FINAL

ASSIGNMENT / RESEARCH
NO.7
IN
PRE-STRESSED CONCRETE DESIGN

SUBMITTED BY:
CASINAS, SHAIRA MAE A.
BSCE-V
MWF (9:30-10:30)

SUBMITTED TO:
ENGR. EVANGELINE CONSTANTINO

1
 Analysis of Continuous Prestressed

Concrete Beams

1 Introduction
This conference is devoted to the development of structural
analysis rather than the strength of materials, but the
effective use of prestressed concrete relies on an
appropriate combination of structural analysis techniques
with knowledge of the material behaviour. Design of
prestressed concrete structures is usually left to specialists;
the unwary will either make mistakes or spend inordinate
time trying to extract a solution from the various
equations.
There are a number of fundamental differences between
the behaviour of prestressed concrete and that of other
materials. Structures are not unstressed when unloaded;
the design space of feasible solutions is totally bounded; in
hyperstatic structures, various states of self-stress can be
induced by altering the cable profile, and all of these factors
get influenced by creep and thermal effects. How were
these problems recognised and how have they been
tackled?
Ever since the development of reinforced concrete by
Hennebique at the end of the 19th century (Cusack 1984), it
was recognised that steel and concrete could be more
effectively combined if the steel was pretensioned, putting
the concrete into compression. Cracking could be reduced,
if not prevented alto- gether, which would increase stiffness
and improve durability. Early attempts all failed because the
initial prestress soon vanished, leaving the structure to be-

2
have as though it was reinforced; good descriptions of these
attempts are given by Leonhardt (1964) and Abeles (1964).
It was Freyssinet’s observations of the sagging of the
shallow arches on three bridges that he had just completed
in 1927 over the River Allier near Vichy which led directly to
prestressed concrete (Freyssinet 1956). Only the bridge at
Boutiron survived WWII (Fig 1). Hitherto, it had been
assumed that concrete had a Young’s modulus which
remained fixed, but he recognised that the de- ferred
strains due to creep explained why the prestress had been
lost in the early trials. Freyssinet (Fig. 2) also correctly
reasoned that high tensile steel had to be used, so that
some prestress would remain after the creep had occurred,
and also that high quality concrete should be used, since
this minimised the total amount of creep. The history of
Freyssinet’s early prestressed concrete work is written
elsewhere (Grote and Marrey 2000).

3
Figure 1: Boutiron Bridge, Vichy
Figure 2: Eugene Freyssinet
 At about the same time work was underway on creep at
the BRE laboratory in England ((Glanville 1930) and (1933)).
It is debatable which man should be given credit for the
discovery of creep but Freyssinet clearly gets the credit for
successfully using the knowledge to prestress concrete.
There are still problems associated with understanding
how prestressed con- crete works, partly because there is
more than one way of thinking about it. These different
philosophies are to some extent contradictory, and certainly
confusing to the young engineer. It is also reflected, to a
certain extent, in the various codes of practice.
Permissible stress design philosophy sees prestressed
concrete as a way of avoiding cracking by eliminating tensile
stresses; the objective is for sufficient compression to
remain after creep losses. Untensioned reinforcement,
which attracts prestress due to creep, is anathema. This
philosophy derives directly from Freyssinet’s logic and is
primarily a working stress concept.
Ultimate strength philosophy sees prestressing as a way
of utilising high ten- sile steel as reinforcement. High
strength steels have high elastic strain capacity, which could
not be utilised when used as reinforcement; if the steel is
preten- sioned, much of that strain capacity is taken out
before bonding the steel to the concrete. Structures
designed this way are normally designed to be in compres-
sion everywhere under permanent loads, but allowed to
crack under high live load. The idea derives directly from
the work of Dischinger (1936) and his work on the bridge at
Aue in 1939 (Sch¨onberg and Fichter 1939), as well as
that of Finsterwalder (1939). It is primarily an ultimate load
 concept. The idea of par- tial prestressing derives from
these ideas since the addition of quite significant amounts
of untensioned reinforcement does not alter the logic
(Emperger 1939).
The Load-Balancing philosophy, introduced by T.Y. Lin,
uses prestressing to counter the effect of the permanent
loads (Lin 1963). The sag of the cables causes an upward
force on the beam, which counteracts the load on the
beam. Clearly, only one load can be balanced, but if this is
taken as the total dead weight, then under that load the
beam will perceive only the net axial prestress and will have
no tendency to creep up or down.
These three philosophies all have their champions, and
heated debates take place between them as to which is the
most fundamental.

2 Section design
From the outset it was recognised that prestressed concrete
has to be checked at both the working load and the ultimate
load. For steel structures, and those made from reinforced
concrete, there is a fairly direct relationship between the
load capacity under an allowable stress design, and that at
the ultimate load under an ultimate strength design. Older
codes were based on permissible stresses at the working
load; new codes use moment capacities at the ultimate
load. Different load factors are used in the two codes, but a
structure which passes one code is likely to be acceptable
under the other.
For prestressed concrete, those ideas do not hold, since the
structure is highly
 Figure 3: Load deflection curve

stressed, even when unloaded. A small increase of load can

cause some stress limits to be breached, while a large
increase in load might be needed to cross other limits. The
designer has considerable freedom to vary both the working
load and ultimate load capacities independently; both need
to be checked.
A designer normally has to check the tensile and
compressive stresses, in both the top and bottom fibre of
the section, for every load case. The critical sections are
normally, but not always, the mid-span and the sections
over piers but other sections may become critical when the
cable profile has to be determined.
The stresses at any position are made up of three
components, one of which normally has a different sign
from the other two; consistency of sign convention is
essential.
If P is the prestressing force and e its eccentricity, A and Z
are the area of the cross-section and its elastic section
modulus (top or bottom fibres),M while M is the applied
moment, then
Pe
P
ft ≤ + − ≤ fc (1)
A Z Z
where ft and fc are the permissible stresses in tension and
compression.
Thus, for any combination of P and M , the designer
already has four in- equalities to deal with.
The prestressing force differs over time, due to creep
losses, and a designer is usually faced with at least three
combinations of prestressing force and moment;
• the applied moment at the time the prestress is first
applied, before creep losses occur,
• the maximum applied moment after creep losses, and
• the minimum applied moment after creep losses.
Other combinations may be needed in more complex
cases. There are at least twelve inequalities that have to be
satisfied at any cross-section, but since an I-section can be
defined by six variables, and two are needed to define the
 Figure 4: Gustave Magnel

prestress, the problem is over-specified and it is not

immediately obvious which conditions are superfluous. In
the hands of inexperienced engineers, the design process
can be very long-winded. However, it is possible to separate
out the design of the cross-section from the design of the
prestress. By considering pairs of stress limits on the same
fibre, but for different load cases, the effects of the
prestress can be eliminated, leaving expressions of the
form:-
Moment Range
Z≥
Permissible stress range
(2)
These inequalities, which can be evaluated
exhaustively with little difficulty,
allow the minimum size of the cross-section to be determined.
 Once a suitable cross-section has been found, the
prestress can be designed using a construction due to
Magnel (Fig. 4). The stress limits can all be rearranged into
the form:-
Z 1
e ≥ − + (f Z + M ) (3)
A P
By plotting these on a diagram of eccentricity versus the
reciprocal of the prestressing force, a series of bound lines
will be formed. Provided the inequal- ities (2) are satisfied,
these bound lines will always leave a zone showing all
feasible combinations of P and e. The most economical
design, using the min- imum prestress, usually lies on the
right hand side of the diagram, where the design is limited
by the permissible tensile stresses.
 Figure 5: Magnel diagram

Plotting the eccentricity on the vertical axis allows direct

comparison with the cross-section, as shown in Fig. 5.
Inequalities (3) make no reference to the physical
dimensions of the structure, but these practical cover limits
can be shown as well.
A good designer knows how changes to the design and
the loadings alter the Magnel diagram. Changing both the
maximum and minimum bending moments, but keeping the
range the same, raises and lowers the feasible region. If the
moments become more sagging the feasible region gets
lower in the beam. In general, as spans increase, the dead
load moments increase in proportion to the live load. A
stage will be reached where the economic point (A on Fig.
5) moves outside the physical limits of the beam; Guyon
(1951a) denoted the limiting condition as the critical span.
Shorter spans will be governed by tensile stresses in the two
extreme fibres, while longer spans will be governed by the
limiting eccentricity and tensile stresses in the bottom fibre
(assuming sagging bending). However, it does not take a
large increase in moment for point B to move outside the
cover limit, at which point compressive stresses will govern
in the bottom fibre under maximum moment.
Only when much longer spans are required, and the
feasible region moves as far down as possible, does the
structure become governed by compressive stresses in both
fibres.

3 Continuous beams
The design of statically determinate beams is relatively
straightforward; the engineer can work on the basis of the
design of individual cross-sections, as outlined above. A
number of complications arise when the structure is
indeter- minate which means that the designer has to
consider, not only a critical section,
 Figure 6: Luzancy Bridge

but also the behaviour of the beam as a whole. These are

due to the interac- tion of a number of factors, such as
Parasitic Moments, Creep, Temperature effects and
Construction Sequence effects. It is the development of
these ideas which forms the core of this paper. The
problems of continuity were addressed at a conference in
London (Andrew and Witt 1951). The basic principles, and
nomenclature, were already in use, but to modern eyes
concentration on hand analysis techniques was unusual,
and one of the principle concerns seems to have been the
difficulty of estimating losses of prestressing force.

3.1 Secondary Moments

A prestressing cable in a beam causes the structure to
deflect. Unlike the statically determinate beam, where this
motion is unrestrained, the movement causes a
redistribution of the support reactions which in turn induces
additional moments. These are often termed Secondary
Moments, but they are not always small, or Parasitic
Moments, but they are not always bad. They are normally
denoted by M2.
Freyssinet’s bridge across the Marne at Luzancy, started
in 1941 but not completed until 1946, is often thought of as
a simply supported beam, but it was actually built as a two-
hinged arch (Harris 1986), with support reactions adjusted
by means of flat jacks and wedges which were later
grouted-in (Fig. 6). The same principles were applied in the
later and larger beams built over the same river.
Magnel built the first indeterminate beam bridge at Sclayn, in
Belgium (Fig.
7) in 1946. The cables are virtually straight, but he adjusted
the deck profile so that the cables were close to the soffit
near mid-span but above the centroidal axis at the internal
support (Magnel 1951). Even with straight cables the sag-
 Figure 7: Sclayn Bridge

ging secondary moments are large; about 50% of the

hogging moment at the central support caused by dead and
live load.
The nomenclature for dealing with these reactant
moments was quickly es- tablished (Guyon 1951b). The
designer needed to distinguish between the actual location
of the cable profile (es), and its apparent position, known as
its line of thrust (ep). The two profiles differ by M2/P . The
cable profile has to fit within the section and is subject to
cover constraints, but it is the line of thrust which has to be
used in the stress calculations.
A good designer can exploit this freedom, but it is also
the cause of problems; the secondary moments cannot be
found until the profile is known but the cable cannot be
designed until the secondary moments are known. Guyon
(1951b) introduced the concept of the concordant profile,
which is a profile that causes no secondary moments; es and
ep thus coincide. Any line of thrust is itself a concordant
profile.
 The designer is then faced with a slightly simpler
problem; a cable profile has to be chosen which not only
satisfies the eccentricity limits (3) but is also concordant.
That in itself is not a trivial operation, but is helped by the
fact that the bending moment diagram that results from
any load applied to a beam will itself be a concordant
profile for a cable of constant force. Such loads are termed
notional loads to distinguish them from the real loads on the
structure. Superposition can be used to progressively build-
up a set of notional loads whose bending moment diagram
gives the desired concordant profile. The question of
whether such a profile can be found will be addressed
later.
Inexperienced designers often stop at this stage of the
design process, but they are clearly then not making full
advantage of the power of prestressing, and their structures
are often uneconomic. Guyon pointed out that it was
possible to move the cable profile by means of a linear
transformation in such a way that the line of thrust remains
unchanged. Having chosen a suitable concordant profile
(which remains as the line of thrust), the designer can then
alter the profile by linear transformations until a suitable
cable profile is achieved. This freedom
allows the engineer to choose lines of thrust which lie outside
the cross-section. Experienced designers often claim that they
“do not bother” with concordant profiles; they simply use
their judgement to choose a secondary moment that they
expect to obtain in the particular structure they are designing.
They can then use modified forms of the eccentricity
equations (3) which allow them to
produce limits on the actual cable profile directly:-
Z 1
es ≥ − + (f Z + M + M2) (4)
A P
A slightly different problem now has to be solved; how to
find a cable profile that not only satisfies the local
eccentricity limits but also generates the required value of
M 2.
The two methods are, in fact, directly equivalent, since a
profile that satisfies one set of constraints will
automatically satisfy the other.

3.2 Analysis to determine M2

The calculation of the secondary moments M2 or the
determination of the line of thrust ep, which are equivalent,
can be done in several ways. The development of these
methods reflects changes elsewhere in analysis techniques,
and of course the adoption of computer techniques. At the
1951 conference, for example, Guyon proposed the method
of nodal points (a variation of the method of fixed points)
(Guyon 1951b); as with most methods at the time the
objective was to minimise the number of simultaneous
equations to be solved. The most common method in use
today is to determine the forces that the cable exerts on the
concrete and then to analyse the beam under those loads.
This can be done at several levels of detail; if the structure
is analysed as a beam the resulting bending moment will be
the total effect of the cable (= Pe p ) and the reactions will
be those that contribute to the secondary moments M2. If a
more detailed method, such as a finite element analysis, is
used, the distribution of cables across the width can be
determined, as can local effects of the cable profile
distribution, which can be particularly important with cables
which have significant horizontal curvature.
Alternatively, for continuous beams, use can be made of
virtual work to derive a set of equations from which the
secondary moments can be determined directly in terms of
the actual cable profile (see, for example, (Burgoyne 1988)).
This is not usually much easier than the cable force method,
but it does allow analytical formulations to be developed
which can be used to derive further theories.

3.3 Existence of line of thrust?

Experienced engineers adopt their own strategies for
designing complex struc- tures. Low (1982), for example,
showed that there were limits on the minimum prestressing
force that is needed in a continuous beam. One minimum
limit was derived directly from the Magnel diagram, while
another considered the
range of eccentricities that have to be allowed for between
the maximum sag- ging region at mid-span and the
maximum hogging region over the piers. If the prestressing
force is not high enough, the eccentricity range is too large
and it is impossible to find a secondary moment that leaves
the cable profile always inside the structure.
But Low also showed that there was a third limit on the
prestressing force which had to be satisfied before a
solution could be found, which he called the “third
equation”. It was later shown (Burgoyne 1988) that this limit
related to the existence of the line of thrust. If the
prestressing force is too low, then a cable placed anywhere
between the upper or lower limits on the cable profile will
give secondary moments of the same sign. Under these
conditions, no concordant profile can exist, so the designer
will never be able to find a satisfactory solution without
increasing the prestressing force. By satisfying Low’s third
limit, the designer is assured that a valid profile exists, even
if it still has to be found.

3.4 Applicability of plastic theory

Prestressed concrete beams are normally checked for
ultimate moment capacity, but that is not the same thing as
saying that plastic theory can be used to design such beams.
Plastic theory can only be used if prestressed concrete
structures have sufficient ductility to allow redistribution of
bending moments as hinges form.
La Grange conducted a study on indeterminate beams
and frames, where he concluded that indeterminate
prestressed concrete structures attained a load at failure
which was just below that predicted by full plastic theory (La
Grange 1961). The small discrepancy was due to the post-
peak softening that occurs in prestressed concrete
structures, so that in order to allow the full set of plastic
hinges to develop, the first hinges undergo some reduction
in moment capacity. However he concluded that the
difference was, in practical terms, negligible.
Prestressing steel is much stronger than normal
reinforcing steel and does not exhibit a well-defined yield
point, so the steel has to have much higher strains before
significant plastic deformation occurs. Some of that strain
capacity is consumed during the act of prestressing but
significant elastic curvatures still have to take place before
yielding can occur. This gives a lower limit on the depth of
the neutral axis at failure.
There is a further complication, because higher strength
steels are typically more brittle than reinforcing steels, with
strain capacities of the order of 3%. Thus designers should
limit the strains that develop in the steel at the ultimate
load, and a limit which is frequently applied is to limit the
additional strain, after prestressing, to 1%. This limit can be
criticised as too low, but it takes account of the fact that the
analysis at the ultimate load uses an average strain along
the tendon, while the strain at crack locations can be
higher. This condition provides an upper limit on the depth
of the neutral axis.
The result of these two conditions is that prestressed
concrete beams can only behave plasticly if they satisfy
relatively narrow limits on the position of the neutral axis,
which in turn provides narrow limits on the section
geometry.
Codes of practice normally aim to force designs into this
narrow band of accept- ability, and they only allow
redistribution if certain conditions on the neutral axis
position are satisfied.

3.5 Secondary moments at the ultimate load?

A closely related question for the designer of continuous
prestressed concrete is whether secondary moments should
be taken into account when the ultimate moment is
calculated. The question is not trivial; secondary moments
can be of the same order of magnitude as the dead-load
bending moments, although distributed in a different way.
The logic behind limit-state codes is to check each
possible failure mecha- nism of the structure, such as
cracking, vibration or collapse. A proper check on the
ultimate limit-state would therefore require determination
of the final collapse mechanism of the structure. When the
final plastic hinge forms the structure becomes a
mechanism, so when the penultimate hinge formed, the
structure must have been statically determinate; secondary
moments do not ex- ist in statically determinate beams. The
moment distribution in the beam can be found purely by
equilibrium considerations which will differ from the elastic
moments by a certain amount of redistribution.
However, in most cases, ultimate moment capacities are
checked on a section- by-section basis by applying factored
values of the elastic load distribution. Some codes make no
mention of secondary moments, but others allow the in-
clusion of M2 in the ultimate load calculation (Mattock
1983). In effect, this condition ensures that the first plastic
hinge forms with a sufficient reserve of strength; up to this
load, the structure has been behaving elastically, so sec-
ondary moments would, indeed, have been present. In a
beam with sagging secondary moments the effect can be to
significantly reduce the ultimate mo- ment capacity that has
to be provided over the piers, and to increase the moment
capacity that is required in the span regions.
There have been some laboratory studies of continuous
beams where the support reactions were measured as loads
were increased until a collapse mech- anism developed. The
magnitudes of these reactions allowed the presence (or
absence) of the secondary moments to be monitored, and
the results showed that the secondary moments
disappeared as the final hinge formed. The plastic hinges
did not form suddenly, but slowly developed through an
elasto-plastic regime (Mattock, Yamazaki, and Kattula
1971).
The existence of secondary moments is an academic
question if the structure is ductile, since the moment
distribution, with or without M2, satisfies the Lower Bound
Theorem if the structure is provided with adequate rotation
capacity, but codes do not always allow these effects to be
taken into account, or limit the amount of redistribution
that can take place.
3.6 Temperature effects
Temperature variations apply to all structures but the effect
on prestressed concrete beams can be more pronounced
than in other structures. The tem- perature profile through
the depth of a beam (Emerson 1973) can be split into three
components for the purposes of calculation (Hambly 1991).
The first causes a longitudinal expansion, which is normally
released by the articulation of the structure; the second
causes curvature which leads to deflection in all beams and
reactant moments in continuous beams, while the third
causes a set of self-equilibrating set of stresses across the
cross-section.
The reactant moments can be calculated and allowed-for,
but it is the self- equilibrating stresses that cause the main
problems for prestressed concrete beams. These beams
normally have high thermal mass which means that daily
temperature variations do not penetrate to the core of the
structure. The result is a very non-uniform temperature
distribution across the depth which in turn leads to
significant self-equilibrating stresses. If the core of the
structure is warm, while the surface is cool, such as at night,
then quite large tensile stresses can be developed on the
top and bottom surfaces. However, they only penetrate a
very short distance into the concrete and the potential crack
width is very small. It can be very expensive to overcome
the tensile stress by changing the section or the prestress,
and they are normally taken into account by the provision
of a mesh of fine bars close to the surface.
A larger problem can arise if thermal stresses act as a
trigger for more damaging cracking, such as the release of
locked-in heat of hydration effects which can occur when a
thick web is associated with thin slabs.

3.7 Construction sequence effects

Prestressed concrete tends to be used for the longer-span
bridge structures, which often means that they are built
sequentially. As a result, the bending moments at the end of
construction differ from those which would be expected if
the bridge had been built in one go (the monolithic moment
). As an example, balanced cantilever construction builds
out from a central pier, so the structure is inevitably in
hogging bending throughout. When the tips of two
cantilevers meet they are joined by an in-situ stitch, or
sometimes by a short suspended span that is usually made
fully continuous. The cantilever will have prestressing cables
at the top to resist hogging bending, while continuity cables
will be introduced across the joint to resist the sagging
bending that will occur later.
The designer has to allow for the temporary condition,
and also for the trapped moments that are induced by the
construction sequence. These trapped moments can be
large, and obey the same rules as the secondary moments,
in that they are brought about by a redistribution of the
dead-load support reactions. The designer may deliberately
choose to use the continuity cables to induce a secondary
moment that reduces the trapped moment.
Further trapped moments can be induced by the use of
temporary prestress- ing cables which are introduced when
the structure is in one configuration, and
then removed later after the support conditions have
changed. For example, in span-by-span construction, where
a long viaduct is built one span at a time, it is sometimes
necessary to introduce temporary cables to resist sagging
bend- ing moments that occur during construction but
which will be removed later. Putting a cable into a two-span
structure (for example), and then removing it once the
structure is more indeterminate, does not leave a zero
stress state; these effects should not be overlooked.

3.8 Creep effects

The final effect that needs to be considered is, appropriately
enough, creep (Bazant and Wittmann 1982). It was
Freyssinet’s original observation of creep that made
prestressed concrete possible since he managed to reduce
the loss of force caused by creep. In simply supported
beams creep causes some loss of prestress and increased
deflections, which may need to be taken into account, but it
does not alter the distribution of bending moments so the
design remains relatively straightforward.
If the structure is indeterminate there is always the
possibility that the bending moments may be altered by
redistribution of the support reactions. If the structure is
built in one piece, all the concrete will be of the same age,
and its effective modulus will change uniformly throughout
the structure. No redistribution of forces is to be expected
under these circumstances.
However, if the concrete is of different ages, the amount
of creep that can occur in the various parts of the structure
will vary, which allows redistribution of moments. It is now
well-established that the structure will creep towards the
monolithic state, and the designer can take the as-built
condition (includ- ing trapped moments) and the monolithic
state as limiting conditions for the behaviour of the beam.
This simplifies the design process.
England has studied the effect of temperature variation
through the depth of the beam. Creep is temperature
dependent and takes place more quickly on the warmer side
of a structure than on the colder side, which can
significantly alter the load distribution. This work was
originally applied to nuclear reactor containment vessels,
where the temperature variation across the thickness can
be of the order of 100◦C (England, Cheng, and Andrews
1984). The work makes use of the concept of a steady-state,
when creep can continue but without redistribution of
stress. More recently, it has been shown that the much
smaller temperature variations that can be expected
through the depth of a bridge deck, which may be of the
order of 5◦C, can also have a significant effect. The speed
with which creep occurs is very heavily dependent on the
relative ages of the concrete in different parts of the
structure (Xu and Burgoyne 2005).

4 Conclusion
The successful design of continuous prestressed concrete
beams cannot be di- vorced from the techniques used to
analyse the structure, and the way these have
developed in the 60 years since the first indeterminate
structures were built is a fascinating reflection on the way
structural analysis has developed over the same period.
It remains the case that designers cannot blindly use
analysis programs with- out fundamental understanding of
the way prestressed concrete behaves.

Acknowledgements
I am grateful to Robert Benaim for his helpful comments on
an earlier draft of this paper.
Figures 4 and 7; Prof Taerwe, University of Ghent. Figure
6; Jacques Mossot (www.structurae.de).

References
Abeles, P. W. (1964). An introduction to prestressed
concrete. London: Con- crete Publications.
Andrew, R. P. and P. Witt (Eds.) (1951). Prestressed
Concrete Statically Indeterminate Structures. Cement
and Concrete Association.
Bazant, Z. P. and F. H. Wittmann (1982). Creep and
shrinkage in concrete structures. John Wiley.
Burgoyne, C. J. (1988). Cable design for continuous
prestressed concrete bridges. Proc. Inst. Civ. Engrs 85,
161–184.
Cusack, P. (1984). Franc¸ois Hennebique: the specialist
organisation and the success of ferro-concrete: 1892-
1909. Trans. Newcomen Soc. 56, 71–86.
Dischinger (1936). German patent no. 535,440.
Emerson, M. (1973). The calculation of the distribution
of temperature in bridges. Technical report, Transport
and Road Research Laboratory. LR561.
Emperger, X. X. (1939). Stahlbeton mit vorgespannten
Zulagen aus h¨oherwertigem Stahl (Reinforced
concrete with additional high strength steel). Berlin:
Ernst & Sohn.
England, G. L., Y. F. Cheng, and K. R. F. Andrews (1984). A
temperature- creep theory for prestressed concrete
continuum beam structures. Journ. Thermal Stresses 7,
361–381.
Finsterwalder, U. (Sept 1939). Eisenbetontr¨ager mit
selbstt¨atiger vorspan- nung (reinforced concrete
beams with self-acting prestressing). Der Bauinginieur .
Freyssinet, E. (1956). Birth of prestressing. Library
translation 59, Cement and Concrete Association.
Translated from French. Published by Travaux, July-
August 1954.
Glanville, W. H. (1930). Studies in reinforced concrete - III:
The creep or flow of concrete under load. Building
Research Technical Paper 12, Dept of Scientific and
Industrial Research, London. 39 pp.
Glanville, W. H. (1933). Creep of concrete under load. The
Structural Engi- neer 11/2, 54–73.
Grote, J. and B. Marrey (2000). Freyssinet, Prestressing
and Europe 1930- 1945. Paris: E´ ditions du Linteau.
Guyon, Y. (1951a). B´eton Pr´econtraint. Paris: Editions
Eyrolles. Translated as Prestressed Concrete by Harris
A.J. (et al): published London, Con- tractors Record and
Municipal Engineering, 1953.
Guyon, Y. (1951b). A theoretical treatment of continuity in
prestressed con- crete. In R. P. Andrew and P. Witt
(Eds.), Prestressed Concrete Statically Indeterminate
Structures, pp. 131–145. Cement and Concrete
Association.
Hambly, E. C. (1991). Bridge deck behaviour (2 ed.).
London: E & FN Spon. Harris, A. J. (1986). Luzancy
Bridge after 41 years. Concrete Quar-
terly 149/June, 6–7.
La Grange, L. E. (1961). Moment redistribution in
prestressed concrete beams and frames. Ph. D. thesis,
University of Cambridge.
Leonhardt, F. (1964). Prestressed Concrete: Design and
Construction. Berlin: Ernst & Sohn.
Lin, T. Y. (1963). Load balancing method for design and
analysis of pre- stressed concrete structures. Journ.
Amer. Conc. Inst. 60/6, 719–742.
Low, A. M. (1982). The preliminary design of prestressed
concrete viaducts.
Proc. Inst. Civ. Engrs 73, 351–364.
Magnel, G. (1951). Continuity in prestressed concrete. In R. P.
Andrew and
P. Witt (Eds.), Prestressed Concrete Statically
Indeterminate Structures, pp. 77–86. Cement and
Concrete Association.
Mattock, A. H. (1983). Secondary moments and moment
redistribution in ACI 318-77 Code. In M. Z. Cohn (Ed.),
Int. Symp. Nonlinearity and Continuity in Prestressed
Concrete, Volume 3, pp. 27–48. University of Waterloo,
Ontario, Canada.
Mattock, A. H., J. Yamazaki, and B. T. Kattula (1971).
Comparative study of prestressed concrete beams,
with and without bond. ACI Journal 68, 116–125.
Schonberg, M. and F. Fichter (Feb 1939). Die Adolf-
Hitler-Bru¨cke in Aue (Sa) (The Adolf Hitler Bridge at
Aue (Saxony)). Die Bautechnik .
Xu, Q. and C. Burgoyne (2005). Effect of temperature and
construction se- quence on creep of concrete bridges.
Proc. Inst. Civ. Engrs, Bridge Engi- neering (in Press).