. Defining delayed
elasticity the released part of the creep deformation, represented in
fig. 13 by the segments PP' or PP" we can state that for the classical
model the final value of the delayed elasticity coincides with the creep
deformation stored in the interval t 0 ~ t ~ t' so that the model can be
defined as totally reversible. For the Dischinger model the delayed
elasticity is zero and consequently this model is totally irreversible,
while in the actual model the delayed elasticity is present but is
smaller than that of the classical model so that it can be defined as
partially reversible.
It is important to observe that if we as a first approximation assume an
elastic modulus constant in time, comparing eqs. (24) and (40) it is easy
to state that when the creep coefficient and the elastic modulus are the
same, the subsequent inequality holds
R(Dl(t,t 1) ~R(c>(t,t 1 ) (41)

where the equality is valid only for t = t'.

According to eqs. (19) (23) (35) and to fig. 13 we derive
J(D) ( t, t 1 ) ( J(A) ( t, t 1) ( J(C) ( t, tf) (42)

so that, from eq. (5), remembering the second of eqs. (3) and (7) we can
proof that
R(Dl(t,t 1) <R(Al(t,t 1) <R!c>(t,t 1) (43)

Eqs. (42) and (43) represent two basic properties of the linear
viscoelastic theory as they allow, when the difference between the
results derived from the two limiting models is sufficiently small, to
assume these results as the solution of the viscoelastic problem avoiding
the use of the complex integral constitutive law (1) which is substituted
by the simpler differential formulations given by eqs. (14) or (32).
6. THE MODELS CEB MC78 AND ACI COMMITTE 209

. . The Models .cEB MC78, [9], [24], [25] and ACI [16], derive from the
f1tt1ng of exper1mental data, adopting analytical formulations similar
to those expressed by eqs. (8), (9). The CEB MC78 Model assumes for the
creep function the sum form:

with

E(t; 1) =11875 3Jfcm(t 1) (MPa)

f ( t') = f ( oo) [ t' ] 2,145

em em t1 + 47

p a ( t') = 08[1 -(-t-')

' t 1 +47
2,145]

I} ( t-t')
d
=[ t-t'
t-t 1+328
1&~2
[
t ]K,(h.l (45)
IJ.t(t) = t+K1 (ho)

K 1 (h 0 ) = exp [ 5 h~ 2 + ln(6,95. h~' 25 )]

K2 (ho> = exp[ 0. 00144 ho - izo
1 1 - ln (1' 005 h~' 295 ')]
cpd = 0,4
h0,1667)]
(~
cp.t = (4,45- 0,035~) · exp [4,4 · 10- 5 • h 0 -
0 ·~: 7 - ln

In the previous relationships fc 11 (t') is the cylindrical compressive

strengtli at time t' , h

0
= 2 A (em)
p
is the notional thickness of the
member measured by the ratio between the area of the transverse section
and one half of the perimeter directly exposed to the exterior where the
relative humidity is €% and times t, t' are expressed in days.
For the shrinkage deformation e 1 h(t) the following expression are assumed
(46)

with
Time dependent behaviour 191

11,8 • h 0 + 16

K4 (h 0 ) exp[- 0,00257 h 0 + Oh~ 2 + ln (0,22 ·h~·')] (47)

e81 (o,ooo775 ~ 3 - 0,1565 ~ 2 + 11,0325 ~- 303,25) ·1o-s

h0,25 ]
€ 82 = exp [ 0 00174 h - 0 • 32 - ln ( -0- )
I O h0 119

In fig. 14, and in fig. 15, taken from Ref. [24], are reported for
various t' the creep function J and the relaxatio n function R obtained
solving by means of the algorithm exposed in the next section the

DIMENSIONLESS CREEP FUNCTION ~IT, TOl•EC26 FFI =3.00

HO =20.0CM ....
r:l''fL-CfiAr'
I llll
'""
6.5 8.\117

6.0
5 .ee-.
5.5
5.0 5.1!9

ij.5
' .SSIS
ij,O
3.5 l.l36

3.0
2.5 •. s..s
2.0 Z.l ..

1.5
1.0
0.5
0.0
I '
,
10 1
. •• 10 2
... ""10 1
10•
T
OATS
Fig.14 Creep function of CEBMC78 Model (from Ref.24)

integral equation (5). In fig. 16 is reported, from Ref. [24] the time
developm ent function P,(t) and in fig. 17 the function c, 2 (h 0 ) related to
the shrinkage deformat ion.

The ACI Model adopts the product form assuming

1-
J(t-t 1} = - - [1 ( t- t ) ' ..~ ]
+ 1,251p (oo,7) ti-O,llB _ __;c.::__;_....:,.
1 0 6 (48)
E(t 1) 10+(t-t 1 ) 0 •6

E (t 1) ( t' )..!
2
(49)
E 28 4 + 0 85 t 1
I
192 [Link]

DIHENSIONLESS RELAXATION FUNCTION RIT,TDI/EC28

H0 .
FFI •3.00
20 OCH
.....-a...
...... 11711111

1.5

.- 1--'
1.0
v..- 1-"""
,......!--
,/ ........ ~
~ \ ~
o. s ~.<:::~"'
r--
t--..
I'.. "' "
......
1'.
~
"'
1.......... 'I"-
.... :-- .....
•
.... I' .:-.....
• 210

- ..
~ r-. 0.117

:m
o. 0
I
• • 10 1
. n
101 10
B:~M_I
•
T
OATS

Fig.15 Relaxation function of CEBMC78 Model (from Ref.24)

where the coefficient ~(~,7) represents the final value of the creep
coefficient for t' = 7 days dependent from the environmental conditions
and from the notional thickness h0 , as explained in detail in [26]. For
the shrinkage defo~ation the following expression is proposed:

C~z
/
v v 2
1.!& j
v ,.,._ I.IJR1'11
I.
- H111LIL
IIIII
II 1.2
Fig.16 Time-development
1111 -
of shrinkage
according CEBMC78 (from Ref.24) aa
~~ 080
70
~"'
0

hoi rrml
c !i) lXI 200 loOO £00 800 >1600
Fig.l7 Influence of notional thick-
ness on shrinkage (from Ref.241
Time dependent behaviour 193

(50)

Expression (50) is valid for t ~ 7 days and the procedures to calculate

the final value c 5 h(~). depending from the relative humidity and from the
notional thickness are widely discussed in [26].
In Table 1 some values of the nondimensional creep function E28 J ( t, t') and

t-t'
t'=7 days
E28 J(t,t') £ (t)-£ (t')
t '=28 days
E28 J(t,t') £ (t)-E (t')
-r I
t'=90 days
E28 J(t,t') £ (t)-E (t')
sh sh sh sh sh sh
(days) £ (m) £ (m) £ (m)
sh sh sh
10 2.627 0.160 1.721 0.076 1.456 0.021
10 2 4.019 0.587 2.552 0.341 2.104 0.124
3
10 5.078 0.800 3.185 0.523 2.598 0.249
4
10 5.496 0.830 3.434 0.552 2.791 0.276
m 5.686 0.833 3.531 0.556 2.842 0.280
Table 1. Creep function and shrinkage function according to ACI Model.

of the shrinkage function (c 5 h(t) - c 5 h(t'))/c 1 h(~) are reported for various
initial times t'.

7. METHODS OF STRUCTURAL ANALYSIS

The constitutive general laws (1), (6) are the basic relationships
for the analysis of redundant viscoelastic structures.
The structural analysis can be performed using the Forces Method or the
Displacements Method and in both cases it drives to the solution of
systems of Volterra integral equations. For this reason it is of great
importance to state analytical procedures for the solution of the
Volterra integral equations in order to express the constitutive law (1)
in a form directly applicable to the structural analysis.
When general viscoelastic models like the CEB MC78 m,odel or the ACI model
are adopted, eq. (1) has to be solved only by means of numerical
algorithms or, in an approximate way, following feasible simplifying
hypotheses while if we assume the classical or the Dischinger model we
can transform eq. (1) into the differential equations (14) or (32) which
can be integrated without great difficulties.
In order to proceed to the solution of eq. (1) in the unknown stress
a(t), we observe that in the integral appearing in eq. (1) both a and J
are depending from t' and it is possible, as indicated in fig.l8 to
express J(t, t') as a function of a(t'), so that the deformation c(tk)-
€ ( tJc) , comparing in eq. ( 1) and calculated at time t = tK, is measured
by the area of the zone (O ABC) limited by the curve J(t,a). Subdividing
194 [Link]

a(t.) a(t )
l. k
a

t t t t
0 1 k
J(t,t')

t t
0

J(t,a)
J(t
A

10
~--1---------~~~--~~~a
a(t) a(t. )a(t.)a(t)
0 l.-1 l. k
Fig.18 Iterative procedure for the
solution of the Volterra integral
equation

the interval (tK - t 0 ) in an adequate number of sub-intervals (t 1 - ti-1),

i = 1,2, ... , k, the application of the rule of trapezia to the integral
comparing in eq. (1) gives the subsequent expression for the evaluation
of the area (0 ABC)

- <tx>
e <tx> -e ="2l~K Co Ct 1 ) -o ( t1-1 ) > [JCtx, t 1 ) +J(tx, t1-1]+o (t 0 ) J(tx, t 0 ) (5'.L)
•1

Introducing the quantities

(52)
Time dependent behaviour 195

with
i <k
i =k

eq. (51) becomes

a (txl
e ( txl = + i ( txl + e• ( txl (53)
E1 ( tx>

Eq. (53) is an algebraic form relating the stress a(tK), the

corresponding total deformation e(tK), and the deformation e•(tK)
depending from the stress path from time t 0 to time tK_ 1 , so that this
algorithm has to be applied in an iterative way implementing it on an
efficient computer program which allows to calculate for each step time
(tK- tK_ 1 ), the related stress a(tK).
The results obtained from eq. (53) are very good if the width of the sub-
interval (ti - ti_ 1 ) is conveniently chosen. As the creep function is
rapidly increasing when ti is near to t 0 a feasible subdivision of the
interval (tK- t 0 ) can be obtained using the following formula, proposed
in [24]
t - t
X 0 = 1,15 k2:2
tk-1 - to (54)
tx -t 0 = 0, OS days , k =1

According to eq. (54), the accuracy of the results is very high but it
is necessary to perform about 90 steps in order to cover a maximum time -
interval of about 104 days, so that this method, when the structure is
quite complex becomes very expensive from the computational point of
view.
A remarkable simplification of the problem is achieved expressing the
area (0 ABC) in the subsequent way

_ a (t 0 ) a (t) -a (t 0 ) (55)
e(t)-e(t)= E(to> (l+cp(t,t 0 ))+ E(to> (1+X(t,t 0 )cp(t,t 0 ) )

substituting to the area (0' A' B C) the equivalent rectangle ~0' A" B"

C), having height and making the dashed areas of fig. 19

equal.
Eq. (55) can be used only if the function X(t,t 0 ) is known and at this
scope in [27] (28] this function is calculated in an approximate way
supposing that the deformation e(t) - i (t) can be expressed in the
following way:
196 F. Mola

tJ(t,crl

'A A' B'

cr(t l cr(t) a
0

Fig.19 Solution of the Volterra in-

tegral equation, algebraic method

€ ( t) - e (t) = a + bq>
(56)

with a, b arbitrary constants.

In this particular case, from eqs. (1), (6) we immediately obtain

R ( t, t 0 ) )
•1(t) =a R(t, t 0 ) +bE(t 0 ) ( 1- E(to) (57)
o (t 0 ) =a E(t0 )

and substituting eqs. (56), (57) in eq. (55) we derive

1
(58)

Eqs. (56), (57) allows to exactly solve the integral equation (1) in a
wide number of cases, as the related X function is independent from the
values assumed by the two constants a and b. From a general point of
view, the deformations or the stresses existing in the actual structures
cannot be expressed by eqs. (56), (57), but their development in time as
regards the most important practical situations, is not quite different
from that presented in eqs. (56), (57), so the algebraic form (55) can
be adopted in order to achieve a sufficiently approximate solution of the
Volterra integral equation (1). For practical applications eq. (55) can
be put in the following compact form

e ( t) = ~
E 1 ( t)
+ e (t) + €0 ( t) (59)

with
Time dependent behaviour 197

E(t 0 )
E 1 ( t)
1+XfP
(60)
= a (t 0) !<1-xl
€0 ( t)
E( t 0) 1 +xfP

so that we obtain an algebraic form which can be directly applied at time

t without performing intermediate steps. The method based on eq. (59) is
called Age Adjusted Effective Modulus Method (AAEMM) and can be further
on simplified, remembering the following inequality, which has been
proven in [29]

..!.
2
[1 + oE ( t) I
at t• to
I oR ( t, tol
at It• to] s X s 1
(61)

If we take the maximum value of x. namely X 1, eq. (59) becomes

e. ( t) __<!j_tl_ + i ( t)
E* ( t)
(62)

Fig.20a X(t,t') and relaxation function ac-

cording to ACI Model

.3+---~--------+--------+--------4
5 10 20 30 40
t (days)
0
Fig.20b variation of X with the initial ti-
time t ,according to ACI model for mass
0
concrete
198 [Link]

and we obtain in this particular case the so called Effective Modulus

Method, by which, as pointed out in fig. 19 the area (O'A'BC) is
approximated by means of the external rectangle (O'A'B'C). A better
approximation can be achieved observing the diagrams of fig. 20a, where,
if we consider times t sufficiently greater than t 0 and times t' not too
much elevated, the constant value X = 0,8 can be conveniently chosen,
avoiding the evaluation of the relaxation function and consequently the
solution of the Volterra integral equation (5). It is interesting to
observe that in fig. 20a the initial value of X for t ~ t 0 is always 0,5.
This is a consequence of a property of the ACI Model for which

aR(t,
at to> 1 - -ao
t• e. - ' as we can see from fig. 20a. Fot this the second
term in the brackets of eq. (61) vanishes and the minimum value of x
becomes 0,5 independently from t 0 • If we assume a creep model for which
aR ( t, to>
at
It•t. 1s
. f. . . . .
1n1te, as 1t 1s the case of the ACI Model for mass1ve
concrete, [16], the minimum value of X is lesser than 0,5 as indicated
in fig. 20b. Values of X(t 0 , t 0 ) near to 0,5 require times of application
of the loads greater than 40+50 days.

8. LINEAR CREEP STRUCTURAL ANALYSIS OF R.C. AND P.C. STRUCTURES

Strictly speaking, R.C. and P.C. structures cannot be considered

rheologically homogeneous as they are made of a viscoelastic material and
of an elastic material. However, when we evaluate the internal actions
in redundant structures, the determination of the reactions of the
. hypers-tatic restraints is quite independent from the presence of the
reinforcement as its amount is very small compared to the volume of
concrete. For this reason, as a first approximation, we can formulate the
hypothesis that as regards the structural analysis the R. C. or P. C.
structures can be considered as homogeneous.
This hypothesis is no longer valid in evaluating the state of stress in
the transverse sections as in this case the reinforcement has a marked
influence in reducing the creep deformations of concrete, so that the
sectional analysis of R.C. and P.C. structures has to be performed taking
into account their non homogeneity. In the same way the structural non
homogeneity has to be considered when the structure consists of
viscoelastic homogeneous parts and elastic parts as it is the case of the
modern cable-stayed bridges.
From the structural analysis point of view the transverse sections of
R.C. or P.C. structures and the cable-stayed bridges can be considered
as homogeneous structures with elastic restraints, represented in the
first case by the reinforcing bars and in the second by the stays.
According to these considerations from a general point of view we can
individualize two basic types of structures, namely the homogeneous
structures and the homogeneous structures with elastic restraints. For
Time dependent behaviour 199

the homogeneous structures it is possible to state fundamental theorems

which allow to perform the structural analysis by using the results
obtained from the elastic analysis at the initial time while for the
homogeneous structures with elastic restraints it is convenient to
proceed on the basis of a general analytical algorithm, named Reduced
Relaxation Function Method [30], [31] [32], in its direct or inverse
form, allowing to study complex structures by means of a set of
independent integral equations which can be solved adopting one of the
methods previously explained.

8.1 Homogeneous Structures

Let us consider a homogeneous structure, subjected to loads and

imposed deformations. Denoting respectively by o~>, e~>, oC:1 , eC:1
the
stresses and the deformations produced by the loads (p) and the imposed
deformations (g), calculated considering a pure elastic structural
behaviour with a reference modulus E0 , let us assume that in the
viscoelastic structure the stresses a and the deformations e can be
expressed as following
(p) (g) R ( t, t 0 )
O=Oe +Oe
Eo (63)

The stresses expressed by the first of eqs. (63) are equilibrated and the
strains deriving from the second of eqs. (63) satisfy the compatibility
conditions as they are linear combinations of elastic solutions and
furthermore, introducing the first of eqs. (63) in eq. (1) and
remembering eq. (5) we obtain the second of eqs. (63) so that these
equations satisfy the constitutive law of the material. As a consequence
of the uniqueness of the solution of the~linear viscoelastic problem,
eqs. (63) represent the required solution and clearly show that the
knowledge of the elastic solutions is sufficient for the calculation of
the strain and stresses in the homogeneous viscoelastic structure. Eqs.
(63) express in a compact form the two fundamental theorems of linear
viscoelasticity.
Let now consider a homogeneous structure subjected to constant loads
applied in t 0 and impose to it an additional restraint at time t 0 • ~ t 0 •

Denoting by the elastic deformation evaluated in the direction of

the redundant action x 1 in the additional restraint, produced by the
external loads and assuming elastic modulus E(t 0 ) , according to second of
eqs. (63) with e!P) = e= 0 , the restrained displacement for t ~ t 0 • is

(64)
200 [Link]

and indicating by u1• the elastic displacement produced by X1 = 1, for the

unknown reaction t..X1 (t) in the additional restraint we can write the
subsequent integral equation expressing that the state of displacement
is compatible with the additional restraint

I
t
d(AX1 (t 1))U19 E(t0 )J(t,t 1) =- Au(pl =- U 8(pl (1p(t,t 0 )-tp(t~,t 0 )) (65)
t;

remembering that
u (J>)
AXle=- - • - (66)
ule

represents the elastic reaction in the additional restraint if the

restraint were present before the application of loads, from eq. (65),
applying the Superposition principle we have:

AX (t) =AX It iJ~p(t',to> R(t,t') dt' (67)

1 le iJt' E( to)
t:
as a particular case, if we have t 0 ·=
t 0 +, as it is the case when the
additional restraint is applied immediately after the loads, and suppose
that

A.X (t) • AX
1 le
(1- R(t,to))
E (to)
(68)

introducing eq. (68) in eq. (65) and remembering eq. (1) we obtain
(69)

coinciding with eq. ( 65) when t~ = t; .

Eq. (68) is the solution of the problem, so that, indicating by a.P the
elastic stresses produced by loads in the structure without the
additional restraint and by a. 1 the elastic stress produced by X1 = 1, we
can write:
a (J>l ( t) = ct!1
(70)
ofJ>J ( t) = a<:1 + A.X1 ( t) a 91

Eqs. (67), (68), (69), (70) are ver.y important as they represent the
basic solutions of the problem related to the variation of the structural
scheme after the application of loads. In this case also the solution of
the elastic problem calculated assuming that the additional restraint is
applied before the application of loads is sufficient in order to
Time dependent behaviour 201

evaluate the state of stress and deformation of the structure at any

time.

8. 2 Homogeneous elastic structures with elastic restraints, cable stayed

bridges

With reference to fig. 21 let us indicate by !(t) the vector of the

cable forces of a cable stayed bridge subjected to sustained loads (g)
and to geometrical actions produced by the prestressing of the cables,
(P).

X (t) X (t)
i n

Fig.21. Cable-stayed bridge

Indicating by tc>(t 0 ) , !,(s) the deformability matrices of the viscoelastic

and of the elastic part and by ~(c) the vector of the elastic
displacements of the connection points between the cables and the
structures having the direction of the redundants, the compatibility
equations according to eqs. (63) and the superposition principle become,

J <:(c) ( C
t
0 ) E(c) ( C0 ) J ( C, t 1 ) + F(s)) dX ( t 1)
0 (71)

at the initial time t 0 we have

_ (F<cl ( t ) + F(sl) -l.a_ (c) + X (P) (72)
• o • 0 _o

so that, indicating by
(73)

the initial elastic solution connected to external loads, eq. (72) gives
202 F. Mola

(74)

Introducing the coupling matrix

(75)
D
. (to)
. = (F<c> .
+ F<s>) -1 •
.
F<c> (to)

from eq. (71) we have

J (Jf' + J?
t
E(c) (t 0 ) J(t, t 1)) d~(t 1 ) = 3_ 0 (g) E(c) (t 0 ) J(t, t 0) +3_ 0 (P) (76)
0

where ~· is the complementary matrix of ~-

Eq. (76) represents a system of Volterra integral equations which can be

conveniently solved in the subsequent way.
Indicating by a
the diagonal spectral matrix (or eigenvalues matrix) of
~and by! the corresponding modal matrix (or eigenvectors matrix),
where the eigenvalues "'; and the eigenvectors K; are given by the
expressions
det I(A)I- Dl = 0
(77)
I(A)1I-DIK1=0

with l. unit matrix, as it is well known from the algebra of linear

transformations, the fundamental relationship between matrices ~. ~ a.
holds:
0 = I t1 D K (78)

Introducing the new unknown vector Y given by the equation:

X=KY (79)

the inserting of eq. (79) in eq. (76), remembering eq. (78) drives to the
subsequent set of independent Volterra integral equations
t
J[(l-(A)~)+(A) 1 E(cl(t 0 )J(t,t 1 ))dY1 (t 1 ) = Y~fE<cl(t 0 )J(t,t 0 ) + Y01 1P> (80)
0

so that, defining the varied creep functions J;*(t,t') by the equalities:

Ji(t,t 1) = E (c / (to) [(1-(A)i) + (A) 1 E(c) (t 0 )J(t,t 1)) (81)
Time dependent behaviour 203

eq. (80) takes its final form

. (82)

Introducing the reduced relaxation functions R;.(t,t'), as the resolvent

kernels of the Volterra integral equations (82), satisfying the integral
equations

aR; t'>
f
t

0
<f,
af ·Ji<t,f> dt=1 (83)

the solution of eqs. (82), remembering eq. (81) becomes

yi = y~f' [ 1 + (w. _ 1 ) Ri (t, t 0 )] + y<!'> Ri (t, t 0 ) (84)

W1 ~ E(c) ( t 0)
0~ E(c) ( t0)

and remembering eq. (79), for the unknown vector X we obtain the matrix
form:
R• (t, t 0 ) ] R• (t, t 0 )
X = K0- 1 [ I + ( 0 - I) • JC 1 X <g> + K " JC 1 X (P)
(85)
- = = = = = E(c) ( t) • _o = Eo • _o

wh~re B•(t, t 0 ) is the diagonal matrix of the reduced relaxation functions

R; (t, t 0 ) .

Eq. (85) can be put in a different form introducing the vector ~~~~lg
which takes place when the elastic restraints are perfectly rigid i.e.
for !(s) = 0. In this case from eq. (73) we derive:
(86)

and comparing eqs. (86) and (73) we obtain

X <g> -- rr 1 x<g> (87)
- o,rig . - o

Introducing eq. (87) in eq. (85), remembering eq. (78) we finally obtain

- •
R" (t, t 0 )
X= [ I-K •
• E(c) (to)
r
•
1
R" (t, t
+K •
= E(c) (to)
0 )
Or
= =
1
l-
R• (t, t 0 )
x~~~ig+K-"--,--,.---
• E(c) (to)
r 1 x<:> (88)
204 F. Mola

Eqs. (75) and (88) allow to state two basic properties affecting the
eigenvalues w. and the vector K· As regards the eigenvalues, let us
consider the ~lastic energy u<cJ stored in the viscoelastic part, and the
energy u<•l stored in the elastic part when we apply to the structure the
vector of forces %
1 in the cables coinciding with the i-th eigenvector
:K; 0

From the fundamental clauses of elastic structural analysis we have:

u<cl
(89)
u<sl ..!_ Kl F(s) K_ i.
2 - •

so that for the whole structure derives

u<stl "' u<cl + u<sl "' ; ~I (F<cl + F(s)) ~i (90)

From the second of eqs. (77) and remembering eq. (75) we obtain:
!il i. ~i. "' :> ~i. "' <:(c) + F(s)) -1 • F(c) • ~i (91)

and consequently
F(cl ~i (92)

Comparing eqs. (92) (90) (89) we derive:

!iJ 0 "' u<cl (93)
' (u<cl + u<sl)

so that all the eigenvalues wi, representing the ratio between the
elastic energy stored in the viscoelastic part and that of the whole
structure, are real and satisfy the inequality:
0 s !iJ i s 1 (94)
The particular case w; = 0 takes place when the elastic part is
infinitely deformable, or in other words when the elastic restraints are
practically negligible, while the upper value W; = 1 corresponds to
elastic restraints highly stiff, practically rigid, for which F<•l = 0.
The inequality (94) allows to calculate in advance the reduced reiaxation
functions and to prepare tables considering W; as a parameter, so that an
easy evaluation of the terms of matrix~· can be made. In fig. 22 it has
been reported one of those tables, calculated using the numerical
algorithm of eq. (53) and adopting the CEB MC78 model.
In order to proof the property related to vector K let us suppose to
Time dependent behaviour 205

prestress the cables in such a way that the total force existing in them,
sum of that produced by external loads (~(g)) and that due to
prestressing (~ (Pl) is made equal to the force ~b~~ig connected to the
external loads if the cables were rigid.
In this case we have
x<g>
_o,rig
(95)

h 0 ·20cm R.H.-70/. t 0 -7deys

Ri<t.t 0 l/E 028 w1=0+[Link] (n=l,2,3, •. ,10)

1.0 w1 =o
I
0.8
::-- :-
[Link] r-- t--1....
~ r--... r--r-
--
r-..._
0.6
~~ :--
?::
i-
1-
Wl=~
:::::r-
0.4 ~~ f.....

0.2
~ [Link]~ ::::~
~:::::

0.0
10 100 1000 10000
t [deysl
t-t0 1doysl 7 ,, 30 60 120 3e 109:1 3300 10000

w,-0.001 .86562 .86:535 .86521 .86503 .116480 .1164:53 .116407 .116373 .116J!56 .!!IIJ49
....o. 0:10 .8411J2 .436611 .43012 .82166 .810110 .791109 .n74:1 .76267 .7':1521 .[Link]
w,-0.100 .83331 .80918 .79671 .7807:1 .76042 .7J691 .6111149 .673JJ .660JJ .~0
..,.a. 1:10 .81788 .74328 .[Link] .74288 .71429 .611163 .63067 •:19:18:1 .:171143 .:1721:1
w,-0.200 .80302 .7:1886 .73634 .70776 .67198 .63157 .56977 .:128:11 .:101164 .500119
w,-0.250 .78869 .73:1110 .70900 .67:114 .6JJ09 .:18616 .:11:177 .46982 .4480:1 .43960
w,-0.300 .n487 .71401 .6833:1 .64479 .:19726 .:14488 .46779 .41856 .39:161 .311674
w,-0.350 .76152 .69337 .65924 .61650 .:16426 .:10729 • 42:107 .37373 .3:1012 .34106
w,.0.400 .74862 .67361 .63656 .:19009 .:1337:1 .47300 .38696 .JJ440 .310:16 .30148
w,-0.4:10 .7361:1 .[Link] .61518 .:16:141 .:10:1:11 .44166 .3:1290 .29943 .27612 .26711
w,-0.500 .72409 .63761 .:19:101 .54230 .4793:1 .41296 .32240 .26936 .24603 .23720
w,-0.550 .71241 .62043 .:17:19:1 .52064 .[Link] .366611 .29:104 .242:11 :21970 .211 II
w,-0.600 .70111 .60486 .5:1793 .:10030 .43249 .362:12 .27046 .21676 .19660 .18830
w,-0.6:10 .69017 .:16963 .54086 . 48118 .41147 .34031 .24832 .19771 .17631 .161132
..,.o .700 .679:15 .57511 .52468 .46320 .39188 .31986 .22837 .17904 . 1:1644 .1:1078
w,-0 .750 .66926 .:16124 .50932 . 40625 .37360 .30099 .21035 .16243 .14267 .13535
..,.o .800 .65928 .54798 .49473 .43026 .35652 . 28357 .19405 .14764 ; 12873 .12176
w,-0.850 .64959 .53530 ,48085 .41517 .340:13 .26745 .17929 .13443 .11639 . 1097:1
..,.0.900 .64018 .52316 . 46764 40090 32556 . 25253 .16569 .12263 .10543 .09913
..,.o.950 .63104 .51152 .45505 .J67J9 .31152 .23670 .15372 .11207 . 09:169 .08971
w,-1.000 .62216 .50036 .44306 .37460 .29433 .22585 .14264 .10259 .06702 .01113:1

Fig.22 Reduced relaxation functions according to CEBMC78

206 F. Mola

So that the forces imposed by prestress, according to eq. (87) are:

( I - D) X (g)
_o.r1g (96)

Introducing eq. (96) in eq. (88) and remembering eq. (78) we obtain

~ ( t) = ~~~~1g (97)

Eq. (97) has a great importance in practical design of cable-stayed

bridges, as it states that if by means of an initial prestressing of the
cable we introduce the forces corresponding to the case of rigid
restraints, these forces cannot vary in time and the state of stress of
the structure is not influenced by creep.

8.3 P.C., R.C. AND COMPOSITE STEEL-CONCRETE SECTIONS

When performing the sectional analysis it is possible to proceed in the

same way followed for the structural analysis, applying the Reduced
Relaxation Function Method in its inverse form. Referring to the section
of fig. 23, indicating by~ the vector of the sectional deformabilities

'¥ (t)
1

Fig.23 P.C. section and strain

diagram

and by p T = 11 - yj the vector of the generalized coordinates, assuming

the Bernoulli-Navier hypothesis we have:
e<c> ,. pT. t
e<•> = f r.! (98)

where e 0 !•P> is the imposed elastic deformation due to the prestressing of

Time dependent behaviour 2f.r7

steel. From eq. (6) and considering that the steel is supposed elastic,
we have:

"'J d[~T!
t
a<cl (t 1) - € 8 h(t 1l] R(t, t 1)
0
(99)
a<sl = E(s) pT Ill
a<sp) = E(sp) p TIll + E(sp) e<:P>

The sectional equilibrium requires that

f
A(c)
pa<c> cJA(C) + ~ pa~s>Ajsl + ')'
j
eajsp) Aj"Pl "'0 (100)

where OT = iN Mi is the vector of the internal actions.

Combining eqs. (99) and (100) and introducing the stiffness matrices A(c)
and ~*!•> of the creeping part and of the reinforcement, according to the
expressions

(101)

and the vector ~h' Q(P) of the internal actions equivalent to shrinkage
and prestressing

aesh ( t') R ( t, t') dA <c> dt' (102)

P at'

(103)

eq. (100) becomes:

f
t
(B(c) R(t, t') + B*(B)) d11J (t 1) "'0 + Osh + o<Pl (104)
E(cl(t 0 ) - - -
0

At the initial time we have:

.
(B(C) + B*(B))
. -
Ill (to) "' 0
- -
+ o<Pl (105)

so that, introducing the interaction matrix

208 [Link]

(106)

eq. (104) can be written in the subsequent form

[(~ ;cS;·<~~~ + ;- ~) cf!<t'> =!<to>+ 111esh<t> (107)

r
where
!uh(t) =we)+ ~·(s) ~sh (t) (108)

is the solution of the problem in the elastic domain connected to the

shrinkage deformation. In order to solve the system of Volterra integral
equations (107) we can proceed by means of the Reduced Relaxation
Function Method in its inverse form, by means of the linear
transformation
(109)

(110)

where K*, g* are the modal and spectral matrices of A·

Introducing the varied relaxation functions

Ri (t,t') = [(o)•i R(t,t')

E(c) (to) + 1-(o)•]. E<cl ( to)
1
(111)

and their resolvents J;(t,t') solutions of the independent Volterra

integral equations

the solution of eq. (107), applying the principle of superposition and

introducing the diagonal matrix J(t,t 1) immediately becomes

f ~ (t, t
t

1jl = rJ ( t, t0) E(c) ( t 0 ) ~1 * ! (t 0) + r E(c) ( t 0)

1) ~- 1 cf!esh ( t 1 ) (113)
0

A particular simplification of eq. (113) is obtained if one supposes that

the shrinkage and creep are affine in time. In this case we have
Time dependent behaviour 209

(114)

so that, remembering eq. (102), we obtain

( 115)

and from eq. (108)

_ R (t, t~) (116)
V_esh ( t) = tesh (oo) (1 ( )
E c (to)

where
1i'esh(oo) = (B(c)
=-
+ B*(B))-1 E(c) (to)
~
I
A(e)
£ dA(c). e.h(oo) I tp (oo) (117)

From eq. (113), remembering eq. (116) and (112), we reach finally the
expression

(118)

which is formally [Link] to eq. (85).

Eqs. (113) or (118) give the sectional deformability vector, so that the
second and third of eqs. (9 9) allow to calculate the stresses in the
steel, while in order to obtain the concrete stresses, indicating by A~
the variation of the deformation vector, the sectional equilibrium gives:
(119)

so that from eq. (101)

- -
![Link] =- B*(S) AljJ
- (120)

In the viscoelastic homogeneous part, according to the fundamental

theorem expressed by eqs. (63) we have:
(121)

(122)

Combining eqs. (122), (120) remembering eq. (121) we finally obtain

= _ g!c)
tJ.a!cl
-. -
(t 0 ) pT B(c)-1 B*(s) [Link]
- (123)
210 [Link]

The reduced relaxation functions R* 1 or the functions J 1 (t, t 0 ) are

the basic elements to calculate the solutions of the preceding problems,
deriving from the application of the Reduced Relaxation Function Method
in its direct or indirect form. The calculation of these two functions,
which are the solutions of the Volterra integral equations (83), (112)
can be made using the methods illustrated in 7. From a general point of
view the functions Ri*, J1 have to be calculated in a numerical way from
the Volterra integral equations (83) and (112), or, in an approximated
form, using expression (58). This expression, defines a finite exact
functional relationship between x. R and J which, as a first
approximation
•
can be considered valid also when the functions R1*, J.*
1
are
1ntroduced. ~- ..
According to this hypothesis, remembering eq. (58) we can write

X= ---=1~­
R• (124)
1- 1
g(c) (to)

Solving eq. (124) for R1* we obtain

(125)

In the same way putting in eq. (58) R1 and J1 , we have:

1
X = ----=1'=---
1 - R1
(126)
g(c) (to)

Remembering eq. (111) and eq. (58) we can write

1 - g(c~to) = 6)j(1- g(c)~to)) (127)

---~~1~-- = ~ (128)
1-R / g(c) (t 0 ) fP

Introducing eq. (128) in (127) and comparing with eq. (126) we obtain

(129)
Time dependent behaviour 211

Eqs. (125), (129) give approximate values for the functions R1* and
They are very simple to use and their accuracy is generally satisfactory
so that for practical purposes these equations can be conveniently
recommended for the analysis of non homogeneous structures.
In an alternate way as explained in [30] , we can derive these functions
adopting the classical viscoelastic model or the Dishinger model in order
to describe the creep behaviour of concrete.
At this scope,for the classical model, inserting (A) 1 1p. in eq. (24) we have

R•<c> = E [1 + (A) lp (1 - ~) (1+"••->] (130)

i 1+(1)1'· i • '·

and remembering eqs. (111) and (22), eq. (112) can be put in the
differential form

(131)

with the initial condition

Integrating eq. (131) and remembering eq. (23) we obtain

;Ttl =
1
-
1 [
1 + ' · (1-(A) 1 - -
• ( ' · -lp )(1+ (1-.. j) ,_,1 (132)
1+(1-(A)~)Ip. E lp.

As regards Dischinger model, from eq. (40) we have:

(133)
R 1•!Dl -- E28 e-"Jf(D)

and from eq. (112) remembering eqs. (111) (40) we obtain the differential
form
oJ(Dl
__1_ + (1 - (A).j) J-'fl =1 (134)
acp<D>
-;Tfl ( t= t 0) = E28

having the solution

-;Tf1 = 1 [ •
1 - (A) 1 e - (1- .. j) fiDI] (135)
E2s (1- (1)*1)

Eqs. (125), (130), (133) and eqs. (129), (132), (135) according to the
212 [Link]

properties of the classical and the Dischinger models satisfy the

inequalities:
R'fl < Rj < Rj (cl (136)

(137)

10 . NUMERICAL EXAMPLES

As a first example let us consider the cable-stayed bridge of fig.

24, taken from [31]. The permanent load, applied at t 0 = 28 days gives
the subsequent initial values, elastically evaluated, for the forces in
the cables
cable n. 1 2 3 4 5 6 7
force (ton) 45,20 67,20 73' 10 70,20 57,7 33,8 339,4

The behaviour of concrete has been described by means of the CEB MC78
Model obtaining the time diagrams of the cable forces, related to their
initial value,~reported in fig. 25 and fig. 26. These values are obtained
from eq. (88) with ?I<:> = 0 as the prestressing is not considered in
this calculation. We can see that the variations in the cable forces are
quite large as the structural non homogeneity is marked .
The increments of the axial force in the cables 1 to 6 and in the cable
7 produce a reduction of the bending moments in the beam which as it can
be argued from eq. (88) tends to assume a distribution similar to that
corresponding to rigid cables. It is finally important to observe the
high accuracy connected to the analysis performed on the basis of the
algebraic creep law which gives the results represented by the small
circles in fig. 25 and 26.
In fig. 27, taken from [32], an analogous problem is studied for load
applied at t 0 = 112 days, using the CEB MC78 Model or the ACI Model to
describe the creep behaviour of concrete.
In fig. 28 the time increments of the cable forces, are reported and a
not negligible difference between the results deriving from the two
models arriving to a maximum value of about 6,5% is observed. In fig. 29
the bending moment diagrams are reported, showing the drastic reduction
of the maximum values produced by creep which, increasing the forces in
the cables, makes the moment distribution to tend toward that
corresponding to rigid cables. This distribution, as indicated by eq.
(88) cannot be achieved, as for t -+ "' we have !I(oo) ~~~.g~ig • This is
possible only applying the prestressing forces given by eq. (96) and in
this case the structure behaves as a structure with rigid restraint,
Time dependent behaviour 213

... ..... :
J!.DO

I:
CABLE
AREA [sq. em)

SECTION 8-8 SECTION A-A

i"i<J.24 Cable stayed bridge (from Re'£.31)

which, according to the fundamental theorem of eq. (63) maintains

constant in time the state of stress, increasing in an affine way the
elastic initial deformation.
As a final example let us consider the composite ste~ ~~~crete section
of a bridge beam illustrated in fig. 30, subjected to the constant
bending moment M0 = 2531 tm and referred to a system of coordinates
having its origin in the centroid of the homogeneized section. For this
section, as reported in [33], the two eigenvalues are respectively ~ 1 • =
0,90672; ~ 2 • = 0,00246, and according to the ACI Model the two reduced
creep functions have been calculated by means of the
214 [Link]

X. (t) /X. (t )
~ ~ 0

1.5
..;::::::.<
~
1.4
1.3
~
~
~2
-
-
!--"'
v~

-
L::::..3
1.2

1.1
b ~ v-- ~

r-..
4

~ ~
~
1.0

.9
1'-
r-- r-.. ...._
t-- f-..G_
.8
28 102 103 104
Fig.25 Relative time-variations of the
forces in cables 1 to 6 (from Ref.31)

X (t)/X 7 (t)
7 0

1.3

1.2
~
~

r
~
1.1
1.0

.9

.8
28
Fig.26 Relative time-variation of the
force in cable 7 (from Ref.31)

iterative numerical procedure of eq. (53). The same functions have been
determined also adopting the two limit models according to eq. (132) and
(135) and by means of the algebraic creep law of eq. (129) with the
approxi~ate constant value X= 0,8. The results reported in Table 2 show
that the bounds of the inequality (137) are very close so that the
sectional analysis performed following the simplified models is of good
accuracy and can be adopted for practical calculations. It is important
Time dependent behaviour 215

+
f--:--J]~
1·~8§;n
-
Fig.27 Cable-stayed bridge (from Ref.32)
10 ----

6(1)

- - - 1 - - - - - j Sll.l

.!l.z~~----
/---- - \..!![Link].u
-------- m.s

SM~-------------4--------------~------~
Nz,th &.IJ.I

~.3.::~~-=·~~··~~---;;-,--,;-(-,:;----_-_-_-_-_-_+-_-_-_-_-_--_--1~:~::
·~ f-------------,_ _____________+------;
Fig.28 Relative time-increments in
cable forces according to CEB and
ACI models (from Ref.32) ----+----··11.0
4~;-------~~~~~,-----~--------t-------;
N 1 !CE~IV
?
--------- ------ll6.1

u;i;:cu I
...-

"'If H),lh)l4,4
~ I
)II
liZ zoo
216 [Link]

ELASTIC--
CEBMC78 -·-·-
ACI
Actions scale-- =SOOtm

ECIXIIIII
I 900 00
ElASTIC r£B
00
ACl
I oo
ISK.
I
...
•so
E ASTit
·!iOU
tU
·2061
At I
-m•
-nso
liS 617. 7190 710)
-•z• ' 11 ·ZSZ6 ·ll6.6
•• <J .m.e

..
IS 0 -t9JS 1910 10 lO.O ·1&.)1
OIS •UOS 1r.9J 117l n IS sou 110.0 loOt••

~ ••u
,

600 -1076
SIS ·lllS
-~l
·ll'l s
_,,
-10&.6
I
I
I II 1"0
ll 7.5
IOZO
lllllf ftZ.S
l!O.O
IZJ1
LS,O -S902 ·2l29 ·liSt ' 14 OD 1111 !906 6961

/ '1\ '
*
/ I \'
1/ I \
,r / \: '
J---" / 'k --!'
,,/.t-- - -r '{...::::-"\ X !
/ 1 S 6 7I 9 Ill

4
Fig.29 Bending moment diagram at t=t and t=lO days (from Ref.32)
0

to note that the inequality (137) is always satisfied by the results of

the general algorithm, for which the accuracy is very high while adopting
the approximate algebraic creep law of eq. (129) the inequality may be
in some cases violated. In the present example this happens for (t-t')
10 2 days but also in this case the difference with respect the
numerical algorithm is very small, so that the algebraic creep law, owing
to its simplicity and accuracy can be conveniently recommended for

76.67
16F 8

272

I 235

Fig.30 Composite steel-concrete section

480 ~ 2)5 j
(from Ref. 33)
Time dependent behaviour 217

Numerical Method Dischinger Model Classical Model Algebraic Model

eq. (53) eq. (139) eq. (132) eq. (129)
t-t' E
- E J E J E J E
28J1 E28J2 28 1 E28J2 28 1 E28J2 28 1 28J2
2
10 2.29 1. 00175 2.31 1.00200 2.28 1.00170 2.26 1.00120
3
10 2.76 1.00185 2.79 1.00220 2.70 1.00180 2.70 1.00190
4
10 2.93 1.00185 2.97 1.00220 2.82 1.00180 2.86 1.00190
Table 2. Reduced Creep Functions, (t-t') in days.

practical purposes. In fig. 31 the stresses in the beam and in the slab
are reported, showing the marked redistribution which takes place due to
creep. The relative increment of the curvature, measured by the slope of
the stress diagram in the elastic beam is about 457. and, as the beam

64

.....
CXl
CXl
0'1
.....

-968 -911
2
Fig.31 Stress diagrams at time t 0 and time t=lO years (kgf/cm )

section is constant along the longitudinal axis, this increment is the

same for the transverse displacements. It is important to observe that
the increase of the transverse displacements is quite small, as the
elastic part is large compared to the concrete slab.

11 . CONCLUDING REMARKS

The creep structural analysis of R.C. and P.C. structures requires

different computational efforts connected to the level of non homogeneity
of the structure and to the peculiarity of the mathematical algorithms
adopted for the solution of the system of integral equations which
governs the problem. If we follow the hypothesis of structural
218 [Link]

homogeneity, as it is possible for R.C. and P.C. structures, affected by

a very small amount of reinforcement, the knowledge of the creep and of
the relaxation function is sufficient in order to predict the structural
behaviour under imposed loads and deformations. As the creep models
suggested by the Codes are expressed only in terms of creep function, the
analysis of homogeneous structures requires the solution of the Volterra
integral equation relating the functions J and R in order to obtain the
relaxation function R. This function is obtained by inverting the creep
constitutive law, or in other words, by determining the resolvent kernel
of the Volterra integral equation expressing the creep law, when the
deformation due to the applied stress has a constant unit value.
The analysis of non homogeneous structures, in particular the
viscoelastic structures elastically restrained, requires more complex
algorithms, which, however, can be reduced to only one general procedure
named the Reduced Relaxation Function Method. This procedure can be
applied in its direct form when the forces method is followed orin its
inverse form when the displacements method is adopted. In both cases two
particular sets of functions, namely the reduced relaxation functions or
the reduced creep functions have to be determined by solving a sequence
of independent Volterra integral equations. The algorithms for the
solutions of these equations can be conveniently chosen in the range of
the approximate algorithms related to simplified creep models or to
simplified algebraic creep law, as in many practical cases the accuracy
of the results is quite satisfactory.
Only when the results obtained following the simplified creep models are
too much different, more refined algorithms as the one based on the
iterative application of the superposition principle are required in the
evaluation of the reduced relaxa~ion or creep functions.
In every case, however, the Reduced Relaxation Function Method stands as
the most efficient tool for the analysis of homogeneous viscoelastic
structures elastically restrained, particularly for the analysis of P.C.
or composite steel-concrete sections subjected to axial load and bending
moment. In this case the problem involves only two unknowns, so that the
calculation of the eigenvalues of the interaction matrix is immediate and
the evaluation of the reduced creep functions which can be performed by
means of the simplified procedures shows in general good accuracy and can
be easily applied by the practitioners.

REFERENCES

1. Troxell, G.E., Raphael, J.M., Davis, R.W.: Long Time Creep and
Shrinkage tests of Plain and Reinforced Concrete, ASTM Proceedings,
58(1958)' 1-20.

2. Reichart, T. W.: Creep and Drying Shrinkage of Lightweight and

Normal Weight Concretes, National Bureau of Standards, U.S.A. NBS
Monograph 74 (1964), 1-30.
Time dependent behaviour 219

3. Hansen, T.C., Mattock, A.H.: Influence of Size and Shape of Member

on Shrinkage and Creep of Concrete: ACI Journal 63 (1966) 267-289.

4. L'Hermite, R., Mamillan, M.: Retrait et Fluage des Betons,.Annales

ITBTP, Supplement 21, 1968.

5. L'Hermite, R., Mamillan M., Lefevre, C.: Nouveaux Resultats de

Recherches sur la Deformation et la Rupture du Beton, Annales ITBTP
18 (1965) 325-360.

6. Wei-Wen Yu, Winter, G,: Instantaneous and Long-Time Deflections of

Reinforced Concrete Beams Under Working Loads, ACI Journal 57
(1960) 29-50.

7. Mamillan, M., Savin, V.: Etude Experimentale sur le Fluage du

Beton. Verification. du Principe de Superposition, Materiaux et
Constructions, RILEM, Vol. 14, 81 (1981) 177-189.

8. Hanson, J.A., :A 10-year Study of Creep Properties of Concrete,

Concrete Laboratory Report N. Sp-38, U.S. Department of the
Interior, Bureau of Reclamation, Denver, Colorado, 1953.

9. FIP (Federation Interantional de la Precontrainte) - CEB (Comite

Eurointernational du Beton): Code Modele pour les Structures en
Beton. CEB Bulletin d'Information N. 124, 1978.

10. AASHTO, Standard Specifications for Highways Bridges, 11th Edn. 1-

467, AASHTO, Washington, D.C. 1973.

11. Italian Ministry of Public Works: Italian Building Code for R.C.
and P.C. Structures, 1985, (in italian).
12. EH-82 Instruccion para el Proyecto y la Ejecucion de Obras de
Hormigon en Masa o Armado, Comision Permanente del Hormigon,
Madrid, 1986.

13. Me Henry, D. : A new Aspect of Creep in Concrete and its Applciation

to Design, Proceedings ASTM, 43 (1943) 1069-1086.

14. Ross, A.D.: Creep of Concrete under variable Stress, ACI Journal,
Vol. 54 (1958)" 739-758.

15. Maslov, G.N., Thermal Stresses in Concrete Masses with Account of

Concrete Creep (in Russian), Gosenergoizdat, 28(1940) 175-188.

16. ACI (American Concrete Institute) Committee 209: Prediction of

Creep, Shrinkage and Temperature Effects in Concrete Structures,
in : Designing for Effects of Creep, Shrinkage and Temperature in
220 F. Mola

Concrete Structures. ACI Publication SP27-3 (1971) 41-93.

17. Rusch, H., Jungwirth, D., Hilsdorf, H., : Kritische Sichtung der
Verfahren zur Berucksichtigung der Einflusse von Kriechen und
Schwinden des Betons, Beton und Stahlbetonbau, H. 68 (1973) 49-60;
76-86; 152-158.

18. Nowacki, W.: Theorie du Fluage, Eyrolles, Paris, 1965.

19. Persoz, B.: Introduction a l'Etude de la Rheologie, Dunod, Paris,

1960.

20. Bland, D.R.: Theory of Linear Viscoelasticity, Pergamon Press,

Oxford, 1960.

21. Dischinger, F.: Untersuchungen uber die Knicksichereit, die

Elastische Verformung und das Kriechen des Betons bei Bogenbrucken,
Der Bauingenieur H. 18 (1937) 487-520; 539-552; 595-621. H. 20
(1939) 53-63; 286-294; 426-437; 563-572.

22. Whitney, C.S.: Plain and Reinforced Concrete Arches, ACI Journal
Vol. 28 (1932) 479-519.

23. Glanville, W.H.: The Creep or Flow of Concrete under Load, Studies
in Reinforced Concrete, part III, Dept. Scientific and Industrial
Research, Building Research Technical Paper N. 12, London (1930).

24. FIP-CEB Manual on Structural Effects of Time-dependent Behaviour

of Concrete, CEB Bulletin d' Information N. 142-142bis, Georgi
Publishing Co. Saint-Saphorin, CH 1984.

25. Rusch, H., Jungwirth, D: Stahlbeton Spannbeton, Band 2, Werner

Verlag, Dusseldorf 1976.

26. Branson, D.E.: Deformation of Concrete Structures, Me Graw-Hill,

New York, 1977.

27. Trost, H.: Auswirkungen des Superposition Prinzips auf Kriech und
Relaxation Probleme bei Beton und Spannbeton, Beton und
Stahlbetonbau H. 62 (1967) 230-238.

28. Bazant, Z.P.: Prediction of Concrete Creep Effects Using Age-

Adjusted Effective-Modulus method, ACI Journal, Vol. 69 (1972) 212-
217.

29. Mola, F.: Methods for the analysis of linear viscoelastic

structures, Studi e Ricerche Vol. 3 Italcementi Bergamo (Italy) (in
italian) (1981).
Time dependent behaviour 221

30. Mola, F.: Linear viscoelastic analysis of non homogeneous

structures, Studi e Ricerche vol. 8 Italcementi Bergamo (Italy) (in
italian) (1986) 119-196.

31. Mola, F., Malerba, P.G., Pisani, M.A. Creep and Shrinkage Effects
on the Cable-Stayed Bridges Behaviour, Proceedings of the
International Conference on Cable-Stayed Bridges, Vol. 1, Bangkok,
Thailand, (1986) 657-667.
. . I .
32. Mola, F., Malerba, P.G., P1san1, M.A.: Structural non homogene1ty
and additional restraints effects on the long term behaviour of
precast R.C. or P.C. Bridges, Proceedings of the Symposium of the
Italian R.C. and P.C. Association, Vol. 2, Stresa Italy (1987) 505-
519 (in italian).

33. Mola, F. : Applications. of the Reduced Relaxation Function Method

to the Analysis of homogeneous viscoelastic structures, Studi e
Ricerche Vol. 4 Italcementi Bergamo (Italy) (in italian) (1982)
211-235.
 TECHNOLOGICAL FACTORS IN THE DESIGN
OF FOUNDATION PILES

C. Viggiani
University of Naples, Naples, Italy

ABSTRACT
It is well known that the behaviour of foundation piles
is markedly affected by technological factors, particularly
for cast in situ piles. Some examples are presented to
substantiate this statement, concerning recent experiences
with large diameter bored piles and continuous flight auger
piles in the pyroclastic soils of eastern Naples area.
A pre-loading cell, aimed at improving the load-settle-
ment behaviour of the base of large diameter bored piles, is
then described.
Finally, some data and comments are presented about
loading tests on instrumented piles and non destructive
integrity test methods, such as sonic logging and mechanical
admittance tests.
224 C. Viggiani

!.INTRODUCTION
Piles are often adopted in the foundations of bridges,
and indeed they represent by far the most frequent solution.
There are many reason for this: the high values of loads,
the frequent occurrence of weak soils, the need of founding
deep enough to be safe against scouring effects.
A complete discussion of all the problems of pile
foundations is clearly out of the scope of this paper; only
some aspects, that are believed to be particularly intere-
sting or topical, well be touched.
As it is well known, the behaviour of the foundation
piles is markedly affected by technological factors. The
influence is particularly significant for cast in situ bored
piles that are the large majority in Italy and are parti-
cularly suited for bridges, due to the ease of reaching
diameters up to 2 m and lengths up to many tens of meters.
This paper will be devoted to a discussion of some of these
factors.
Most of the experiences reported, though being of gene-
ral relevance, have been collected in the pyroclastic soils
of eastern Naples area. The essential features of the
subsoil in this area are therefore presented in par. 2.
Some striking examples of the influence of technologi-
cal factors on the behaviour of large diameter bored piles
(par. 3) and continuous flight auger piles (par. 4) are
reported.
A particular type of pre-loading cell, aimed at impro-
ving the load settlement behaviour of the base of large
diameter bored piles, is described in par. 5. The results of
a full scale investigation on instrumented piles are repor-
ted and discussed, pointing out the features connected with
the use of the pre-loading cell.
The marked and somewhat random influence of technologi-
cal factors makes the theme of controls particularly signi-
ficant. Some aspects of loading tests on instrumented piles
are discussed in par. 6, while non destructive integrity
test methods, such as cross hole sonic logging and mechani-
cal admittance tests, are described in par. 7.
Foundation piles 225

2. THE SUBSOIL OF EASTERN NAPLES AREA

The plain lying east of the city of Naples is being

interested by an intense constructio n activity, in
connection with enterprises like the new Directiona l Centre
with its high rise buildings, the new Law Court, roads and
hydraulic infrastruct ures. Most of these structures are
founded on piles; a large amount of experience on their
behaviour became therefore available.
The subsoil of the area had been thoroughly investi-
gated by a number of Authors (CROCE, PELLEGRINO, 1967;
RIPPA, VINALE, 1982; VINALE, 1988), and appears rather uni-
form in its essential features (fig. 1).
starting from the surface and going downward, the fol-
IO

~ Sands filling
~ erosion ditches

~ Volcanic ashes and

~ organic soils Pozzolana

r:: : : j St r atified sands ~ Volcanic tuft

~
~ Cemented levels within the stratified sands

Fig. 1. Schematic constitution of the subsoil in eastern Naples area

 [Link]~wrn

lowing soils are typically found:

- made ground;
- volcanic ashes and organic soils:
- stratified sands;
- pozzolana, loose or weakly indurated;
- volcanic tuff.
The made ground has a thickness of 3 - 4 m and consists
essentially of masonry remains and tuff fragments embedded
in a silty brown matrix of pyroclastic nature.
The ashes and organic soils represent the bottom of
the marsh that occupied this part of the plain until less
than a century ago. The ashes have a silty texture and
exhibit a slight plasticity: the organic soils are generally
pyroclastic sediments with diffused organic matter, but
sometimes peat layers are found.
The stratified sands are alluvial soils of volcanic
origin (sand, pumices, lapilli and ashes) washed out from
the surrounding hills and deposited under lagoonal or marsh
environment. The thickness of each layer ranges between some
tens of centimeters and some meters.
The sands are underlain by a pyroclastic formation
consisting of pozzolana (cohesionless or slightly indurated)
and tuff; they belong to the same geological formation, and
differ only by the degree of diagenesis.
The upper part of the formation, starting at a depth of
25 - 30 m, consists generally of cohesionless pozzolana:
then indurated pozzolana and tuff are found.
The tuff is a soft reef, with a compressive strength
ranging from 20 to 100 kgjcm : the thickness of the layer is
usually of some tens of meters.
At some places, depending on the conditions prevailing
at the time of deposition, the tuff is missing and the
pyroclastic formation consists entirely of cohesionless' and
slightly indurated pozzolana.
The ground water table is found at a depth of about 2 m
below the ground surface, that has an average elevation of
5 m above mean sea level.
Foundation piles 227

3. LARGE DIAMETER BORED PILES

Where the tuff is found within 30 - 35 m from the
surface, the majority of high rise buildings have been
founded on large diameter bored piles penetrating the tuff
for a length of 2 - 3 times their diameter.
Fig. 2 reports some CPT results referring to two diffe-
rent sites; it may be seen that the characteristics of the
soils overlying the tuff at the two sites are practically
identical, except for the depth of the tuff that is 28 m at
a site and 33 to 38 m at the other one.
Site 1 Site 2

lkg/cmq) lkg/cmq)
4BB 4BB

38

z z
lm) lm)

Fig. 2. Profiles of the point resistance of CPT in sites 1 (tests a in

fig. 3) and 2 (tests b, c and d in fig. 3)

In both sites bored piles socketed into the tuff, with

a diameter of 0,8 m, have been adopted; the results of some
loading tests are reported in fig. 3.
At the site 2, piles have been drilled with bentonite
mud. The results of 7 proof loading tests are included in
the narrow band a in fig. 3; it may be seen that the scatter
of results is rather low. The displacements of piles head
are smaller than the elastic shortening of the shaft of the
piles, thus indicating the occurrence of a significant side
resistance.
At the site 1, in order to avoid the practical problems
connected with the management of drilling mud, piles where
drilled within a thin metallic casing that was left in the
soil. Proof loading tests (curves d in fig.3) gave very
unsatisfactory results. The retrieval of the casing after
228 C. Viggiani

concreting the pile improved the behaviour (curves c). Fi-

nally a vibrator was adopted to install a metallic casing
for the whole pile length; the soil within the tube was then
removed and the socket in the tuff drilled by an auger;
after concreting, the casing was retrieved, always by vibra-
tion. This technique gave results (band b in fig. 3) compa-
rable to those obtained in the site 2.
The data reported in fig. 3 appear very effective in
exemplifying the influence of the construction techniques on
the behaviour of the foundation piles, even in a case (piles
bearing on a rock base) where such an influence was not
expected to be significant.
0 100 200 3 0
0

10

15

s
(mm)

Fig. 3. Results of loading tests on bored piles with a diameter of 0,8 m,

end bearing on the tuff~ a) n. 7 tests on piles with L = 35 - 40 m
bored with bentonitic mud; b)n. 4 tests on piles with L ~ 30m, ~
red within a casing introduced in advance by vibration; c) piles
with L = 30 m bored with a casing; d) piles drilled with a casing
left in the soil
Foundation piles 229

4. CONTINUOUS FLIGHT AUGER PILES

Continuous flight auger piles (VAN IMPE, 1984, 1988;
MASCARO!, 1985; VAN WEELE, 1988) are increasingly used in
many circumstances; contractors claim that they offer the
advantages of both driven and bored piles without the re-
lated disadvantages; in particular, they are claimed to be
relatively insensitive to construction procedures.
Auger piles are installed by means of a hollow stem
with an inner diameter of 10 - 15 em, provided with a screw
on its outer surface. The auger is inserted into the soil
by the combined action of an axial thrust and a torque; the
ratio between the vertical travel and the revolutions, howe-
ver, is always less than the pitch of the screw. Accordin-
gly, the penetration implies a lateral compression of the
soil and also the removal of some soil.
After the auger has reached the desired depth (fig. 4),
the temporary closure plate below the central stem is pushed
away by pumping concrete or mortar through the stem and the
auger is lifted, removing from the ground the soil within
the screw. Being generally the volume of the removed soil
less than the volume of the pile, the net resulting effect
is a compression of the soil surrounding the pile; the
resulting situation is intermediate between that of a bored
and that of a driven pile.

Fig. 4 . Stages in the installation of a continuous flight auger pile (TREVI)

230 C. Viggiani

The procedure allows a safe and rapid installation of

piles with a diameter ranging from 40 em up to 1 m and a
length up to 30 - 35 m. It is suited to weak soils, produces
an effective horizontal compression of the soil, acts
without the vibration and noise of driven piles.
A major shortcoming of auger piles is the difficulty of
placing a reinforcement cage. This must be lowered into the
shaft immediately upon completion of concreting, and usually
it must be of a short length.
In order to improve this aspect, the diameter of the
central stem has to be substantially increased and the width
of the spiral welded at the outside decreased.
The pile known with the trade name of "Presso-Drill",
for instance, is characterized by a ratio between the diame-
ter of the stem d0 and the overall diameter d ranging
between 0,7 and 0,9.
The scheme is reported in fig. 5. After the auger has

Fig. 5 . Procedure of installation of a pile type Presso-Drill, developed

by Icels Pali (MASCARDI, 1985)
Foundation piles 231

reached the desired depth under the combined action of an

axial thrust and a torque, a reinforcinq caqe and a concre-
tinq tube are lowered into the stem. While concretinq, the
auqer is lifted by jacks contrastinq upon the concretinq
tube, that stops the caqe hinderinq its liftinq and com-
presses the soil below the temporary bottom closure plate.
Piles of the latter type have been extensively adopted
in eastern Naples area. In many applications, for instance,
piles with d = 0,5 m, d = 0,6 m, L = 16 m have been
adopted; loadinq tests qave values of the bearinq capacity
around 250 t.
At a site where the subsoil was characterized by the
CPT results reported in fiq. 6, piles with d = 0,71 m, d =
0,96 m, L = 21,5 m exhibited a bearinq capa~ity exceedinq
700 t (CAPUTO, VIGGIANI, 1988).
This kind of results is within the range obtained, in
the same soils, with driven piles.
A number of CPT have been performed, before and after
the construction of some piles, at a distance of about 1 m
from the pile axis; in fiq. 7 (CAPUTO, VIGGIANI, 1988) the
ratio of the mean values of q measured after (q ) over the
values measured before constr&ction (q~ is plof~ed against
depth. As in driven piles, a qeneral improvement of the soil
properties is obtained.
For another site, located around 500 m apart, the soil
profile and some CPT and SPT results are reported in fig. 8.
A large multi- storey garage has been founded in this site
on Presso-Drill piles with d = 0,66 m, d = 0,8 m, L = 12,5
m. In this case proof loa&inq tests have shown that the
~ ~- o 10 qc (MPa) 20

.....
T •.
made ground
0

...;:
~
·~
peat
"'"· ..
.
<i
~
~
stratified 70 -

~ I
Q .•·

sand
.., ....
4 .4_

'f'.
.... 'I' peat ~ It,
·.T· -
....... 20 L-..
p
.... coheslon/ess
[Link] :::J
c;
pozzolana z
...·... 1 {m)
1. ..• indurated
-4. pozzo/ana 30
a. ...
L
r-,
Fig. 6. constitution of the subsoil and average CPT results at a site in
the eastern Naples area
232 C. Viggiani

0 2 2
0

.~
5 I 1- 5
L

10
\ i- 10

z
(m~
II ...... z
{m)
~ ~~--~----~----~--~
Ratio between the values of qc measured after (qca) and before
(qcb) the installation of the piles of the Presso-Drill type

Fig. 7. Site of fig. 6 Fig. 9. Site of fig. 8

Nspr
0 10 20 30 40 50
qcfMPa)
0 10 20
=~~y~----- 0+---~----4---~----~

silty sand
with
interbedd
:-1:. pumices 10+---~---+
and

Fig. 8. Constitution of the subsoil and results of CPT and SPT at a

site close to that of fig. 6
Foundation piles 233

behaviour of piles was less satisfactory, and very similar

to that of bored piles.
The results of CPT performed before and after the
installation of the pile in the vicinity of some piles have
been processed as in fig 7 and are reported in fig. 9. An
overall decrease of the penetration resistance in the vici-
nity of the piles is evident.
It can be concluded that constructional factors affect
also the behaviour of auger piles.
In order to understand the reasons of this influence,
let us attempt an analysis of the penetration of the auger
into the soil.
From a kinematical viewpoint ([Link]), if vis the rate
of penetration and n that of revolution, in a time interval
Atwe have:
volume of displaced soil:

71'do_
__ 2

vd = v /Jt
4

volume of soil removed by the screw:

,.
va =
4

where 1 is the pitch of the screw.

If v = nl, v = 0; the screw penetrates into the soil
without removing a~y material. If v = o the auger acts as an

~------ d ------~

Fig. 10. Geometrical properties of the continuous flight auget

Archimedean screw, removinq a volume of soil:
w 2 2
va .ax = - Dl At(d - d 0 )
4

In order to have a net compression effect, the

displaced volume must exceed the removed one. From the
condition:
it follows that:
d2
v ~ nl (1 - - 0 - (1)
d2

Is this ratio between the rates of penetration and

revolution feasible? To answer this question, let us analize
the mechanics of the penetration.
Durinq the penetration (fiq. 11) the screw is acted
upon by: an axial thrust Q, derivinq from the weiqht of the
equipment plus the reaction of anchors, if any: a resistance
P of the soil on the temporary bottom plate closinq the
central stem: an axial force N produced by the torque M.
The latter can be expressed:
2M wd ..... ltqq,
N = (2)
d + do Jrdtqct> - 1

aJ
M~

...... tp• ....

Fig. 11. Forces acting upon the auger during the penetration
Foundation piles 235

having assumed that the coefficient of friction between the

soil and the screw equals tg~.
To have penetration, the sum of Q plus M must exceed P;
hence:

d + d0 1rdtg~ + 1
M ::?: (P - Q) (3)
2 7fd - 1tg~

The force N is transferred to the soil as a tangential

stress on the cylindrical surface of the screw; in order not
to have a shear failure, the shearing strength r . of the
soil should not be exceeded. llm
N
f = --
7fdZ
<
-
f (4)
lim

Assuming r . = ktg~fZ/2, where k is an earth pressure

coefficient, it 11!ollows from (2) and (4):

7fdtg~ + 1
1rdz 2 tktg~ (5)
4 w-d - 1tg~

Assuming the equ~lity in (3) and (5), combining them

0
and expressing Pas 7rd0 Nq z;4, the following expression of
Q is obtained:
,. z d 2 z
=- --(N - 0- - 2ktg~ - ) (6)
4 d q d2. d

The axial thrust Q (eq. 6) needed for penetration

vanishes when:
·z z N d2
= 0 or = _Q;o...__ _o_
d d 2ktg~ d2

The value of Q reaches a maximum at:

z N d2
9 _o_
= d2
d 4ktg~
236 C. Viggiani

The maximum value is:

Qmax
.,. d4
_o_
= N2
1i 32
q
d
4

The trend of Q/1(d3 as a function of z/d is plotted in

fig 12 for typical values of k, Nq and d~d. It may be seen
that, when a torque M expressed by eq. (3) is available, in
order to get a penetration without soil removal an axial
thrust Q is needed: its value increases with increasing
penetration depth, reaches a maximum and then decreases and
eventually vanishes.

Fig. 12. Axial thrust Q needed for penetration as a function of the depth
of penetration z
These findings agree with the common experience that
everyone has done when screwing a screw into the wood. At
the beginning a substantial thrust on the screwdriver is
needed, otherwise the wood is stripped: but once the screw
has penetrated a sufficient depth, only a torque is needed
to continue the penetration.
The maximum value of the axial thrust needed for pene-
tration without soil removal (fig. 13) increases with in-
creasing the ratio d~d.
Coming back to the auger piles, it may be concluded
that, in order to obtain a "good" pile, i.e. a pile where
soil displacement exceeds soil removal:
at the beginning of the penetration, a large axial
thrust is needed:
after reaching a sufficient depth, only a large
torque is needed:
Foundation piles 237

~
100

/v
v
10

Nq 50

/ Q25 Q75

Fig. 13. Maximum value of the axial thrust needed for penetration as a
function of the ratio d0 /d
both Q and M, and hence the power required to the
equipment, increase with increasing dold·
For the existing equipment (MASCARDI, 1985) Q ranges
from 2 to 10 t and M from 2 to 16 tm. It may be easily
realized that, in the majority of practical cases, the
available equipment have not a power sufficient to penetrate
the soil without removal. The penetration rate v, as already
stated, is then always lower than Dl, and some soil removal
during ther penetration can not be avoided. As a consequen-
ce, the soil surrounding the pile loosens, and the penetra-
tion becomes possible; the bearing capacity of the pile,
however, decreases.
If the condition (1) is satisfied, the net effect of
the penetration is a compression of the soil, and the re-
sulting behaviour of the auger pile approaches that of a
driven pile. If the eq. (1) is not satisfied, because the
equipment is not powerful enough for the conditions (soil
properties, dimensions of the pile) prevailing, the beha-
viour of the resulting auger pile tends to that of a bored
pile.
The evidence reported above is in overall agreement
with these conclusions. The "good" auger piles of the first
site (fig. 7) had a ratio 1/d in the range from 20 to 30 and
a ratio d~d = 0,75; these values are compatible with an
installati~n without an excessive removal of soil.
On the contrary, the piles of the second site (figs. 8
and 9), that behaved as bored piles, were characterized by
more unfavourable values of the ratios 1/d (15) and dold
(0,87).
238 C. Viggiani

5. PRE-LOADING CELL
As it is well known (see, for instance, Whitaker, 1976)
the base resistance of the large diameter bored piles is
mobilized after substantial vertical displaceme nts, its
ultimate value being attained at a displaceme nt in the range
of 15% - 25% of the pile diameter d.
As a consequence , the design of foundations with large
diameter bored piles is controlled by settlement rather than
bearing capacity considerati ons: in the usual design prac-
tice (WRIGHT, REESE, 1979) the working load is essentially
resisted by the side friction and only a small fraction of
the available base resistance can be mobilized.
In this situation, the use of pre-loading cells at the
base offers considerab le potential as a means of improving
the load- settlement behaviour of the pile base in order to
attain a better mobilizatio n of the base resistance at
working loads and, finally, a more economical design.
Early attempts of this kind had been based on mere
cement grouting below the pile base, through one or more
outlets at the pile bottom (SIMONS, 1961): various provi-
sions to obtain a uniform grout distributio n have been tried
(DIAMANT!, 1973). These systems proved unsatisfact ory in
some instances, especially in fine grained soils where a
uniform grouting is not possible and the grout mixture comes
in an irregular pattern of breakthroug hs ("claquages ").
Such experiences have led to the concept that, rather
than trying to grout the soil below the pile base, it is
convenient to reconsolida te it by means of a displaceme nt
grout acting essentially as a flat jack.
The pre-loading cell developed by Lizzi (1976) is based
on this concept. As indicated in fig. 14, it is formed by
two circular perforated steel plates having about the same
diameter of the pile, spaced 20 - 30 mm apart and completely
independen t from each other. The two plates are wrapped
together by an outer impermeable PVC membrane and an inner
canvas envelope, forming a closed cell: two or more injec-
tion tubes extend from the cell to the surface.
The cell is dipped at the bottom of the drilled hole,
together with the reinforceme nt cage, if any; the pile shaft
is then concreted.
Once the shaft concrete has hardened, grouting of the
cell is started through one of the injection pipes: the
Foundation piles 239

SECT I ON

PLAN

0 0 0 0
0
® 0 0 ®
0
0
0 0 ® 0 0
0
® ·o 0 ®
0 0
0 0 0 0 0
0 ® 0 ® 0 @ 0
0 0 0 0
0 0 0
0
® 0 0
®
0 0
0 0 ® 0 0

® 0
0
0 ®
0 0 0 0

Spacer

Fig. 14. Pre-loading cell develo~ed by Lizzi (1976)

second pipe acts initially as a vent and it is closed when
an outflow of grout is recognized.
After filling the cell, under increasing pressure the
grout makes the two plates separate and move apart; under
further increasing pressure, the PVC wrapping bursts leaving
free access to the grout into the soil and allowing free
movement of the plates. With this technique, a sort of flat
jack with unlimited travel is formed, compressing uniformly
the soil and pushing the pile upward; since no valves or
other kind of obstacles prevent the free flow of the grout,
the pressure measured at the surface allows a very reliable
estimate of the upward force exerted upon the pile bottom.
The injection is stopped when.
- either the pile top begins to undergo some upward di-
splacements (side resistance of the pile shaft
approaching its ultimate value),
- or the volume of injected grout increases without in-
creasing the pressure (a sort of bearing capacity fai-
lure of the soil below the pile base is being approa-
ched).
If needed, the injection can be carried out in more
than one stage.
Of course, besides the flat jack action, some improve-
ment of the $Oil around the pile base by grout impregnation
is likely to occur, especially in coarse grained soils.
240 C. Viggiani

Lizzi (1976) suggests to obtain a preliminary estimate

of the ultimate bearing capacity Qult of the pile from the
maximum pressure that can be applied to the pre-loading cell
during the pre-loading stage. He argues that, if the maximum
grouting pressure p. produces an upward displacement of the
pile top, then it may be stated that:
p.A
1
= S: p.A
1
~ p (7)

where A is the area of the pile base, s the ultimate side

resis~ance and P the ultimate base resistance. Alternative-
ly, if the maximum grouting pressure produces a steady
increase of the injected volume without displacing the pile,
it may be stated that:
p.A
1
~ S (8)

In both cases:
Q
ul t
=P + S ~ 2p.A
1
(9)

Eq. (9) represents an estimate of a lower bound of the

bearing capacity. It is based on the assumption that: (i)
the pre-loading cell actually acts as a flat jack: (ii) tha
values of s and P for upward and downward loading of the
pile are equal. Both assumptions appear to be conservative.
Some full scale observations have been carried out by
Viggiani and Vina1e (1983), who performed load tests on four
piles with a length of 42 m and a diameter of 1,5 m (piles A
and C) or 2m (piles Band D). The test area is located, as
usual, in the eastern Naples area, in a site where the tuff
ismissing: the piles bear in the slightly indurated pozzo-
lana and were drilled under bentonite mud.
All the piles were instrumented in 8 sections by strain
gages extensometers, allowing the determination of the di-
stribution of the axial load along the pile shafts. Piles c
and D were provided with a pre-loading cell of the type
(iescribed above.
Fig. 15 reports the volume Vt of grout injected into
the cells during the pre loading stage, as a function of the
injection pressure pi .
The maximum pressure attained was equal to 4,0 MPa for
pile C and to 4,4 MPa for pile D: the corfesponding volumes
of pumped grout have been of 2,5 and 4,1 m respectively. On
the basis of previous experiences in similar soils, the
final volumes of indurate~ displacement grout should be of
the order of 1,2 and 2,0 m respectively.
It may be noted that the ratio between these volumes
and the area of the pile base is equal to about 0,65 m for
Foundation piles 241

s~------------------------------~
3

4 ,.,.)\\ \
,. ,.
/
\
\

rl " \\
I \ p
/ \1
' ' \(/ o----J' p1'I e D

0~------~-------r------~------~
0 2

Fig. 15. volume Vi of grout injected during the pre-loading stage of the
piles c and D as a function of the injection pressure Pi

Q <MNl Q <MNl
0 2 4 6 8 0 2 4 6 6 10 12 14
0
1 -
Stage 1\ lljA Stage P1 WI
<MPal <MNl <MPal <MNl
2- 10 <D _1.0 1.8 1.2 3.8
@ 2.5 4.4 2.4 7.5
3-
@ 3.6 6.4 4.4 13.6
4- 20
pileD
5-

6- 30
®
7 - <m>

6 - 40
tl
P1
Fig. 16. Distribution of the axial load Q along the shaft of the piles
C and D during the pre-loading stage

both piles; this length corresponds to the travel of an

ideal flat jack expanding with a constant section equal to
the cross section of the pile. The actual travel is likely
to be much smaller, due to lateral expansion of the grout
bulbs.
During the grouting operations, the distribution of the
axial load along the shaft of the piles was measured; some
of the results obtained are reported in fig. 16.
 242 C. Viggiani

These findings substantiate the idea that the pre-

loading cell acts essentially as a flat jack; as a matter of
fact, the axial load measured near the pile base is very
nearly equal to the grout pressure multiplied by the cross
section of the pile. This upward force is resisted by side
friction and, accordingly, it decreases moving upward along
the shaft of the pile, vanishing near the pile top.
30 days after injecting the cells, the four piles were
load tested to failure, except pile D for which the capacity
of the reaction frame was exceeded before attaining failure
the scope of this paper; in fig. 17, anyhow, the load-
settlement curves of the four piles are reported. Besides
the curve total load Q - top settlement s, also the curves
side resistance s-settlement at mid length and base resi-
stance P-settlement at the base are reported.
A marked improvement of the performances of piles
equipped with the pre-loading cell may be observed. The
total bearing capacity increases by 30% - 40%, the increase
being mostly accounted for by a corresponding increase of
the base resistance P. Still more evident is the improvement
of the load-settlement behaviour, due to an earlier mobili-
zation of both the base and side resistances.
In fig. 18 the distributions of the axial load along

P, S,Q <MN) P, S,Q <MNl

5 10 15 20 25 0 5 10 15 20 25 30

p s Ql
I
5 I
I

(
I

10
s
<cml Pile A d=1.5m no PLC Pile C d=1.5m with PLC
15~--------------------~

0~~~~--~--~--~

5
failure ofjl
l '\ \
Q
reaction 1
I
l I
\
\ frame
10 I I \
l

s
(em) Pile 8 d = 2m no PLC Pile 0 d=2m with PLC
15~--------------------~

Fig. 17. Load - settlement curves for the four piles during the loading
tests. Total load Q, siae resistance s and point load P are plo~
ted respectively against the settlement of the top, the mid point
of the shaft and 'the base of the pile.
 243

QziQ
0.5 1

Pile C
d =1.5 m
with PLC

- 10

20

- 30
<ml d=2m
no PLC
40

Fig. 18. Distribution of the axial load Qz along the shaft of the four piles
during the loading tests

the shaft of the piles at various load levels are reported.

In piles A and B, not equipped with the pre- loading
cell, the mobilization of P and S has been very nearly
proportional, and P accounts for 10% - 15% of the total
applied load over the whole range of loads.
In piles c and D, with the pre-loading cell, the per-
centage of load taken by the pile base increases with in-
creasing load level, reaching 30% - 45% of the total load in
the failure range.
The data collected allow a check of the procedure to
evaluate a lower bound of the bearing capacity presented
before (eq. 7 to 9). The relevant data are listed in the
following table.
pile d(m) P. (MPa) 2p.A(MN) S(MN) P(MN) Qult (MN)
c 1,5 I 4I 0 1412
1 11,5 8 19
D 2,0 4,4 27,6 19,0 8 28

Eq. (9) appears to be satisfied in both cases, thus

substantiating the proposed procedure; it must be said,
however, that this result may be somewhat fortuitous s1nce
eq. (8) is not satisfied by the results obtained on the pile
D.
244 C. Viggiani

6. LOADING TESTS ON INSTRUMENTED PILES

Loading tests of foundation piles are rather common in
the construction practice, and are indeed the most satisfac-
tory procedure to check the successful! construction of a
pile (proof tests) or to measure its bearing capacity (tests
to failure). The test set up and procedures are well known
(MARCHETTI, 1975; AGI, 1984).
For small and medium diameter piles the maximum test
load is in the range of some tens to a few hundreds of tons.
In these cases, a number of piles between 2% and 4% of the
total is usually proof tested in Italy.
In the case of large diameter piles, the test load
attains many hundreds and even thousands of tons; the test
is therefore extremely expensive. For this reason, the
number of tests that are performed in practice is very
small, just for those piles that need a careful control
because of the very high service loads.
The attempt of getting a maximum of data from such
expensive tests appears therefore justified. This can be
done by instrumenting the shaft of the pile to determine the
distribution of the axial load and the transfer curves
across the base and along the lateral surface.
Besides providing valuable informations for the founda-
tion design in the particular case under investigation,
these findings allow the development of semiempirical design
criteria (VIGGIANI, 1973; MEYERHOF, 1976; GHIONNA et al.,
1979). And indeed the availability of a wealth of data
referring to relatively uniform soil types and to well
defined construction techniques is believed to be a prere-
quisite for a substantial improvement of our capacity of
predicting the behaviour of piles.
Accordingly, some informations on the instrumentation
of piles are given below, with the aim of promoting a dif-
fusion of these techniques, that are relatively simple and
whose cost is not high in comparison to the total cost of a
loading test.
Further details can be found in a number of papers
(MARCHETTI, 1975; WRIGHT, REESE, 1979; CAPUTO et al., 1989;
ROCCHI et al., 1989).
The load is applied to the test pile by a hydraulic
jack, and it is normally determined by measuring the oil
pressure in the jack. It has been found that this measure-
ment is very unreliable, because of the random occurrence of
Foundation piles 245

significant friction between the ram and the cylinder; the-

refore the provision of a load cell interposed between the
jack and the reaction frame is mandatory.
Rugged load cells, suited for use under the adverse
conditions prevailing on a construction site but yet very
accurate and reliable are now currently available; their
full scale capacity reaches even 1000 tons (fig. 19).
The distribution of the axial load along the shaft of
the pile may be determined in the following ways.
- Direct measurement of the load by means of load
cells interrupting the continuity of the shafts at
various depths (PRICE, WARDLE, 1983).
- Measurements of the vertical displacements of a
number of points at various depths by means of tell-
tales (SNOW, 1965) or amovable extensometers (BAGUE-
,LIN, JEZEQUEL, 1975).
- Measurements of the axial strain in some sections by
means of suitable strain gages (REESE, O'NEILL,
1988; CAPUTO et al., 1989).
The direct measurement is the most satisfactory techni-
que, since it is unaffected by factors such as the effective
area of the pile and the compression modulus of the con-
crete. Unfortunately, it is very expensive and may be adop-
ted only in special cases.

Fi g . 19 . Hydraul ic jack and stra i n gages l oad cell; f ull scale 1000 t
246 C. Viggiani

The writer has systematically adopted the strain mea-

surement technique, with strain gagrs bounded to steel bars
with square cross section o,sxo,s em and a length of 40 em.
The strain gages are covered by a watertight resin protec-
tion.
These "strain bars" are fixed to the reinforcement cage
(fig. 20) in a number of at least 3 (better 4) in each
measuring section. The number of sections varies between 3
and 8, depending on the length of the pile, the subsoil
conditions and the li_berality of the sponsor.

Fig. 20. Strain bar fixed to the reinforcement cage

Soon after concreting the shaft, the strain bars are
checked and a systematic survey of the strain is carried out
during the curing of the concrete (fig. 21). The increment
of strain due to the shrinkage of the concrete must be
negligible at the time of load test, otherwise it could
affect all the subsequent measurements.
During a load test on an instrumented pile the follo-
wing data are collected:
- the curve of the applied load Q versus the settlement s
of the pile head:
- the values of the vertical strain Ez in the instrumented
sections at the various load levels.
To calculate the values of the axial load N from the
values of the axial strain Ez the axial rigidity EA of the
pile should be known (E = Young modulus of the concrete: A =
area of the cross section of the pile, taking into account
the reinforcement) .
For a pile cast in situ the area A can be determined
only as an average value over a certain pile length, by
monitoring the volumes of concrete and the levels of the
Foundation piles 247

16m

z= 24m

25
' t (g)

Fig. 21. Strain measurements at various depths in the shaft of a pile

during the curing of the concrete
surface of the concrete in the pile shaft during concreting.
The values of E depend on the curing condition prevailing in
the different instrumented sections, such as overburden
pressure and occurrence of ground water, as well as on the
composition of the concrete and the procedure of casting.
For a given pile it is usually assumed that the rigidi-
ty EA is constant over the shaft; with this assumption, the
distribution of N along the shaft is proportional to that of
the axial strain Ez and may be obtained from it once the
value of N in any section is known.
The value of N in the upper first instrumented section
may be taken as known, because this section is purposely
located near the pile head and the fraction of load taken by
side friction is negligible at that level.
A better approximation may be easily obtained, if
needed, by assuming a realistic distribution of the~ side
resistance between the head of the pile and the £irst
instrumented section.
From the measurements of the settlement of the head of
the pile and of the axial strain in the instrumented sec-
tions, the displacement of all the points of the pile may be
248 C. Viggiani

determined; it should be always directed downward. Sometimes

apparent upward displacements result from the data; should
this be the case, the assumption that EA is constant would
be no more acceptable and more sophisticated elaborations
would be needed accounting for the observed variations of A,
if any, for the non linearity of the concrete (variations of
E with the load level), for informations on the quality of
the concrete obtained by sonic logging, if any.
The final result of the elaboration is a diagram of the
axial load N along the shaft of the pile for each applied
load Q. Transfer curves of side shear may also be developed;
an example is reported in fig. 22.
An example of the kind of results that can be obtained
by loading tests on instrumented piles has been reported in
the previous par. 5.
25 I
Pile A
20

~15 3+10 m
1-
~
_,. -.--- ---- ---- 1o~15

ft/
-- --
20.;35
5

-.,.-
,-::;~: -=--'<=-
- -- -35+40
--
0 ~
;.'-
1----- 0+3
r-.J.5+20

0 2 4 6 8 10 12
a(z) (em)

Fig. 22. Transfer curves of side resistance at various depths. Pile A of

fig. 17
Foundation piles 249

?.INTEGRITY TESTING
The expensiveness of the loading tests on large diame-
ter piles makes the non destructive integrity testing me-
thods very interesting. They allow a rapid and cheap check
of the quality of the pile and make it possible to test a
large proportion of the piles (or even all the piles).
The test methods most widespread in Italy are the sonic
logging and the mechanical admittance (CARABELLI, LIBERATI,
1985).
As it is well known, the former requires 2 or 3 access
tubes for sondes to be cast in the pile. This implies a pre-
selection of the pile to be tested, with the possibility of
greater care being taken in the construction of these parti-
cular piles. Of course, if all the piles are tested this
disadvantage becomes an advantage.
Sonic logging is conducted from hole to hole by hori-
zontal scanning, dipping a transmitter probe in one hole and
a receiver in another; the logging is based on recording the
velocity of sound waves in the concrete between the trans-
mitter and the receiver. The results give continuous and
detailed informations about the quality of the concrete and
the occurrence of defects such as voids, porous concrete or
soil inclusions.
The cost is in the range of 200.000 - 400.000 lit per
pile, in addition to the cost of the access tubes. If these
are made by steel, as it is desirable, they can be computed
in the reinforcement area.
In the mechanical admittance method a steady pulsating
force or an impact force is applied to the pile head by a
small vibrator or by a light hammer. The induced velocity of
-the piie head as well as the applied force are then simulta-
neously monitored; from these data it is possible to re-
trieve informations regarding the integrity and stiffness of
the pile, its length and the occurrence of major changes in
its cross section.
The cost is in the range of 50.000 - 250.000 lit per
pile, and no previous preparation of the pile is needed.
The sonic logging, at present, is believed to be more
exhaustive and reliable; however, it gives only informations
on the quality and integrity of the concrete in the pile
shaft.
The mechanical admittance method is much more uncer-
tain, since expert interpretation of the results is needed;
250 C. Viggiani

Fig. 23. Results of sonic logging of a bored nile with d=2m and L=42m

nevertheless, it offers some potential of estimating the

bearing capacity of the pile and the pile-soil stiffness at
low strain. Recently, this method is evolving towards a
dynamic load test, in which a substantial force is applied
to the head of the pile (e.g., by the drop of a pile hammer)
(HERITIER, 1989).
In fig. 23 (VIGGIANI,VINALE, 1983) the results of sonic
logging on a large diameter bored pile are reported. A
striking defect is evident on a stretch of 2,5 m near the
head of the pile.
A large inclusion of soil and drilling mud was actually
detected by excavating around the head of the pile (fig.
24); it probably occurred during the withdrawal of the guide
casing.
In other cases the results obtained are less clear. At
a site in the eastern Naples area many hundreds of bored
piles, part end bearing on the tuff and part floating,were
instalied with diameters ranging from 0,8 m to 1,5 m and
lengths from 32m to 43 [Link] proof loading tests are
available, in addition to sonic logging of all the piles.
Some defects have been detected on four of the twenty
test piles; an example is reported in fig. 25. Nevertheless
(fig. 26) the results of the loading tests on the "defec-
tive" piles can not be distinguished from those on "intact"
piles.
From this and many similar evidences it may be con-
cluded that the integrity test methods, at their present
Foundation piles 251

Fig. 24. The upper section of the oile of fig. 23, after excavation
development stage, have an essentially semiqualitative cha-
racter
If the results of the test are satisfactory, the good
quality of the concrete of the pile shaft is warranted;
though being very useful, this information is far from being
exhaustive.

Fig. 25. Sonic logging of a bored nile with d = 0,8 rn, L 34 rn

252 C. Viggiani

If the test evident iates some defects , the assessm ent

of their gravity is not straight forward .
end
d{m) bearing float ing
0.8
1,0
•... 0
6
~ L=36m
1,25
., 'Q :~: defective

6
1,5
• 0

%
l%oi
•
36

36
! 36
4

g ~31
'"'j'1
~Vi'. eJ7

8 ~'~{*:\l4
2 8 038

•
32
8 ~
'~=
~
38
~
~

0
35 4b 45 50
4~;r (kfW:mzJ

Fig. 26. Settleme nt at the maximum test load measured on "intact" and
"defectiv e" piles
Foundation piles 253

8. CONCLUDING REMARKS
In this paper some data and comments concerning the
influence of technological factors on the behaviour of
foundation piles have been presented, without even
attempting a systematic and exhaustive state-of-the-art.
Due to the somewhat random influence of technological
factors and to the scatter of the properties of natural soil
deposits, even within the same foundation significant diffe-
rences occur in the behaviour of supposedly identical piles.
Some data on the variability of the load-settlement
behaviour, as observed in loading tests on a substantial
number of piles, are listed in the following table (CAPUTO,
VIGGIANI, 1989).
PILES RESULTS OF PROOF LOADING TESTS
Qe
SITE FOUND. Type * L (m) d (m) n. of
(t)
atQ = Q at Q a I,SQ at s .. s at s=s 1 ,SQ
tests s(mm) 'S(mm) v s(mm) 'Shnm) v Vn Vo

A 1 0 8,0 0,40 12 60 1,89 a, 78 0,41 4,36 2,31 0,53 0,23 0,25

2 D 18,0 0,52 17 120 3,35 1,02 0,30 11,68 8,14 0, 70 0,16 0,19

I
B 3 D 15,0 0,50 16 80 I ,43 0,51 0,36 2, 76 1,07 0,39 0,22 0,20
c 4 B 14,5 0,45 16 60 1,10 0,21 0,19 2,40 0,68 0,28 0,11 0,12
0 5 B 20,0 0,80 22 110 0,92 0,20 0,21 2,03 0,69 0,34 o, 13 0,13
E 6 B 15,0 0,60 9 115 I ,26 0,50 0,40 5,05 2,29 ; 0,59 0,21 0,27
7 B 15,0 0,60 14 115 1,10 0,51 0,46 4,49 2,87 i 0,64 0,20 o, 22
F 8
9
B
B
20,0
18,0
0,40
0,60
19
14
40
75
0,96
0,88
0,46
0,42
0,48
0,48
I ,99
I, 78
0,82
0,82
I 0,41
0,46
0,25
0,29
0,17
0,23
10 B 18,0 0,80 6 ISO 2,10 0,63 0,30 5,56 1,91 0,34 0,17 0,12
11 B 22,0 0,50 9 58 1,64 0,87 0,53 3,61 1,98 I o,ss 0,37 0,28
12 B 22,0 0,80 19 ISO 2,83 0,99 0,35 8,24 4,45 1 o,s4 0,16 0,16
G 13 0 21,5 0,35 18 so 2,34 0,42 0,18 4,03 0,86 0,21 0,19 0,24
H 14 'B 9,0 0,50 22 30 0,25 0,04 o, 17 0,46 0,07 0,15 0,10 0,08
15 B 9,0 0,50 10 60 0,85 0,14 o, 17 1,44 0,16 o, 11 0,11 o,oS
16 B 11,0 0,50 11 60 0,92 0,13 0,14 1,61 0,27 0,17 0,12 0,13
I 17 A 15,5 0,60 6 110 2, 77 0,91 0,33 6,05 2,17 0,36 0,25 0,25
L 18 A 12,6 0,80 10 140 5,40 2,34 0,43 12,71 5,03 0,40 0,21 0,13

:tD = driven pi]("; B == boreU pile~ A== auger pile; Q8 "' service load; S =mean value; S = standard deviation; v 5 =
coefficient of variation of the settlement; vQ = coefficient of variation of the load

From the viewpoint of the design, this kind of results

could engender a serious lack of confidence. The design of
pile foundation is indeed based essentially on an empirical
or semiempirical approach. Any rational analysis is necessa-
rily based on oversimplified models and allows but an eva-
luation of the relative weight of the different factors.
The main role of the theory is actually to provide a
logical framework to arrange the results of the experience,
in order to generalize them and make them available in other
cases.
 Isn't this exactly the scheme of " ••• rational princi-
ples that the mind imposes to the sensible data so that they
can be part of the experience ••• ", that characterizes the
Kantian rationalism?
Even with his feet in the mud of a pile yard, the
geotechnical engineer remains holder of a rational culture!
Foundation piles 255

9. REFERENCES
AGI (1984) Raccomandazioni sui pali di fondazione.
Roma.
BAGUELIN F., JEZEQUEL J.F. (1975) Mesure des elongations
dans les pieux et tirants a l'aide de extensometres
amovibles. Travaux.
CAPUTO V., VIGGIANI C. (1988) Some experiences with
bored and auger piles in Naples area. Symp. on deep
foundations on bored and auger piles, Ghent. Balke-
ma, Rotterdam.
CAPUTO V., GAMBACORTA F., VIGGIANI C. (1989) Pali trivella-
ti di grande diametro nei terreni piroclastici del
Napoletano. XVII Convegno Italiano di Geotecnica,
Taormina.
CAPUTO V., VIGGIANI C. (l989) Intervento alla discussione,
Sessione n.2, XVII Convegno Italiano di Geotecnica,
Taormina.
CARABELLI E., LIBERATI G. (1985) Controlli non distruttivi
dei pali di fondazione. Conf. di Geotecnica di Torino,
XII Ciclo.
CROCE A., PELLEGRINO A. (1967) Il sottosuolo della citta
di Napoli. Caratterizzazione geotecnica del territorio.
VIII Convegno Italiano di Geotecnica, Cagliari.
DIAMANT! R. (1973) Intervento alla discussione, Sessione 1,
XI Convegno Italiano di Geotecnica, Milano.
GHIONNA V., JAMIOLKOWSKI M., LANCELLOTTA R. (1979) Valuta~
zione della capacita portante di un palo di fondazione.
Conf. di Geotecnica di Torino, IX Ciclo.
HERITIER B: (1989) Capacite portante des pieux par charge-
ment dynamique. La methode CEBTP. XII Int. Conf. Soil
Mech. Found. Eng., Rio de Janeiro.
LIZZI F. (1976) Pieux de fondation Fondedile a cellule de
precharge. Construction, n. 6, Paris.
LIZZI F., VIGGIANI C., VINALE F. (1983) Some experiences
with pre-loading cells at the base of large diameter
bored piles. VII Asian Reg. Conf. Soil [Link].,
Haifa.
MARCHETTI s. (1975) Prove di carico verticale statico
su pali con o senza strumentazione. Conf. di Geotecnica
di Torino, VII Ciclo.
MASCARO! c. (1985) Esecuzione e cenni sul dimensionamento
dei pali trivellati con elica continua. Conf. di
256 C. Viggiani

Geotecnica di Torino, XII Ciclo

MEYERHOF G.G. (1976) Bearing capacity and settlement of pile
foundations . Journ. Geot. Eng. Div., [Link], n. GT3.
PRICE G. WARDLE I.F. (1983) Recent developmen ts in pile-
soil' instrumenta tion systems. Field Meas. in Geomecha-
nics, Zurich.
REESE L.C., O'NEILL M.W. (1988) Field load tests on
drilled shafts. Symp. on deep foundations on bored and
auger piles, Ghent. Balkema, Rotterdam.
RIPPA F., VINALE F. (1982) Experiences with CPT in eastern
Naples area. 2nd ESOPT, Amsterdam.
ROCCHI G. , ALBERT L. F. , VACCA 0. , NARDOCCI A. , SALVI
M., MONTINARO N. (1989) Prove di carico strumentate a
rottura su pali di grande diametro trivellati in
sabbia. XVII Convegno Italiano di Geotecnica, Taor-
mina.
SIMONS M. ,H. (1961) Discussion, Session 3b, V Int.
Conf. Soil Mech. Found. Eng., Paris, vol. 3.
SNOW R. (1965) Telltales. Foundations facts, Raymond Inter-
national.
VAN IMPE w. (1984) Pile loading test results on an Atlas
screwpile. 6th Budapest Conf. Soil. [Link] . Eng.
VAN IMPE W. (1988) Considerati on on the auger piles design.
Symp. on deep foundations on bored and auger piles,
Ghent. Balkema, Rotterdam.
VAN WEELE A.F. (1988) Cast in situ piles. Installatio n
methods,so il disturbance and resulting pile behaviour.
Symp. on deep foundations on bored and auger piles,
Ghent. Balkema, Rotterdam.
VIGGIANI c. (1973) Significate e limiti delle teorie e
dei procedimen ti di calcolo per la progettazio ne dei
pali di fondazione. Studi e rendiconti, Corso di Perfe-
zionamento nel1e costruzion i in c.a., Politecnico di
Milano.
VIGGIANI C., VINALE F. (1983) Comportame nto di pali
trivellati di grande diametro in terreni piroclastic i.
Rivista Italiana di Geotecnica, [Link], n.2.
VINALE F. (1988) Caratterizz azione del sottosuolo di
un• area campione di Napoli ai fini di una microzona-
zione [Link] vista Italiana di Geotecnica ,[Link], n. 2.
WHITAKER T. (1976) The design of pile foundations . Pergamon
Press, Oxford.
WRIGHT S.J., REESE L.C. (1979) Constructio n of drilled
shafts and design for axial loading. u.s. Dpt. of
Transporta tion: Drilled Shafts Manual, vol I.
Washington .
 COMPUTER BASED STRUCTURAL ANALYSIS OF BRIDGES

S. Odorizzi
University of Padua, Padua, Italy

SUMMARY

Some applications are presented for the computer based structural

analysis of bridges. The possibility of employing general purpose
computer programs is also discussed, with specific reference to finite
element programs, to get an effective solution of a variety of problems
typical of such structures. The methods here proposed are not at all
exhaustive of the subject; they are aimed to characterize the attitude
with which the user of any actual computer program should face a
particular category of structures.

1. INTRODUCTION.

Most of the general met. hods for the computer based modelling and
analysis of structures and continuous bodies are well established as
far as their theoretical assumptions and analytical developments are
concerned, as well as the features more strictily pertaining to the
practice of their being coded into computer programs.

The applications to particular structures - that is structures

entering categories with respect to the problems brought up and to the
methods of solution - induce two typicaJ attitudes, reflecting two
classes of problems partially complementary.
258 S. Odorizzi

Any structure can be handled by general purpose programs. At the

most the performances of such programs can be improved by writing a
procedure chaining the logical sequence of the operations and reworking
the results, to comply also with the requirements of given standards.
For instance a finite element fully three-dimensional model is commonly
applicable.

On the other hand, and more frequently under condition of

recurrence of a specific problem, a particular program should be
implemented. This is normally quite far from the general procedures for
the structural modelling, as tends to reproduce the hand calculations,
the assumptions and the approximations of which are accepted with the
aim of speeding up the solution, obtaining everyday professional
results.

Fancing to have enough time to study a problem thoroughly, both

the approaches should co-exist, entrusting the designer with the
decision of employing a numerical model when the complexity of the
matter or some conditions laying within the problem spoil the
application of an approximated fixed-schema program.

In this paper the methods and their applications are roughly

classified into three groups.

The first group includes the appl [Link] which are not
implementing numerical procedures established within the methods of
computer based structural analysis, such as those programs - directly
written by the user or committed to specialized software houses - which
are reproducing a hand calculation for a given practical purpose. For
instance the problem of drawing from the stress resultants the stress
distribution over a section of a structural member falls into this
category. As these applications do not give rise to any general subject
of discussion, they are not mentioned in what follows.

The second category covers the so-cC~.ll ed semi -numerical or

semi-analytical methods. These formulations are quite general, but the
applications are suitable for particular structures, including most of
the bridges. Here the solution for slab bridges as a series of
continuous plates connected to beams having rotatory and flexural
inertia is presented, as well as the finite prism method.

The third category concerns the numerical methods and, in

particular, the finite elemenL method. A general discussion is proposed
and some details are given on the implementation of a variable
thickness thick plate element and on the geometrically non-linear
analysis of combined beam and cable structures.
Computer based structural analysis 259

2. ANALYSIS OF SLAB BRIDGES AS A SERIES OF CONTINUOUS PLATES CONNECTET

TO BEAMS HAVING ROTATORY ANU FLEXURAL INERTIA.
The application is fully described in references [1] and [2].
Here the discussion is proposed as a useful implementation of the
finite strip method, an extension of which is presented in the next
paragraph.

The method is introduced referring to the following figure 1,

showing a rectangular simply supported bridge deck, with ledgers only
at the ends.

Fig. 1 Schema of the bridge deck.

The slab is divided into strips of constant or linearly variable

thickness according to geometric conditions or position and arrangement
of loads. In the longitudinal direction the stiffness coefficients of
beams and slabs are espressed in Fourier's series, imposing the
continuity and the equilibrium equations to each term of the expansion.
The loads are also described in terms of Fourier's series.
Transversally to the deck the integration is carried out in exact form.
Hence the problem is reduced to a one-dimensional one, and from an
operative point of view any harmonic contribution is similar to the
analysis of a continuous [Link] on elastic supports. The precision
depends only on the number of harmonic terms employed.

Developing the stiffness matrices and the equivalent nodal loads

of the problem the authors start from further assumptions. These are

- that the deck plate is a flat plate, connected to simply supported,

constant cross-section beams, having rotatory and flexural inertia
and rigidly bounded to the end ledgers;
260 S. Odorizzi

- that the ledgers are perfectly rigid in their own plane and without
torsional stiffness;

- that bending stresses and in-plane stresses of the plate are

independent of each other;

- that the loads are transversally constant over any strip;

- that concentrated loads are only in corrispondence of subdivision

lines.

Therefore the position of the longitudinal beams and the

hypotheses on the loads give the conditions for the strip subdivision
of the deck.

In the above references the Levy's solution is adopted for the

plate. The corresponding stiffness matrices can be obtained by formally
expressing each term, and hence imposing to any harmonic a vertical
displacement (fig. 2)

w
n
'¥
on
. (nnx)
s1n - 1- fJn =0 (1)

and a rotation (fig. 3)

fJ
n
= ~on [nnx)
sin--
1
w
n =0 (2)

Such displacement and rotation are on the edge i of the strip, while
the opposite edge J is fully clamped.

z
Fig. 2 Reference schema for the imposed displacement.
Computer based structural analysis 261

z
Fig. 3 Reference schema for the imposed rotation.
The corresponding stress parameters are obtained

2 2 2
n n: sinh2a + a
B
n n (3)
2
sinh2a - a
n n

sinh a cosh a + a
n n n
B a0 n . (nn:x)
Sln --
- 1

2nn: sinh a cosh a - a

n n n
(Mcpn) I B
sinh a - a
n n

2 2 2
n n: sinh 2a + a
n n
(r cpn) 1 B
2
1 sinh2a - a
n n

2n2n:2 a sinh a
n n
(MwJ j = B
2 ron . (nn:x)
s1n - 1-
12 sinh 2a - a
n n
262 S. Odorizzi

B sinh a + a cosh a
_______
n ____ n r
n _______
. [nrrx)
s1n --1-
sinh 2 a - a 2 on
n n

2nrr sinh a - a cosh a

n n n
(MIPJ j = B
2
l sinh 2 a - a
n n

2n 2 rr 2 a sinh a
n n
(T (/Jn) j B
2
12 sinh 2 a - a
n n

where

M is the bending moment distribution,

T is the shear force distribution,
- w refers to the vertical displacement,
- qJ refers to the rotation,
- i refers to the edge i,
- j refers to the edge j,
- n is the harmonic number,

and

Es 3
8 = (4)
12( 1-v 2 )

nrra
a = -1- (5)
n

Hence the stiffness coefficients are determined employing the

virtual work equation.

The equivalent nodal loads are drawn referring to a strip which is

simply supported at the ends and fully constrained along the
longitudinal edges. To this strip the following load is applied

q(x,y) nrrx
q 0 sin --
1- (6)
Computer based structural analysis 263

obtaining after some algebra the equivalent nodal forces.

Then the overall stress state is determined imposing the

equilibrium conditions along each subdivision line. This yields for any
harmonic contribution to a linear system of equations of the form

K w = F (7)
n n n

Once the displacements are obtai ned, the stress parameters can
also be drawn by proportion. Eventually all the harmonic contributions
are summed up.

3. THE FINITE PRISM METHOD.

The finite strip method, an application of which has just been

described combining strips and beams for the analysis of bridge decks,
roots back to references [3] and [4].

The idea is extended to three-dimensional problems as a

combination of finite elements and Fourier's series. Approaches of this
type are well established and have been worked out in quite a remote
past of informatics [5], [6], [7], [8] . Reference [9] deals with simply
supported bridge boxes; reference [10] gives an almost complete review
of the method.

The main advantage of the method has been pointed out in the past
as the possibility of substantially reducing the computational efforts
and the computer core memory requirements, as a consequence of dealing
with a three-dimensional problem through a series of two-dimensional
ones. Nowadays the practical use of the method depends probabily more
on the possibility of having a 'natural' descriprion of problems and
loads.

The methods applies to structures as those in figure 4.

To focus the leading idea of such approach, the deflection of a

simply supported beam (fig. 5) is firstly introduced.
264 S. Odorizzi

~~-----------~~

DDDDDDDDDD lo o o o o o o o o o ol

Fig. 4 Typical structures to which the finite prism method applies.

Fig. 5 Simply supported beam under distributed loading.

The differential equation for a beam in bending is

d4y
q(x) = EJ - ( 8)
dx 4
Computer based structural analysis 265

in which y must satisfy the boundary conditions

y =0 at x =0 and x = (9)

0 at x =0 and x =I

The deflection y can be expanded in Fourier's series as

(10)

where m refers to the harmonic term, and am is the unknown amplitude of

the mth harmonic.

The loading q(x) is defined by the Fouries's series

( 11)

where bm is the amplitude for the harmonic m, and can be obtained using
the Euler's formula

b
m
2f Jq(x)
1
0
• (mnx)
Sln --1-- (12)

Therefore, for a given harmonic of load the problem is one of

finding the unknown amplitude am which uniquely describes the deflected
profile for that harmonic. Substituting equations (10) and (11) into
equation (8) leads to

14 b
m
am = (13)

The final deflected profile can be obtained carrying out the

summation (10) over an adequate number of terms.

This seri~s converges rapidly. For instance the mi~~pan

deflection corresponding to a triangular distributed load is determined
266 S. Odorizzi

with prec1s1on of three digits using only three terms of the series,
and with precision of two digits using only the first term of the
series.

To generalize, reference is made to the prismatic solid of figure

6, which can be also generated by sweeping the section through a
circular arc. According to the displacement approach, the displacements
are described in terms of harmonic functions along the longitudinal
direction, while maintaining the shape functions typical of the finite
element formulation in the cross-section. Therefore the convergence
depends on the number of harmonic terms as well as on the finite
element subdivision.

Fig. 6 Typical prismatic situation: a-straight prism, b-curved prism.

In what follows, to give some details of the method, an

isoparametric finite element formulation is proposed, based on
parabolic shape functions, and simple basic funcions in the
long~tudinal direction.

The displacements for a prismatic body can be expressed as

a
T
= [u,v,w] (14)
Computer based structural analysis 267

where

co n
u = E LNI ulI
1=1 1=1
Sln a-
. (lnz) (15)

co

E LN1
n
v = VI
I
sin(l:z)
1=1 1=1

co n
w = E LNI
1=1 I= 1
WI
I
. enz)
s1n -
a

N1 are the shape functions, expressed in terms of curvilinear local

coordinates E; and 11, n denotes the number of nodes per element and 1
refers to the harmonic number.

Due to the nature of the basic functions, the following boundary

conditions must always apply

dw _
u = v = dz - 0 at z =0 and z =a (16)

These describe a simple support condition such that al the ands of

the prism the in-plane displacements and the axial strains are both
zero.

The stiffness matrix of the prism gives

F = Ko (17)

where

(18)
268 S. Odorizzi

F and a represent respectively the vector of the equivalent nodal

forces and the vector of the unknown displacements for the lth
harmonic.

A typical submatrix of K can be written as

~~ :; = J ~~f ~ ~; dv
v
(19)

where the range of integration is taken over the whole solid volume of
the prism and can be performed numerically.

The strain matrix !3· by introducing the displacement components

(15) becomes

(20)

where

aN 1
0 0
ax
aN l
0 0
8Y
0
-N lhr
0 -
81 a
-l
= (21)
aN aN
l l
0
ay ax
0 0 0

0 0 0
Computer based structural analysis 269

0 0 0

0 0 0

0 0 0
gl = (22)
-l

0 0 0

Nhr
aN l
l- 0
a
ax
Nhr aN 1
0 1-
a ay

A transformation is needed to relate the strain in curvilinear

coordinates ~ and ~ to strain in global coordinates x,y. Such
transformation may be written as

au 1 av 1 aw 1 au 1 av 1 aw 1
a;- a;- ax a~ ~ ~
J-1 (23)
au 1 av 1 aw 1
au 1
av 1
aw 1
ay ay ay a~ 811 811

where J is the jacobian matrix

ax ay
a~ a~
J = (24)
ax ay
aT/ a11

and
270 S. Odorizzi

L
n
X = N1 (~.TJ)X1 (25)
1=1

n
y = L
1=1
N1 (~.TJ)Y1

The D-matrix, g1v1ng the stress-strain relationship, for a

three-dimensional isotropic material can be written as

D =[ ~11 ~
-
l
-22
(26)

where

(1-v) v v 0
E v (1-v) v 0
D
-11
= (l+v) (1-2v) (27)
v v (1-v) 0
0 0 0 (1-v)

E
~22 = (t+v) (1-2v)

assuming the stresses are given the following sequence

0"
y
0"
z
't
xy
't
xz
(28)

Hence equation (19) becomes expllcitely

Computer based structural analysis 271

Bm sin [mnz)
-J a
(29)
i? cos (mnz)
~J a

Since the off-diagonal submatrices of D are zero, the products of

sine and cosine terms vanish, and the above-equation (29) becomes

K1 ,m -
[ -l,j ] e I[ ]TB1
-1
D
-11
-m
~J sin (lnz) . (mnz) dV
a_- s1n a- + (30)

I[~~ r~22 ~
v

cos ( 1:z) cos (m:z) dV

v

Due to the orthogonal nature of the basic functions, there are

further simplifications

fsin[':" sin [•:z

0
J J dz =
{~
if

if
h~m

l=m
(31)

f ['""] [""]
0
cos a sin a· dz =
{;
if

if
l:;t:m

l=m

and hence

K1,m = 0 for l:;t:m (32)

-I. J

Therefore the overall stiffness matrix becomes diagonal and each

harmonic term decouples from the others

~2 11
K .... ~ .... J (33)
272 S. Odorizzi

A typical submatrix becomes

= ~ I ( [~:r ~11 ~~ [~:r~22 ~~

A
+ J dA (34)

where A represents the cross-sectional area of the prism.

The loads are also resolved into Fourier's series functions,

obtaining for each harminic a vector of equivalent nodal forces. To
make an example, the case of a patch load is considered, starting from
the following expansion

px [ p~ sine:z)

E = py = [ p~ sine:z) (35)

pz [ p: sine:z)

where px and py and pz represent the Fourier coefficients of the load

for the lth harmonic.

The equivalent nodal forces are

sin2 e:z)

sin2 e:z) dz dS (36)

Performing the inner integration and simplifying this reduces to

Computer based structural analysis 273

l
px

I NldS
l
r-l =
a
2
pl
y
(37)
s
1
pz

This integration can be performed numerically over the surface

length of the element on which the load is applied. If the load is
limited by z1 and z2, its amplitude reduces to the form

(38)

Similar procedures can be employed for different loading patterns.

Once the the system of equilibrium equations has been solved for
each harmonic, the overall displacements are obtained by summing all
the contributions.

To extend the-procedure to a circular prism, it is sufficient to

interprete the displacements with respect t.o a cylindrical coordinate
system, replacing the coordinate z by the angle ~ and the span of the
prism by the angle «. The displacements become

L LNl
00 n
= [ln~) (39)
u
l sin«-
u1
1=1 1=1

L LNl v1l sin(1 :~)

00 n
v =
1=1 l =1

L LNl w1l sine:~)

00

w=
1=1 1=1

and the strain components become

T
[au av _! aw + -v au+ av (40)
f:
ax ' 8r ' r a~ r ' ar ox '
274 S. Odorizzi

.!. au + aw .!. av + aw w]
r a1J ax ' r 8{} ai - r
An application of the method is shown in figure 7, where a curved
composite box girder bridge is analysed.
In figure 8 the transverse load distribution is discussed for a
prestressed concrete bridge.

\ c

Fig. 7 Application of the finite prism method to the analysis of a

curved composite box girder bridge.
A - Finite prism discretisation (span of the bridge = 70 m,
radius = 450 m).
8 - Deformed configurations at 1/3 of the span and at the
mid-span.
C - Detail of the [Link] stress field at 113 of the span
and at the mid-span.
Computer based structural analysis 275

TIT . • ~A

Fig. 8 Parametric study of the transverse load distribution for a

prestressed concrete bridge.
A - 3 beams assembly. Influence coefficients of the midspan
deflect ion, corresponding to loads appl led with unit
intensity in turn over each beam.
8 - Same for a 4 beams assembly.
C - Same for a 5 beams assembly.

The finite prism method, as just presented in its essential form,

and the previous application to the analysis of slab bridges as a
series of continuous plates connected to beams having rotatory and
flexural intertia, are formally restricted by the particular boundary
conditions due to the basic functions which are employed. As a matter
of fact, studying a continuous multi-span bridge or a bridge slab with
intermediate ledgers, leads only to formal complications, as such
problems can be solved by a former computaton of the influence
coefficients of the reaction forces on the corresponding displacements
(or points of the intermediate ledgers) and by imposing subsequently
the continuity. On the basis of the superposition criterion, this leads
only to the introduction of some further loading conditions, whose
effects - affected by coefficients drawn by a system of compatibility
equations, and hence different for each external loading condition -
have to be added to the external loading condition which is in turn
under consideration.
276 S. Odorizzi

4. FINITE ELEMENT MODELS . AN APPLICATION OF VARIABLE THICKNESS THICK

PLATES.

Applying generalized finite element model, the behaviuor of a

class of structures, having the characteristics of the one shown in
figure 9, can be effectively investigated taking advantage of
particular thick plate elements of variable thickness .

SIDE VIEW

WI111~1 1111 1111 11111 111 1 1111 R jl •

1

!:!~ I
5[ 2i11111 il l l il

LONGITUDINAL CROSS SECTION

AI sl !!•..a

f: I 1 ·~
"""
, • •'SO

'::: ~
I'

:I' · sl
SECTION A-A SECTION 8 - 8

Fig. 9 Hyghway overpass . Posttensioned concrete slab deck.

 Computer based structural analysis 277

In fact such models are advantageous with respect to fully

three-dimensional models not only depending on the reduction of one
dimensional degree but above all on the possibility of obtaining the
results in a straightforward way for engineering applications.

Plate bending problems had a number of formulations within the

finite element method. To give a reference to some early approaches,
non conforming elements are worked out in [ 11] and [ 12], proving
convergent under most practical conditions and today still used for the
design and analysis of slabs.

Here is shortly outlined the formulation of a thick plate element,

based on the assumptions adopled by Mindlin [ 13]. The element is
derived in details in reference [14], and extended to variable
thickness thick plates in [15] and [16]. The assumptions are : that the
deflections (w) of the plate are small; that the normals to the
midsurface before deformation remain straight but not necessarily
normal to the midsurface after deformation; that the slresses normal to
the midsurface are negligible irrespective of the loading. Therefore
the displacement field can be uniquely specified by an independent
variation of the lateral displacement w and of two angles 11x and fJy
defining the direction of the line originally normal to the midsurface
(fig. 10). In particular 11x and fJy represent average rotations,
including angles lbx and lby, denoting the average shear deformations,
yielding to:

w w

c5 = 11
X
= aw
-
ax + lb
X
(41)
aw
+ lb
(}
y ay y

and

lb =
[~::] (42)
278 S. Odorizzi

Fig. 10 Deformation of the cross-section of a thick plate of

homogeneous material.

Referring to bending moments and shear forces taken as in figure

11, where the x-y plane is taken to coincide with the midsurface of the
plate, the stress resultants are defined by the espresslon

M
X
fa- X
z dz

M= My = fa-y z dz (43)

M f
xy T Z dz
xy

l
and

Q=
[: l = [f ' n dz
f • yz dz (44)

where the integrals are calculated over the range z=-t/2 to z=+t/2, if .
t is the plate thickness.
Computer based structural analysis 279

Fig. 11 Sign convention for bending moments and shear forces.

The total potential energy of the plate may be written as

n = 21 J (~) T ~ + [91 T! dA - Jqw dA (45)

A A

where

ax
8-0
y
(46)
By

- [aiJx _Y]
+ BiJ
ay ax
280 S. Odorizzi

xx, x_y, x_xy are measures of bending deformation, and q is the

distributed load per unit area.

The stress-strain relationships may be written in the following

form

M
-
= D-r-X (47)

and

Q = D~ (48)
- -s-

where for an isotropic homogeneous material

v
D = 1 (49)
-f

and

D
-s
= 2(1
Et
;:;-r.--:-~
v)a: [~ ~] (50)

where a:=B/5 allows for warping of the cross-section. For cellular

plates or sandwich plates, the matrices D can be readily modified for
instance as discussed in [17].

Substituting into (45) one obtains

n = ~ J ([!lr~r! + [!lr~s !l dA- J qw dA (51)

A A

Following standard procedures for isoparametric finite elements, a

typical submatrix of K, linking nodes p and q may be written as

K = K + K (52)
-pq -pqf -pqs

where

K
-pqf ==
JJ [B
-pf
]TD B
-f -qf
dx dy (53)
 Computer based structural analysis 281

and

K (54)
-pqs
=JJ
where

8N l
0 0
ax
aN l
0 0
By

~.=[-!r•-]=
8N 8N
l l (55)
0
8y ax
-sl
-- --- - --- -
8N l
- N 0
l
ax
8N l
0 - N
ay

N are the shape functions, and the 8-matrix is splitted into the part
associated with bending deformations-and the p~·t associated with shear
deformation.

A straigtforward extension of the above element can be obtained

coupling the plate with a membrane element.

The main advantages of the above formulation are:

- the reliability of the finite element, particularly if this is

based on parabolic ~hape functions;

- the plain description of the loads;

- the interpretation of the results directly in terms of stress

resultant parameters and boundary reactions;

- minimum hardware requirements;

- practically negligeable numerical instability.

282 S. Odorizzi

On the contrary there are obviuos restriction s to the possible

applicatio ns, in particular if the element is not coupled with a
membrane one.

An applicatio n is shown for the slab bridge of figure 9. The

finite element mesh adopted and some results are given in figure 12.

Fig. 12 Highway overpass.- Discretiza tion with variable thickness

thick elements of half structure Bending distributio n
correspond ing to a given loading.

5. GENERAL APPLICATIONS OF THE FINITE ELEMENT METHOD.

The previous applicatio ns have been discussed to point out that

tailoring a general approach to a given category of problems leads to a
more direct solution or at least to a solution which is easier to be
used in practice.

However today performanc es of hardware and software systems

address to the use of non specific tools in a number of fields,
including the structural analysis. A general purpose finite element
program for the analysis of the static and dynamic answer of structures
and continuous bodies - characteri sed by linear or geometrica lly or
material non-linear behaviour - can be effectivel y employed for the
Computer based structural analysis 283

majority of problems. Preference is given to the reliability of

programs coming from high credit technical staffs, widely tested and
well established so as to comply with the requirements of the· most
sophisticated modelling, giving also the opportunity to practically
work in an homogeneous computational environment. In fact the time
needed to input the problem data as well as the time needed to read the
results is normally much longer than the time needed to run the
analysis : that phases of the work have then Lo be favoured, trying
also to keep the possibility of errors to a minimum.

Nevertheless there are some areas where general purpose programs

co-exist with specific applications. This happens for instance when a
procedure within a general purpose program is difficult to be
controlled, like in the case of geometrically non-linear analysis of
cable structures. A hypostatic problem arises, where a starting
configuration has to be chosen satisfying the equilibrium and
compatibility conditions, while being achievable in practice. Such type
of formulation is discussed in [ 181, from which the example of a
continuous suspension bridge is taken (fig. 13), as well as the
stability analysis of a bow-string girder (fig. 14).

A 11111111111111111111111111 i~ IY·~~r I ~\Yf! 11111111111111111111111111 SYMMETRIC

TEST LOAD

IOI~IIOI~IIOI~IIcriDni16D(~I?aiii~~DI~ai~IIOI~IIcri~IILI____________________JI ~~~c

w- 4.8 kN/m (cable and pipe)

w- 0.22 kN/m

1 108 1 249 249 1 1o6 1

ELEVATION
/ . 609.5/5

1II I I II II mJj~TP1 ~f1ll1

I I II I I I I II

PLAN

Fig. 13
 284 S. Odorizzi

ELEVATION

PLAN
- UNEAR ELEMENT

~ PARABOUC ELEMENT

c Vr VI

V - horizontal displacement (em)

a - Cable a becomes slack at 1.20 w
b - Cable b becomes slack at 1 .28 w
o~~~~~~-=~~v C - Cable c becomes slack at 2.20 w
200 400 600 BOO (em) d - Cable d yelas at 3.00 w

Fig. 13 Continuous suspension bridge carrying a gas pipeline.

A - Layout and loading.
8 - Finite element model.
C - Load/deflection behaviour (asymmetric wind loading).
Computer based structural analysis 285

a a

i
8

47600
k'i t i i i JT N Qd• Ql• 89.9 kN/m
LOADING 1

c LOADING 2
Ql• 48.72 kN/m
Qd• 41.18 kN/m

Fig. 14 Buckling analysis of a bow-string girden.

A - Elevation.
B - Front view.
C - Finite element discretisation .
286 S. Odorizzi

A general purpose finite element program gives also overall

information on the structure, pertaining to particular features of the
specific problem. For instance one can limit himself to obtaining the
load subdivision in a complex situation. To explain this method the
example of figure 15 is considered. It is a three-span continuous
reinforced concrete bridge, with four variable section longitudinal
T-beams. To find the loading configurations giving any maximum stress
resultant at any sect ion of a given beam a fully three-dimensional
finite element model is employed , drawing the influence coefficients
of the loads on the deflections, for instance for the schemas 1 to 6
applied with unit intensity in turn over an external beam and over an
intermediate one. Combining then with the cross schemas A and 8, the
conditions for the maximum stresses are obtained. Some results of the
analysis are listed in the following table.

Loading Tot. Defection Rei. load Absolute load

cond. trans.
load (tim) T1 T2 T3 T4 T1 T2 T3 T4 T1 T2 T3 T4

C.C.1 12,554 2,248 2,169 1,629 0,973 0,320 0,309 0,232 0,139 4,021 3,879 2.914 1,740

1,316 1,323 1,046 0,687 0,301 0,303 0,239 0,157 2,959 2,974 2,352 1,544
C.C.2 9,629
0,604 0,654 0,498 0,316 0,292 0,316 0,240 0,153 2,865 3,102 2,362 1,499

C.C.3 14,503 1,689 1,689 1,270 0,787 0,311 0,311 0,234 0,145 4,507 4,507 3,389 2,100

0,707 0,704 0,704 0,707 0,251 0,249 0,249 0,251 2,155 2,141 2,141 2,155
C.C.4 8,6
0,497 0,490 0,490 0,497 0,252 0,248 0,248 0,252 2,167 2,133 2,133 2,167

C.C.5 9,010 2,269 1,875 0,968 0,065 0,438 0,362 0,187 0,013 3,950 3,264 1,682 0,113

1,343 1,112 0,548 0,008 0,446 0,369 0,182 0,003 2,882 2,386 1,176 0,017
C.C.6 6,482
0,617 0,560 0,251 -0,010 0,435 0,395 o,1n -0,007 2,812 2,552 1,144 -0,046
Computer based structural analysis 287

'92.50 l 225.00 225.00 225.00

'92.50 l

g
IIi
N
N

C1
~l!!llllllllllllllllllllllll!llllllllllll!llllllll~

14.10 1 11.65 14.50 11.65 14.10 1

C2
~!llllllllll!ll!llllllllllllllll!lllllll~lllllllllllllllllllllllllll!!llllllllllllllllllllt

14.10 1 11.65 14.50 11.65 14.10 1

Fig. 15
288 S. Odorizzi

CJ
!lllllllllllllllllllllllllllllllllllllll~ ~lllllllllllllllllllllllllllllllllllllll!

14.10 1 11.65 14.50 11.65 14.10 1

C4
~llllllllllllllllllllllllllllllllllllll!lllll!ll!@ll!!llllllllllllllllllllllllllllllll!lllllllllllllllllllllllllllllllllllllll~

14.10 '
11.65 14.50 11.65 14.10 1
C5
~IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIHIIIIIIIIIIIIIIIIII!
~ ~

14.1o 1 11.65 14.50 11.65 14.1o 1

C6
~lll!lllllllllllllllllllllllllllllllllll~lllllllllllllllllllllllllllllllllllllllllllllllll~

14.10 1 11.65 14.50 11.65 14.1o 1

01

[Link] 1 350

02
q1f ~
~1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
[Link] 1 350

Fig. 15 Three span continuous reinforced concrete bridge.

A - Longitudinal cross section.
B - Transversal cross section.
C - Longitudinal loading patters.
D- Transversal loading·pattern.
E - Finite. element discretization.
Computer based structural analysis 289

The design of any section is then reduced to the standard one for
a T-beam. It is remarkable that, being the model employed only for the
search of deflections, the number of elements is very small: in fact
the solution is within the deformation approach. Hence the application
is highly effective.

On the other hand if a fully three-dimensional model is meant to

work out the stress state and the material is non-homogeneous, the
outcoming needs to be interpreted. For instance a three-dimensional
model of the highway overpass of figure 9 (solved above with thick
plate elements) gives graphical results in terms of stress contours or
contours of combinations of stress values (fig. 16).

Fig. 16 Highway overpass. Contour of the shear stresses over a section

at the intermediate support.

In conclusion the main difference between the use of a specific

program and the use of a general purpose program is that in the first
case the solution is conditional on a fixed model, which is assumed to
be representative for both the preliminary calculation and the final
design, while in the second case the modelling is fully free and can
each time suit the design concept and the desired degree of precision.
To make an example reference is made to the problem of figure 17,
relating to a posttensioned concrete contiunuous slab bridge of three
spans in a bend. To detail the corresponding finite element model to
the limits of the method, a thr·ee-dimenslonal discret isat ion should be
290 S. Odorizzi

employed, including the actual path of the tendons. The nodal forces
equivalent to the prestressing force over any element can be computed
as described in ( 19]. There are but many theoretical and practical
reasons against such approach. From the practial point of view the
model comes out to be complicated to be managed - in particular in the
preliminary design phase, where changing the path of a tendon could
lead to rework the model completely -, the computation is heavy - with
consequences which are not negligible, especially in the handling of
the results -, and the results are difficulty be interpreted. From the
theoretical point of view, even if the program provides for the
elastic-viscoplastic behaviour of three-dimensional continua, the
experimental data are inadequate to lay down a three-axial
characteristic relationship for stresses and strains and for their
gradients . Following the above method leads rather to an investigation
on the research side than to obtaining effective information for the
design. A more direct and useful model is based on shell elements,
combining the plate bending effects and the in-plane membrane stresses,
and coded with non-linear constitutive relationships for layered
materials. If this is the model, then the path of the Lendons must be
described material ising at each point their [Link] ty with respect
to the mid-plane of the plate. Figure 17 shows such type of
application.

k:. A

Fig. 17
Computer based structural analysis 291

Fig. 17 Posttensioned concrete continuos slab bridge of three spans in

a bend.
A- Finite element model of the slab.
B - Finite element model of the path of the tendons.
C - Prestressing force Deformed configuration.
D - Dead load - Deformed configuration.
E - Superposition of C and D - Deformed configuration.
292 S. Odorizzi

A further example (fig 18) illustrates the versatility of a

general purpose program. It is an aluminium truss bridge with an upper
orthotropic deck plate. The bridge has been designed to be employed
provisionally over variable spans, which are obtained assembling a
given number of elementary parts. The model reflects the same
characteristics. Shell elements are employed for the deck plate, taking
into account an orthotropic constitutive relationship, which has been
obtained working out the load/displacement law on a detailed model of a
portion. Then a mixed beam and shell model is set up for any elementary
part of the bridge and for the end structures. Each one of these is
handled as a substructure of a chain structure of variable size, the
analysis of which becomes straigthforward both with respect to static
and dynamic loads.

Fig. 18
Computer based structural analysis 293

Fig. 18 Aluminium truss bridge with an upper orthotopic deck plate.

A - Finite element model of an intermediate component.
8 - Finite element model of an end component.
C - Overall assembled structure .
D - Dead-load - Deformed configuration.
E - First free vibration frequence .
294 S. Odorizzi

In favour of the appl iacat ion of a general purpose program, to

work in an homogeneous computational environment, a practice normally
takes place reducing the difference towards a specific program. In
fact, whenever the application repeats itself, it turns out to be
natural to code within a general purpose program a procedure for the
automatic mesh generation and handling of the results, fixing once and
for all a satisfactory model and the logical puth of the use, including
the references to specific standards. On the actual stage this is
probabily the most professional attitude. By the way the programs we
are talking about make the internal coding of a procedure quite easy,
allowing the different functions of a process to be directly addressed
and stored in a journal file to be accessed afterwards for new
applications.

6. CONCLUSIONS.

Some applications have been presented of the computer based

structural analysis of given classes of bridges. Some other
applications, while being coded for different specific purposes, suit
quite well a number of cases. Eventually the general purpose finite
element programs free the modelling of any structure, and hence of the
bridge structures. The outline which is here proposed is not at all
exhaustive of the subject nor gives definite categories. However the
attitude is indirectly described with which the user of any actual
computer program should face this kind of problems to get an effective
solution in practice.

REFERENCES

1. Casellato A., Vescovi U. , Vi taliani R., CALCOLO DI SOLETTE DA

PONTE COME SERlE DI LASTRE CONTINUE COLLEGATE A TRAVI DEFORMABILI
PER FLESSIONE E TORSIONE, Giornale del Genio Civile, ott-die. 1976,
411-422.

2. Casellato A., Vitaliani R., Vescovi U., STRESS ANALYSIS OF BRIDGE

SLABS WITH MICROCOMPUTERS, Proc. First Int. Conf. on Engineering
Software for Microcomputers, Venezia, 2-5 aprile 1984, 119-133.

3. Cheung Y. K., THE FINITE STRIP METHOD IN THE ANALYSIS OF ELASTIC

PLATES WITH TWO ENDS SIMPLY SUPPORTED, Proc. Inst. Civ. Eng. 40,
1968, 1-7.

4. Cheung Y. K., FOLDED PLATE STRUCTURES BY THE FINITE STRIP METHOD,

Proc. A.M. S.C. E., 94, 1969, 2963-2979.
Computer based structural analysis 295

5. Grafton P.E., Strome P.R., ANALYSIS OF AXISYMMETRIC SHELLS BY THE

DIRECT STIFFNESS METHOD, J.A. I.A. A. 1 1963, 2342-2347.

6. Ahmad S., Irons B. M. , Zienkiewicz 0. C. , CURVED THICK SHELL AND

MEMBRANE ELEMENTS WITH PARTICULAR REFERENCE TO AXISYMMETRIC
PROBLEMS, Proc. 2nd Conf. on Matrix Methods in Structural
Mechanics, Wright-Patterson Air Force Base, Ohio, 1968.

7. Loo Y. C., Cusens A. R., A REFINED FINITE STRIP METHOD FOR THE
ANALYSIS OF ORTHOTROPIC PLATES, Proc. I.C.E., 48, 1971, 85-91.

8. Deist F. H., AN EXTENSION OF THE TWO-DIMENSIONAL FINITE ELEMENT

METHOD TO REPETITIVE THREE-DIMENSIONAL SITUATIONS, The South
African Mechanical Engineer, 140, 1969.

9. Zienkiewicz 0. C., Too J. J. M., THE FINITE PRISM IN THE ANALYSIS

OF THICK SIMPLY SUPPORTED BRIDGE BOXES, Proc. Inst. C. E., 53,
1972, 147-171.

10. Too J.J.M., TWO DIMENSION, PLATE, SHELL AND FINITE PRISM
ISOPARAMETRIC ELEMENTS AND THEIR APPLICATIONS, Ph.D. Thesis,
C/PH/14/71, Swansea 1971.

11. Zienkiewicz O.C., Cheung Y.K., THE FINITE ELEMENT METHOD FOR
ANALYSIS OF ELASTIC ISOTROPIC AND ORTHOTROPIC SLABS, Proc. Inst.
Civ. Eng. 28, 1964, 471-488.

12. Cheung Y.K., King I.P., Zienkiewicz O.C., SLAB BRIDGES WITH
ARBITRARY SHAPE AND SUPPORT CONDITIONS : A GENERAL METHOD OF
ANALYSIS BASED ON FINITE ELEMENTS, Proc. Inst. Civ. Eng. 40,
1968, 9-36.
13. Mindlin R.D., INFLUENCE OF ROTATORY INERTIA AND SHEAR ON FLEXURAL
MOTIONS OF ISOTROPIC ELASTIC PLATES, J. Appl. Mech. 18, 1951,
31-38.
14. Hinton E., Owen D.R.J., FINITE ELEMENT PROGRAMMING, Academic Press,
1977.

15. Hinton E., Schrefler B. A., Natali A., A FINITE ELEMENT METHOD
FOR VARIABLE THICKNESS PLATES, Proc. XXII ICES Conf., McGill
University, Montreal, 1979.

16. Natali A., Odorizzi S., VARIABLE THICKNESS PLATES BOUNDED TO

AN ELSTIC ISOTROPIC HALF SPACE : A FINITE ELEMENT APPROACH, Proc.
XXI ICES Conf., Padova, 1978.

17. Hinton E., Razzaque A., Zienkiewicz O.C., Davies J.D., A SIMPLE
FINITE ELEMENT SOLUTION FOR PLATES OF HOMOGENEOUS, SANDWICH AND
CELLULAR CONSTRUCTION, Proc. lnst. Civ. Eng. 59, 1975, 43-46.
296 S. Odorizzi

18. Schrefler B. A., Odorizzi S. , Wood R. D., A TOTAL LAGRANGIAN

GEOMETRICALLY NON-LINEAR ANALYSIS OF COMBINED BEAM AND CABLE
STRUCTURES, Computer and Structures, 17, 1983, 115-127.
19. Majorana C., Odorizzi S., Vitaliani R., ANALISI AD ELEMENT!
FINITI 01 MEMBRATURE TRIDIMENSIONALI PRESOLLECITATE, A
COMPORTAMENTO ELASTO-VISCOPLASTICO, La Prefabbricazione, sett.
1984, 583-594.