CIE4160 Textbook February 2019

CIE3150/4160
Prestressed concrete
February 2019
Faculty of Civil Engineering and
Geosciences
Prof.dr.ir.Dr.-Ing.h.c. J.C. Walraven

dr.ir.drs. C.R. Braam
Foreword
This textbook explains the principles of prestressed concrete and its application. It
contains the background information for the lectures CIE3150/CTB3335 (Concrete
Structures 2) and CIE4160 (Prestressed Concrete).
This book is basically the English translation of the existing book “Voorgespannen
Beton”, that has been used already for a number of years as background material for the
courses.
This second English version is an update from the Dutch code NEN 6720 (VBC) to a
Eurocode (EN 1992-1-1). In spite of all efforts there might be inconsistencies, specially
since (national annexes to) the Eurocodes are regularly updated. The authors are grateful
to everybody who brings errors to their notice and for any constructive criticism.
prof.dr.ir. J.C. Walraven

dr.ir.drs. C.R. Braam
February 2019
Contents
1. Introduction
1.1 The principle of prestressing
1.2 Application of the prestressing force
1.3 Curved tendon profiles
1.4 Prestressed concrete versus reinforced concrete
1.5 Literature
2. Materials and methods of prestressing

2.1 Prestressing steel
2.2 Methods of prestressing
2.2.1 Prestressing with post-tensioned steel, with bond
2.2.2 Prestressing with post-tensioned steel, without bond
2.2.3 Prestressing with pre-tensioned steel
2.3 Points of attention during prestressing
2.4 Important characteristic values of prestressing steel
3. Centrically prestressed structures

3.1 Introduction
3.2 Centrically reinforced element subjected to axial load
3.3 Calculation of the elongation required during prestressing
3.4 Design of a prestressed tensile member
3.4.1 Introduction
3.4.2 Global design
3.4.3 More detailed calculation of the loads
3.4.4 Deformations of the column
3.4.5 Calculation of the elongation of the prestressing steel during
tensioning
4. Prestressing of structures subjected to bending

4.1 Loading by prestressing
4.2 The influence of the geometry of the cross-section
4.3 Prestressing with pre-tensioned steel
4.4 Prestressing with post-tensioned steel
4.4.1 The transmission of the prestressing forces
4.4.2 Prestressing losses caused by elastic deformation during the
prestressing of a member with more than one prestressing tendon
4.4.3 Friction losses
4.4.4 Wedge set / anchorage losses
4.5 The fictitious cross-section
4.6 Choice of the shape and dimensions of the cross-section
4.7 Calculation example of a fully prestressed beam
4.7.1 Introduction
4.7.2 Calculation of the losses due to friction
4.7.3 Situation during prestressing
4.7.4 Situation in the serviceability limit state (SLS)
4.8 Literature
5. Statically indeterminate structures

5.1 Determination of the prestressing load
5.2 Input of equivalent prestressing loads in a frame-analysis program
5.3 Prestressed frames
6. Shrinkage, creep and relaxation

6.1 Introduction
6.2 General concepts
6.2.1 Creep
6.2.2 Shrinkage
6.2.3 Calculation of the creep deformation with the principle of
superposition according to Boltzmann (1876)
6.3 Simplified methods for the calculation of the shrinkage and creep behaviour
of structures
6.3.1 The effective-modulus method
6.3.2 The method of Dischinger
6.3.3 The method of Trost
6.4 Relaxation of the concrete
6.5 Relaxation of the prestressing steel
6.6 Redistribution of stresses in structural parts loaded in compression
6.6.1 Axially loaded, symmetrically reinforced cross-section
6.6.2 Eccentrically loaded, asymmetrical cross-section
6.7 Redistribution of forces in statically indeterminate structures
6.7.1 Redistribution of forces and moments in statically indeterminate
structures
6.7.2 Redistribution of moments caused by creep in the case of a change of
the structural system
6.7.3 Redistribution of moments due to creep in the case of a sudden support
settlement
6.7.4 Effect of a slow support settlement
6.7.5 Summary of the calculation of redistribution of moments and forces
according to Trost
6.8 Literature
7. Bending moment resistance
7.1 Design philosophy
7.1.1 Minimum reinforcement
7.1.2 Rotational capacity
7.1.3 Principles of moment resistance design
7.2 Moment resistance according to the equilibrium method
7.3 Determination of the resistance of statically indeterminate structures
8. Shear
8.1 Introduction
8.2 Shear behaviour of elements not containing prestressing
8.2.1 Reinforced elements without shear reinforcement
8.2.2 Reinforced elements with shear reinforcement
8.3 Prestressed elements without shear reinforcement, serviceability limit state
8.4 Prestressed elements loaded in shear, ultimate limit state
8.4.1 Elements without shear reinforcement
8.4.2 Structures with shear reinforcement, prestressed with straight
prestressing tendons
8.4.3 Structures with shear reinforcement, prestressed with draped tendons
8.4.4 Influence of ducts in the cross-section
8.5 Vertical prestressing of the web
8.6 Literature
9. Crack width control

9.1 Introduction
9.2 Bond between steel and concrete
9.3 Crack formation in a reinforced concrete tensile member
9.4 Long term effects
9.5 The transition point between the crack formation stage and the stabilised
cracking stage
9.6 The effective tensile area around the reinforcement
9.7 Summary of the most important expressions for crack width control in
reinforced concrete
9.8 Some applications
9.9 Crack width control in prestressed concrete
9.9.1 Introduction
9.9.2 Behaviour of a concrete tensile member with reinforcing and
prestressing steel
9.10 Example of crack width control for a combination of prestressing and
reinforcing steel
9.11 Literature
10. Detailing of prestressed concrete structures
10.1 Introduction
10.2 Introduction of prestressing forces
10.2.1 Introduction of prestressing forces through anchors
10.2.2 Introduction of the prestressing force by bonding
10.2.3 Anchorage capacity of pre-tensioned steel
10.2.4 Intermediate anchors
10.3 Effects of prestressing tendon curvature
10.4 Construction joints and joints between elements
10.5 Indirect support of structures
10.6 Coping with deformations caused by prestressing
10.7 Design of a dapped-end beam.
10.8 Literature
11. Partial prestressing

11.1 Introduction
11.2 Definition of partial prestressing
11.3 Applications of partial prestressing
11.4 Economic degree of prestressing
11.5 Design method
11.6 Working example - three-span bridge
11.7 Literature
12. Prestressing with unbonded tendons

12.1 Introduction
12.2 The behaviour of structures with prestressed unbonded tendons
12.3 Serviceability limit states
12.4 Bending moment resistance
12.5 Shear resistance
12.6 Slabs prestressed by unbonded tendons
12.7 Other applications
12.8 Literature
13. External prestressing

13.1 Introduction
13.2 Technology of external prestressing
13.2.1 Prestressing steel
13.2.2 Saddles at points of tendon deviations
13.3 General design processes
13.4 A practical application: the Long Key Bridge in Florida
13.5 Analysis of the behaviour of an externally prestressed bridge under an
increasing load
13.6 Failure mechanisms of structures with unbonded tendons
13.7 Application of external prestressing for the retrofitting of structures
13.8 Literature
Chapter 1 - Introduction
1. Introduction
1.1 The principle of prestressing
Concrete is a building material that is strong in compression, but relatively weak in tension.
Therefore, the occurrence of tensile stresses in the structural design makes that the designer
has to take action. There are two ways to address this:
 embed another material strong in tension in the concrete, in which case it is denoted as
reinforced concrete (cracking is not prevented, but crack widths should be limited);
 compensate for the tensile forces, which can be achieved by arching or prestressing.
This course focuses on the principle of prestressing.
Consider a statically determinate simply supported beam of a homogeneous, linear-elastic

material as shown in fig. 1.1. The beam must be able to carry its selfweight (permanent load)
and a variable load. The largest stresses in the beam are generated at midspan. It is assumed
that the strains ε and stresses σ vary linearly over the height of the beam (plane cross-sections
remain plane; Bernoulli’s hypothesis).
εct σct (-)
εcb σcb (+)

Fig. 1.1 Stresses as a result of a permanent load (qG) and variable load (qQ) in a beam that
behaves linear elastic
Because of the relatively low tensile strength of concrete, a reinforced concrete beam will
start to crack at an outer fibre at a relatively small load. In the opposite outer compressive
fibre, the compressive strength of the material is then far from reached. After cracking, the
longitudinal reinforcement takes over the tensile force (fig. 1.2). Although the beam is
cracked, it can sustain the load provided that a sufficient amount of steel reinforcement is
applied. Provided that the crack widths in the serviceability limit state (SLS) are sufficiently
limited, the expected service life time is not affected.
εct
εcb
Fig. 1.2 Behaviour of reinforced concrete after cracking
April 2018 1-1

Thus, structures in reinforced concrete should meet crack width requirements. Moreover, the
structure should meet SLS deflection requirements and should have sufficient resistance at the
ultimate limit state (ULS). Usually this is not a problem: by a rational choice of the
dimensions of a structural element (e.g. height h  lmin 35 for floor slabs), deflection
requirements are met implicitly. In addition, the longitudinal reinforcement required to resist
the bending moment at the ULS is in most cases sufficient to meet crack width requirements.
However, a number of factors exist that limit the application of reinforced concrete:
1. The load bearing capacity of an element cannot be increased unlimitedly by still further
increasing the amount of reinforcement. For a reinforcement ratio of over 0,02 (depending
on the strength class of the concrete and the type of reinforcing steel), the compressive
strength of the concrete is governing at ULS and the reinforcing steel will not reach its
yield strength. As a result, an undesired brittle fracture of the element can occur. In
addition, it often becomes practically difficult to apply such a high amount of reinforcing
steel in the cross-section of the element.
2. When the span increases, the bending moment caused by the selfweight increases more
than proportional ( 18 qG l 2 ). The ratio between the variable load that can be applied and the
permanent load becomes more and more unfavourable. Furthermore, problems may arise
with regard to limiting the deflections below certain limit values.
Crack formation at SLS can be prevented by applying an eccentric compressive force Pm on

the beam. In this way, high compressive stresses are generated at the bottom side of the beam,
while small tensile stresses occur at the top side (fig. 1.3a). By superposition of the stresses
caused by Pm (fig. 1.3a) and the permanent load (selfweight + static loads) of the beam (fig.
1.3b), the stress distribution shown in fig. 1.3c is obtained. At the top side, small or no tensile
stresses at all occur, whereas at the bottom side of the beam compressive stresses occur. In
order to keep the beam free of cracks, the possible tensile stresses should not exceed a certain
limit value. Also the compressive stresses must be kept below a limit value.
Due to the variable load, compressive stresses are generated at the top side of the beam and
tensile stresses at the bottom side. This stress distribution has to be superposed on the stress
distribution that is already present (fig. 1.3c). The result is shown in fig. 1.3d. Again, tensile
stresses (bottom side) and compressive stresses (top side) must be limited.
Fig. 1.3 Stresses in the cross-section at midspan, as a result of the prestressing force (P),
permanent loading (G) and variable loading (Q)
April 2018 1-2

1.2 Application of the prestressing force

Several methods have been developed to apply a prestressing force to concrete. The most
important ones are:
1. The use of fixed points in between which the structure is prestressed.
2. Pouring of the concrete around tensioned tendons of high-quality steel. This is called pre-
tensioning: prestressing with pre-tensioned steel.
3. Tensioning of the tendons after the concrete has hardened. This is called post-tensioning:
prestressing with post-tensioned steel.
Method 1: Prestressing between fixed points

An element is placed between two fixed points (fig. 1.4a). The compressive force is applied
by a hydraulic jack. The fixed points may be cliff faces or very solid soil masses. The high
initial prestressing force will be highly reduced due to creep and shrinkage (shortening) of the
concrete. Therefore, this method of prestressing is suitable only for temporary structures,
unless regular post-tensioning is possible (for example by jacking). Therefore, this method is
only rarely applied.
rigid support
a structural member
jack
jack element pre-tensioned steel

pretensioning steel
A A
b
prestressing steel (bar or cable) jack
c sheath element
Fig. 1.4 Various prestressing methods: (a) prestressing between rigid supports; (b) pre-
tensioning; (c) post-tensioning
Method 2: Pre-tensioning
Tendons are positioned, stressed and anchored before the concrete is cast in the mould (part
A-A of Fig. 1.4b). Once the concrete has sufficiently hardened, the ends of the tendons are
gradually or instanteneously released from the external fixing points. From this moment on,
the concrete element is prestressed. The ends of the tendons are cut off. The prestressing force
April 2018 1-3

is transferred from the tendons to the concrete over a certain distance (the transmission length
lpt2) by the bond stresses between concrete and tendon. The magnitude of the transmission
length lpt2 depends on tendon type and its cross-section, surface profile (roughness) and initial
stress (lpt2 = 300 - 700 mm). This method is very suitable for the pre-fabrication of elements.
Method 3a: Post-tensioning with bonded tendons

Special profiled ducts are installed in the mould before the concrete is cast. Tendons are
installed in the ducts before or after casting of the concrete. After casting and hardening of the
concrete, the end faces of the concrete element are used as supports for the jacks and
anchorages are used to stress the tendons (see fig. 1.4c). Since the tendons are placed in ducts,
they can deform relative to the concrete. After tensioning, the tendons are anchored. The
prestressing force is now transferred from the anchorage plates to the concrete. The ducts (or:
sheaths) are then injected with a special grout. The grout bonds the tendons to the duct,
enabling the transfer of forces from the tendons to the concrete. Moreover, the tendon is
protected against corrosion.
Method 3b: Post-tensioning with unbonded tendons

Tendons are first coated with grease or a bituminous material and then covered by a smooth
plastic sheeting. This prevents corrosion of the tendon. The tendon is then put in position
before the concrete is cast. After the concrete has sufficient strength, the tendons are stressed
to a pre-determined force and anchored. Because the tendon can slip in its sheeting and the
sheeting – concrete interface is relatively smooth, there is hardly any force transfer by bond
from concrete to steel.
1.3 Curved tendon profiles
The beam from fig. 1.3 is prestressed to such a level that the tensile stresses in the critical
cross-section caused by the permanent and variable load (bottom fibre at midspan) are
compensated for by compressive stresses. However, this way of prestressing generates tensile
stresses at the top side of the beam near the supports, which are not compensated for by the
bending moment caused by the permanent load (fig. 1.5: cross-section B-B).
Therefore, at the supports, the anchorages of the tendons should be positioned higher, for
instance within the kern area of the cross-section. Then, no tensile stresses occur. If the centre
of gravity of the anchorages is outside the kern area, tensile stresses occur. These stresses
should not exceed a certain limit value. In most cases, the centre of gravity of the tendons at
the support is chosen such that it coincides with the centre of gravity of the concrete cross-
section. The tendon now is no longer straight, but has a curved shape. When a parabolic
tendon profile is used (fig. 1.6), the stresses resulting from the prestressing force have the
same profile as the stresses caused by a uniformly distributed load.
April 2018 1-4

qG
unloaded structure fully loaded structure
stresses in
section A-A
by Pm by qG by qQ
stresses in
section B-B
by Pm only
by Pm only
Fig. 1.5 Stress in the cross-section at midspan and support, respectively
section A-A section B-B
Fig. 1.6 Parabolic tendon profile
Because of the curvature of these tendons, not only forces Pm act on the end faces of the
beam, but also forces perpendicular to the tendon profile. This distributed load acts in upward
direction and is called the curvature pressure (qp).
In most cases, not one but several tendons are applied, see fig. 1.7. The tendons are then
positioned such that their overall centre of gravity more or less coincides with the parabolic
profile as shown in fig. 1.6. The tendons are spread over the end faces to provide space for the
anchorages and to have a more uniform distribution of the (often high) anchor forces.
April 2018 1-5

Fig. 1.7 Tendon layout when several tendons are applied
The pre-tensioning principle can be applied too (fig. 1.4). To obtain a similar distribution of
the prestressing as presented before, the tendons are kinked as shown in fig. 1.8. Note that this
is only possible when using tendons that are suited to be bent, for instance strands (not bars).
In most cases, a number of strands is not raised at their ends but is (almost) parallel to the
beam axis. This also has the advantage that a steel tensile tie (see section 8.4.3) is present at
the bottom side over the full length of the beam.
Fig. 1.8 Kinked tendon profile applied with pre-tensioned steel
The slope of the tendon near the support has an additional positive effect: the upward directed
vertical component of the prestressing force reduces the shear force caused by the loads,
which results in a reduction of the shear force and, if required, less shear reinforcement
(stirrups).
1.4 Prestressed concrete versus reinforced concrete

Thanks to an optimised use of the strong points of concrete (compressive strength) and steel
(tensile strength), it is possible to design more slender structures in prestressed concrete than
in reinforced concrete (see fig. 1.9). This means a reduction of structural height at the same
span length, which has a positive effect on selfweight and on the forces exerted on the
foundation.
April 2018 1-6

Fig. 1.9 Test loading in the U.S. in 1954; span 9,2 m, thickness of the prestressed slab
50 mm, with a cast topping of 100 mm
It is of course also an option to increase the span when using the same structural height. The
absence or strong reduction of crack formation in SLS has advantages with respect to water
tightness and corrosion resistance. Therefore, offshore structures are often prestressed.
Fig. 1.10 Prestressed offshore gravity-based structure, built at Rotterdam (1992)
As an example, fig. 1.10 shows the construction of the F3 P1 offshore platform, designed for
the F3-field in the North Sea. This structure is built up with cells, which are prestressed at the
water retaining side. Thanks to the absence of cracks, the stress cycles in the steel are
relatively small. Therefore, fully prestressed concrete has a very high resistance against
fatigue.
April 2018 1-7

Besides the advantages mentioned, prestressing has the disadvantage that it is often
expensive. The tendons themselves and, especially, the anchorages are more expensive than
the traditional reinforcing steel. Furthermore, the tensioning and grouting activities are quite
labour intensive.
Prestressed and reinforced concrete both have their own advantages and disadvantages. For
each application, investigation has to reveal which of the two options is the best.
Until recently, a fundamental difference existed regarding the design of reinforced concrete
and prestressed concrete. A structure in reinforced concrete was often designed with respect
to the ultimate limit state (ULS): when the dimensions of the concrete cross-section and the
reinforcement met the requirements at the ULS (i.e. a failure criterion; strength), then in most
cases the conditions with respect to the serviceability limit state (SLS; e.g. crack width
control, deflections) were met automatically.
For prestressed concrete, on the other hand, the design was based on the SLS. The design was
performed such that the concrete tensile and compressive stresses for the most unfavourable
load combination in SLS were within certain limits and the magnitudes of the corresponding
deformations were checked. Only afterwards, it was checked whether the requirements at the
ULS were met. The ULS requirements were most often met and it appeared that SLS
requirements were governing.
However, several decades ago it became clear that this clear separation between designing in
reinforced concrete on one hand and prestressed concrete on the other, not always resulted in
the most economical solution. Situations can occur in which a combination of both design
systems gives better results. This combination is denoted as partially prestressed concrete.
The designer then has to be aware that a reduction of prestressing will result in crack
formation at the SLS. This aspect requires special attention because prestressing steel is more
corrosion sensitive than reinforcing steel. Therefore, in most cases extra steel reinforcement is
applied to restrict the crack width.
In several countries, the application of “partial prestressing” has already been incorporated in
daily engineering practice. In the Netherlands, designers were very reluctant to apply partially
prestressed concrete, also because of the lack of specific regulations, codes and design rules.
With the introduction of TGB 1990 “Loads and deformations” (NEN 6702) in 1990, and the
“Regulations for concrete - Structural requirements and calculation methods” (NEN 6720,
also known as VBC 1990 and VBC 1995), the existing distinction between reinforced
concrete and fully/limited/partially prestressed concrete almost disappeared. As stated, this
was thanks to the introduction of a uniform design method for both reinforced and prestressed
concrete.
In a design according to NEN 6702 / NEN 6720, the prestressing has to be modelled as a load
applied on the structure. By doing so, a gradual transition between reinforced concrete on one
hand and fully prestressed concrete on the other, with all possible intermediate options, is
acquired. This integral approach is often referred to as “structural” concrete, no longer using
“reinforced” or “prestressed” concrete [1.2]. This appoach is used in EN 1992-1-1 too.
April 2018 1-8

To summarise: the technique of prestressing has considerably increased the opportunities

when designing in concrete. The technique of prestressing has developed from an exotic
special way of constructing (fig. 1.9) to one of the most important construction technologies,
of which many successful examples can be given (fig. 1.11).
Fig. 1.11 Prestressed structure for the subway ring line in Amsterdam
1.5 Literature
1.1 Edwards H., “The innovators of prestressed concrete in Florida”, Journal of the PCI,
Special Issue for the FIP-Conference in Stockholm, June 6-10, 1982.
1.2 Bruggeling A.S.G., “Constructief beton”, Cement 1987, no. 1, 2, 3, 4, 6, 7, 9, 10, 12.
Bruggeling A.S.G., “Science into practice”, Heron, Vol. 32, 1987, no. 2.
April 2018 1-9

April 2018 1-10

Chapter 2 – Materials and methods
2. Materials and methods of prestressing

2.1 Prestressing steel
Reinforcing steel was used in the first attempts to prestress structures. A calculation will
now demonstrate whether reinforcing steel is a suitable material to exert a prestressing
force on a structure for a long period of time.
A member having a length of 10 m is pre-tensioned with reinforcing steel B500

(subscript s) along its neutral axis (fig. 2.1). It is assumed that the initial stress in the steel
after prestressing is σsmax = 350 N/mm2, which is assumed to induce an initial
compressive stress in the concrete σc = -12 N/mm2.
steel bar in untensioned state
(pre)tensioning of the steel
17,5 mm
casting of the concrete
17,5 mm
elastic shortening (Δlce) of the

concrete caused by the release of the
prestressing force
14,1 mm
final state condition after shrinkage

(Δlcs) and creep (Δlcc) of the
concrete 14,9 mm
Fig. 2.1 Stress-losses if reinforcing steel is applied to axially prestress a concrete

member
Before releasing the bar, the steel strain is:

350
s   1, 75 103
200 10 3
April 2018 2-1

The corresponding total elongation of the reinforcing steel:
ls  1, 75 103 10 103  17,5 mm
When applying a modulus of elasticity of the concrete Ec =35·103 N/mm2, the elastic
shortening of the concrete (subscript c) caused by releasing the prestressing force is
approximately:
12
lce  10 103  3, 4 mm
35 10 3
Creep of the concrete (plastic deformation under constant loading; subscript c) causes an
additional time-dependent shortening, which is about 2 to 3 times the elastic deformation,
i.e.:
lcc  8,5 mm
Shrinkage of the concrete (shortening due to drying of the concrete; subscript s) results in
a shortening on long term of about 0,3·10-3, or:
lcs  0,3 103 10 103  3, 0 mm
The total shortening caused by elastic deformation, creep and shrinkage is about 3,4 + 8,5
+ 3,0 = 14,9 mm. This implies that only 17,5 – 14,9 = 2,6 mm of the original steel bar
elongation of 17,5 mm remains. As a result, the initial tensile stress in the steel of
350 N/mm2 is reduced to (2,6/17,5) · 350 = 52 N/mm2, and the concrete stress σc is down
from -12 N/mm2 to (52/350) · -12 = -1,8 N/mm2 (compression).
It can be concluded that reinforcing steel is not suited to apply a prestressing force.
To apply prestressing successfully, the shortening of the concrete has to be kept small
compared to the elongation of the prestressing steel used. This can be achieved by
applying steel with a much higher tensile strength and strain at failure. Therefore,
prestressing steel has a tensile strength of 900 to 2000 N/mm2.
This high steel quality is achieved by:

 altering the chemical composition of the steel, for example by increasing the carbon
content;
 improving the microstructure of the steel by heat treatment or mechanical treatment
(for example cold deformation).
Several types of prestressing steel are available.
April 2018 2-2

Bars having a diameter of Ø15 to Ø50 mm
The steel type ranges from Y1030H to Y1230H, in which the number indicates the
nominal tensile strength in N/mm2. The bars are made from steel that is hot rolled into
bars (code H). Subsequent processing (e.g. accelerated cooling, cold stretching,
additional tempering) might be required to achieve the required mechanical properties.
The bars can be plain (code P) or ribbed (code R) (fig. 2.2).
Fig. 2.2 Ribbed (left) and plain steel (right) prestressing bars
Ribbed bars have better bond properties and offer the advantage that they can be cut and
anchored at any position, or can be extended by coupling. The modulus of elasticity is
205 GPa for bars that are either rolled only or as rolled stretched and tempered. For bars
that are as rolled stretched only, the E-modulus is significant lower and is about 165 GPa.
The possible combinations of bar diameters and steel strength according to EN 10138-4
are given in table 2.1
Table 2.1 Properties of prestressing bars (EN 10138-4)

type diameter area tensile strength Rm [MPa]
d [mm] Sn [mm2]
1030
1100
1230
15,0 177 R
20,0 314 R
25,5 511 P
26,0 531 P P
Y….H 26,5 552 R R
32,0 804 P, R P, R
36,0 1018 P, R P, R
40,0 1257 P, R P, R
P = plain bar, R = ribbed bar
Ep = 205 GPa for as rolled and as rolled stretched and tempered bars
Ep = 165 GPa for as rolled and stretched bars
Note:
EN 1992-1-1 cl. 3.3.6 presents an E-modulus of 205 GPa for bars and wires; 195 GPa for
strands.
April 2018 2-3

Wires with a diameter of Ø3 to Ø10 mm

These wires are cold drawn (code C) and heat-treated; the steel quality ranges from
Y1570C to Y1860C. The wire surface can be plain, grooved or ribbed (fig. 2.3).
Fig. 2.3 Ribbed and plain prestressing wires
The modulus of elasticity of bars and wires is about 205 GPa. Properties of wires
according to EN 10138-2 are listed in table 2.2.
Table 2.2 Properties of cold deformed prestressing wires (EN 10138-2)

diameter area tensile strength Rm [MPa]
2
d [mm] Sn [mm ]
type
1570
1670
1770
1860
3,0 7,07 x
3,2 8,04 x
4,0 12,57 x
5,0 19,63 x x
6,0 28,27 x
6,9 37,39 x
Y….C 7,0 38,48 x
7,5 44,18 x
8,0 50,27 x
9,4 69,4 x
9,5 70,88 x
10,0 78,54 x
Ep = 205 GPa
April 2018 2-4

Strands, constructed from 3 or 7 wrapped wires

Figure 2.4 shows a strand of seven wires, six of which are spun around a straight central
wire. The diameter of the wires ranges from 2 - 5 mm. The central wire has an at least
2,0% greater diameter than the outer helical wires. The lay length is between 14 and 18
times the nominal strand diameter.
Fig. 2.4 Seven wire strand
The properties of strands according to EN 10138-3 are given in table 2.3.
Table 2.3 Properties of 3 and 7 wire prestressing strands (EN 10138-3)

number diame- area tensile strength Rm [MPa]
of ter Sn [mm2]
type
1700
1770
1820
1860
1960
2060
2160
wires d [mm]
5,2 13,6 x x x
6,5 21,1 x x
Y….S3 3 6,8 23,4 x
7,5 29,0 x
6,85 28,2 x
7,0 30,0 x x
9,0 50,0 x x
11,0 75,0 x
Y….S7 7
12,5 93,0 x
13,0 100 x
15,2 140 x x
16,0 150 x x
18,0 200 x
12,7 112 x
Y….S7G
7 15,2 165 x
compact 18,0 223 x
Ep = 195 GPa
The steel type ranges from Y1670 to Y2160 (note: In NL the maximum steel quality
allowed for use is Y1860, except for foundation piles in which higher strengths are
allowed). The number of wires used is part of the code used: Y1860S7 denotes a 7 wire
strand constructed of strands having a tensile strength of 1860 N/mm2. The modulus of
elasticity is about 195 GPa.
The specification G, in for example Y1820S7G, implies that it is a so-called compacted

strand.
April 2018 2-5

Stress-strain relationship
Figure 2.5 shows the stress-strain diagram (σ-ε diagram) for a number of different steel
types. Prestressing steel often has no pronounced yield point. In these cases, as an
alternative for the yield point, the value fp0,1k is used, where fp0,1k is the stress that, after
unloading, causes a permanent deformation (plastic strain) of 0,1 %.
s [N/mm2]
2000 yield value s

Y1860 cold-worked
cold-workedsteel
prestressing steel
FeP 1860 steel
Y1770 cold-worked
FeP 1770 cold-workedsteel
steel yield value 0.1
tensile strenght
Y1670 cold-worked
FeP 1670 cold-workedsteel
steel
failure
1500
Y1230
FeP 1230
Y1030
FeP 1030
1000
reinforcing steel
B500B
FeB 500 HK
500
FeB 400 HWL FeB 400 HK
FeC 220
0
0.1% 5 10 15 20
 [%]
Fig. 2.5 Stress-strain relationships of reinforcing steel and prestressing steel
The  - relationships given in fig. 2.5 cannot be used directly in the calculation of the
bending moment resistance in ULS (the failure bending moment). Therefore, EN 1992-1-
1 uses schematised  - diagrams (fig. 2.6). The position of the slope discontinuity in
this diagram is chosen such that the actual curve of fig. 2.5 is described quite accurately.
Note:
The subscript k refers to 'characteristic' value. This is a lower bound value derived from
test results, for instance from tensile tests on a prestressing bar or compressive tests on
concrete cubes. The characteristic value of the tensile strength of the prestressing steel fpk
follows from the characteristic value of the maximum force resisted by the steel and its
cross-sectional area. The design value of the strength follows from the characteristic
value, divided by a material factor γs.
Note that the NL National Annex to EN 1992-1-1 prescribes γs = 1,15 for reinforcing
steel and γs = 1,1 for prestressing steel (EN 1992-1-1 table 2.1N).
April 2018 2-6

Fig. 2.6 Schematic stress-strain relationships for prestressing steel according to EN

1992-1-1 fig. 3.10
The value of fp0,1k can be calculated from EN 10138 by using Fp0,1k and Sn. According to
EN 1992-1-1 fig. 3.10, it is also allowed to assume that fp0,1k = 0,9 fpk. From this
assumption it follows that fpd = fp0,1k / γs = 0,9 fpk / 1,1.
The NL National Annex states that εud = 0,9 εuk. According to EN 10138 the minimum
value of εuk = 3,5 %.
Often, 7-wire strands from steel type Y1860S7 are used. Their nominal cross-sectional
area is often 100 or 140 mm2 (identification diameter Ø13,0 and Ø15,2 mm,
respectively), see Table 2.4.
Table 2.4 Data of prestressing strands used in tendons.
steel type Y1860S7

characteristic diameter (mm) 13,0 15,2
cross-section of steel (mm2) 100 140
design value of failure load (Ap fpk / γs) (kN) 169 235
Maximum prestressing force after tensioning (kN) 140 194
pm0 = 0,75 fpk =0,75·1860 = 1395 MPa
Maximum prestressing force during tensioning (kN) 149 207
p,max = 0,80 fpk = 0,80·1860 = 1488 MPa
The prestressing force depends on the maximum stress allowed after anchorage of the
prestressing steel (according to EN 1992-1-1 cl. 5.10.3: the minimum value of 0,75 fpk =
0,75·1860 = 1395 N/mm2 and 0,85 fp0,1k = 0,85·0,9·1860 = 1423 N/mm2 (Y1860)).
April 2018 2-7

During stressing EN 1992-1-1 cl. 5.10.2.1 allows at the anchorage side a maximum stress
that is the minimum value of 0,8 fpk = 0,8·1860 = 1488 N/mm2 and 0,9 fp0,1k =
0,9·0,9·1860 = 1507 N/mm2 (Y1860).
When the jack has an inaccuracy of less than 5%, it is allowed to overstress to 0,95 fp0,1k
= 0,95·0,9·1860 = 1590 N/mm2 (Y1860) (Note that overstressing is not allowed
according to the Dutch National Annex. This is included in table 2.4).
Note that prestressing bars are not allowed to be overstressed.
2.2 Methods of prestressing

In Chapter 1 the following methods of prestressing are presented:
1. prestressing with post-tensioned steel, with bond;
2. prestressing with post-tensioned steel, without bond;
3. prestressing with pre-tensioned steel.
2.2.1 Prestressing with post-tensioned steel, with bond
Ducts are installed in the formwork or mould and the concrete is cast. As soon as the
concrete has developed sufficient compressive strength, the tendons (bars, wire bundles
or bundles of strands) are installed, tensioned and anchored. Shortly after installing the
anchorages, the open space between the tendons and the ducts is injected with a special
grout to protect the very corrosion sensitive prestressing steel. Finally, for corrosion
protection, the anchorages are covered, for instance with concrete.
Note:
According to EN 1992-1-1 cl. 5.10.2.2 the minimum concrete strength fcm(t) required to
be allowed to start stressing (stepwise per individual tendon) is 50% of the minimum
required compressive strength for full prestressing as given in the ETAG certificate of the
prestressing system. From this minimum value fcm(t) to the strength given in the
certificate, the prestressing force of a tendon can be stepwise increased from 30% of the
maximum tendon force up to the full 100% of maximum tendon force. Additionally,
during stressing, the concrete compressive stress caused by prestressing and other loads
should not exceed 0,6 fck(t). In the case of pre-tensioned steel, this value may be
increased to 0,7 fck(t).
The ducts for the tendons are created in the concrete using special thin-walled steel
sleeves (ducts, sheaths) as shown in fig. 2.7. The wall thickness of these ducts is about
0,2 to 0,4 mm. The ducts are produced in lengths of about 6 m and extended by
couplings. The couplings are sealed to prevent undesired leakage because this might lead
to problems during prestressing or injecting of the prestressing elements. The prestressing
is carried out by hydraulic jacks, examples of which are shown in figs. 2.8 and 2.9.
April 2018 2-8

Fig. 2.7 Steel sleeves (ducts, sheaths; producer: Drosbach)
Fig. 2.8 The tensioning of a bundle of strands (system DSI)
April 2018 2-9

Fig. 2.9 Jack and oil pump
In fig. 2.10, the several stages during prestressing are demonstrated.
Injection should be carried out using moderate pressures; a too high pressure might lead
to pushing away (push-out) of the concrete cover on the sleeve, the generation of splitting
cracks in the concrete, or the penetration of grout into empty sleeves close to the duct
being injected.
During the injection of grout, the entrapped air accumulates at high positions in the
sleeve. Vent tubes are installed at these positions to release the air. This enables a
complete injection of the sleeve. A poorly performed injection with inclusion of air may
eventually lead to corrosion damage. The costs of repairs carried out later will often by
far exceed the original construction costs.
Several techniques exist for the anchorage and coupling of prestressing steel. A number
of often applied methods is discussed.
April 2018 2-10

step 1:
Before installing the jack, an accessory (A) is
attached to enable accurate centring of the jack.
(A)
jack
wedges (B)
Step 2:
When the jack is stressed, the conical wedges
(B) are loosened.
travel
Step 3:
During tensioning, the prestressing tendon is
pulled out of the structure.
wedges
Step 4:
When the desired tendon force is reached, the
wedges are pressed hydraulically.
Step 5:
The oil is drained form the jack, thus
decreasing the jack force. The jack is removed
after full release of the force.
Fig. 2.10 Stressing of a tendon with a hydraulic jack
April 2018 2-11

Anchorage by means of anchor plates

Figure 2.11a shows a system for a prestressing bar, and fig. 2.11b for a tendon with
strands. Several methods are available to attach the bars, wires or strands to the plates.
(a) (b)
Fig. 2.11 Anchorage with bearing plates (system Dywidag)
These methods are schematically shown in fig. 2.12.
Type Wire Strand Bar

Anchoring
Wedges
Upsetted
heads
Nut
Fig. 2.12 Several methods to attach the bars, wires and strands to the anchor plate
In split-wedge anchorages, three-parted or four-parted conical wedges are used, the inside
of which has fine and sharp serrations and the outside is smooth. The principle of this
type of anchorages is illustrated in fig. 2.12 (first row) and fig. 2.13.
The BBRV system realizes the anchorage by mechanical upsetting of the wire ends (fig.
2.12 second row). As presented before in fig. 2.11, anchoring can be achieved by nuts as
well (fig. 2.12 last row).
April 2018 2-12

anchor plate
three-parted
conical w edges
Fig. 2.13 Split-wedge anchorage of a tendon composed of strands
Because large concentrated forces are transmitted through the anchor plates to the
concrete, the concrete compressive stresses usually will exceed the standard allowable
values. The application of spiral reinforcement (fig. 2.14) creates a volume in which an
external confining pressure is generated. This allows the concrete to resist these high
compressive stresses (EN 1992-1-1 cl. 3.1.9). The spiral reinforcement is an integral part
of the prestressing system and is also presented in the ETAG certificate of the system.
In addition, splitting reinforcement is applied, in order to prevent the generation of
splitting cracks in the concrete because of the introduction of large concentrated forces,
(also see Chapter 10).
Fig. 2.14 Spiral reinforcement to resist the high compressive stresses in the zone where
the prestressing force is transferred from the steel to the concrete
April 2018 2-13

Anchorage by means of a solid plate anchor

In this type of anchorage, the spiral reinforcement is replaced by a thick internal
cylindrical plate. Figure 2.15a shows such an anchorage for a prestressing bar and
fig. 2.15b for a tendon with strands.
a b
Fig. 2.15 Solid plate anchorage (system Dywidag)
Blind (dead-end, fixed-end) anchorage

This type of anchorage is applied when a cable can only be (or: has to be) prestressed at
one end. Figure 2.16 shows some solutions. The strands of the cable in fig. 2.16a are
spread out at the end, which allows the tendon force to be transmitted to the concrete by
bond and by the bulbs at the ends of the strands. When insufficient bond length is
available, an extra anchor plate can be applied (fig. 2.16b). This type of anchorage is used
in large, massive structures.
For the construction of offshore structures with sliding formwork, a different type of
blind anchorage is applied for the prestressing in vertical direction (fig. 2.16c). The types
of anchorage shown before would be very unpractical in this case, because the blind
anchors would have to be cast in concrete at the bottom of the structure. This implies that
the tendons are installed already at the start of the sliding process, which causes a lot of
inconvenience during construction. This is overcome by initially installing only the
sleeves during the casting process. After the positioning of the prestressing anchorage on
top of the sleeve, the complete assembled prestressing element is lowered and the first
few meters at the bottom of the sleeve are injected. After hardening of the grout, the
prestressing elements are anchored by bonding. Bond is improved by applying plugs at
the lower ends of the strands. After tensioning the prestressing element, the rest of the
sleeve is injected. Additional reinforcement to distribute the high local load and to
prevent splitting of the concrete might be required in the transmission zone of the
prestressing force at the blind anchor.
April 2018 2-14

sheath
a bond head anchorage (Dywidag)
80m
strand
grout pipe
forced on
anchorage
block
5m
b flat anchorage c special bond anchorage (Dywidag)
Fig. 2.16 Blind end (dead-end, fixed-end) anchorage
Coupling anchors
It might be required to couple prestressing elements, for example during staged
construction. The most basic type is the screw coupling (fig. 2.17).
Fig. 2.17 Basic coupler to join bars (Dywidag)
Prestressing elements consisting of strands can be coupled by connecting the individual

strands (fig. 2.18), or by applying intermediate anchor plates and split wedges (fig. 2.19).
Fig. 2.18 Coupler to continue unstressed tendons individually
April 2018 2-15

Fig. 2.19 Coupler consisting of a multiplane anchorage body
2.2.2 Prestressing with post-tensioned steel, without bond
In this system no grout is injected. The grout is replaced by an anticorrosive agent around
the tendon. This is done during the manufacturing process. The strands are provided with
a layer of protective grease and are in a plastic sleeve (fig. 2.20). The strand now is
protected against corrosion for a long period of time. Attaching a mono-strand to an
anchorage is shown in fig. 2.21.
durable anticorrosive grease
PE-tube strands
Fig. 2.20 Unbonded single strand tendon

PE-plug
cast anchor
anticorrosive grease PE-sheath
PE-sleeve PE-sealing tape strand
wedges
anticorrosive grease
safety hanger
Fig. 2.21 Solid plate anchorage of unbonded tendon
April 2018 2-16

Thanks to the small cross-section of the tendon and the small required concrete cover, a
relatively large distance can be realised between the tendon and the neutral axis of the
concrete cross-section. This is especially important for thin structural elements such as
floor slabs.
2.2.3 Prestressing with pre-tensioned steel
This method of prestressing is mainly used for the industrial production of concrete
elements such as piles, floor and roof slabs and beams for bridges and industrial
buildings. This system is very well suited for the production of standardised elements.
Because the prestressing force is transferred by bond to the concrete over the so-called
transmission length (lpt2 ; EN 1992-1-1 cl. 8.10.2.2 & eq. (8.18)) only wires or strands are
used. The wires have some surface profile to limit the required transmission length. A
pronounced surface deformation would result in high bond stresses which, in turn, might
cause large splitting stresses in the surrounding concrete and even result in crack
formation.
Prestressing beds have lengths up to 200 m. Figure 2.22 shows plants for the
manufacturing of prestressed hollow-core slabs. After the tensioning of the tendons
between the abutments at the ends of the bed, concrete is cast continuously or per section
over the full length of the bed.
After hardening of the concrete, the concrete slab is cut into pieces of the required length
by a circular saw (fig. 2.23). The tendons are cut as well and they slip relative to the
concrete over the so-called transmission length (the ends of the tendons are pulled 0,1 –
0,7 mm into the concrete), and transfer the prestressing force to the concrete by bond (fig.
2.24). Because of the lack of concrete cover at the ends of the tendons, elements that are
applied outside (such as bridge beams) should be protected against corrosion by, for
example, bitumen.
April 2018 2-17

Fig. 2.22 Factory for the production of prestressed prefab elements cast after tensioning
of the steel (prestressing with pre-tensioned steel)
April 2018 2-18

Fig. 2.23 Cutting of the elements at the required length
strand
tension force
before
before sawing
sawing
through
steelstress
po
σp,max
saw cut
slipping of the wire bond stress
after after
sawingsawing
through
σp,0
pi
po
transmission length σp,max
stress decrease due to elastic
shortening of the concrete
Fig. 2.24 Stress in strands or wires before and after being cut (prestressing with pre-
tensioned steel)
In section 1.3 it was already mentioned that the positions of the tendons at the ends of the
beam should be such that at the top side no or just minor tensile stresses are generated.
For concrete elements with a small structural height, this can be achieved by straight
tendons. The large structural heights as applied in bridge beams require tendon profiles
that have raised ends (fig. 1.8). A similar effect can be obtained in, for instance, roof
beams by an increased height at midspan (fig. 2.25): at the supports, the prestressing
force is close to the centre of gravity of the cross-section whereas it is considerably lower
April 2018 2-19

in the cross-section in the middle of the beam (namely in the area loaded in tension by
permanent and variable loads).
centroidal axis prestressing steel
Fig. 2.25 Roof beam with variable height
The most important advantages of pre-tensioned steel are:

 It is not necessary to inject the prestressing ducts.
 The prestressing of the separate wires or bars is relatively easy to perform since only
small jacks are required (relatively low forces).
 The relatively expensive anchorage devices are not required.
2.3 Points of attention during prestressing
A concrete structure tends to deform during prestressing. The deformation should not be
restrained by the formwork or rigid supports. If, however, a restraint occurs, the forces
resulting from the restrained deformation should be accounted for properly.
During prestressing, both the prestressing force and the elongation of the tendons are
measured. In advance, their minimum and maximum values should be prescribed by the
designer. This information has to be available at the construction site in a “prestressing
protocol”. This protocol should also contain the order of prestressing in case more
prestressing tendons are applied. It also has to be indicated whether just one or both sides
of the tendons have to be stressed.
To prevent the formation of shrinkage cracks, it is an option to prestress the young

("green") concrete as soon as possible after casting. This has the additional advantage that
the formwork can be removed quickly and the production process can be speeded up,
making optimal use of the relatively expensive equipment. When the concrete strength is
still low, the tendons can be stressed stepwise.
The designer should be aware of the fact that already stressed tendons will loose some
prestressing force because of the additional elastic shortening of the concrete caused by
the following stressing of other elements (EN 1992-1-1 cl. 5.10.5.1 & eq. (5.44)).
In section 2.2.2 it was stated that the maximum steel stress allowed after anchorage of the
prestressing steel (EN 1992-1-1 cl. 5.10.3) is the minumum value of 0,75 fpk and
0,85 fp0,1k.
During stressing, it is allowed to use an increased maximum stress to compensate for the
stress losses due to wedge set, friction and elastic deformation of the concrete. EN 1992-
April 2018 2-20

1-1 cl. 5.10.2.1 restricts the stress to the minimum value of 0,80 fpk and 0,90 fp0,1k. When
the jack has an inaccuracy of less than 5%, it is allowed to overstress to 0,95 fp0,1k.
However, the NL National Annex prohibits overstressing.
In section 2.2.1 it was already mentioned that, according to EN 1992-1-1 cl. 5.10.2.2, the
minimum concrete strength fcm(t) required to be allowed to start stressing the tendons
(stepwise per individual tendon) is 50% of the minimum required compressive strength
for full prestressing as given in the ETAG certificate of the prestressing system. From
this minimum value fcm(t) to the strength given in the certificate, the prestressing force of
a tendon can be stepwise increased from 30% of the maximum tendon force up to the full
100% of maximum tendon force.
Additionally, during stressing, the concrete compressive stress caused by prestressing and
other loads should not exceed 0,60 fck(t). In the case of pre-tensioned steel, this value may
be increased to 0,70 fck(t).
Permanent compressive concrete stresses of over 0,45 fck(t) are allowed provided that the
designer accounts for non-linear creep (EN 1992-1-1 cl. 5.10.2.2 (5)).
The design should meet the requirements from EN 1992-1-1 cl. 7.3.1 with regard to
allowed steel stresses or crack widths in SLS (see Chapter 9).
2.4 Important characteristic values of prestressing steel
In table 2.5, the most important characteristic values of a number of frequently used
prestressing steel types is collected. The data are from EN 1992-1-1. For more detailed
information on prestressing steel, reference is made to EN 10138.
Table 2.5 Mechanical properties of prestressing steel.

steel type tensile fracture 0,1% maximum tensile stress slope modulus of
type strength strain proof- during during initial discontinuity elasticity
stress pre- pre-stressing stress in the  -
stressing with accurate diagram (ULS)
jack
fpk fpk/s pu fp0,1k p,max p,max pm0 fpd Ep
MPa MPa ‰ MPa MPa MPa MPa MPa GPa
Y1030H bar 1030 936 35 927 773 773 773 843 205 or 170
Y1670C wire 1670 1518 35 1503 1336 1428 1253 1366 205
Y1770C wire 1770 1609 35 1593 1416 1513 1328 1448 205
Y1860S7 strand 1860 1691 35 1674 1488 1590 1395 1522 195
More information about prestressing systems available in the Netherlands, such as the
cross-sectional area and the composition of the different prestressing elements, the
required ducts, the minimum bending radii, etc, are included in the appendices.
April 2018 2-21

April 2018 2-22

Chapter 3 – Axially prestressed
3. Axially prestressed structures

3.1 Introduction
A practical example of an axially prestressed element is the so-called “suspension

column”. Suspension columns are applied in buildings that should require as little as
possible floor space at ground level. Figure 3.1 shows an example of such a building in
which the floor slabs are suspended by suspension columns from a supporting structure at
the top of the building.
19
18
17
16
15
14
13
12
11
10
suspension column suspension column
9
2A
01
02
03
Fig. 3.1 Building with suspension columns
Compared to reinforced tensile columns, prestressed columns have the advantage that
they can remain uncracked during the service life, and, as a result, deform less. In
reinforced columns, a high reinforcement ratio would be required to limit crack widths.
Figure 3.2 schematically shows a comparison between a prestressed and a reinforced
column subjected to axial tension and provided with one steel bar.
April 2018 3-1

On the vertical axis, the subscripts refer to:

s reinforced concrete;
p prestressed concrete.
r rupture (cracking) of the concrete;
u the ultimate load that can be resisted;
rep the load representative for serviceability limit state (SLS) behaviour.
With respect to the behaviour, three phases can be distinguished:
Phase 1: The tensile strength of the concrete has not yet been reached. The column is
uncracked and has a high stiffness.
Phase 2: The column is cracked. The deformation of the steel bar is reduced by the
concrete in between the cracks, which is still active by bond (tension
stiffening; transfer of forces between steel and concrete). The stiffness of the
column decreases with ongoing crack formation.
Phase 3: The reinforcement has reached its yield strength and is in a limit state. The
bond forces transferred by the concrete between the cracks are now of minor
importance since the deformation of the steel bar is governed by the position
where yielding occurs.
1 2 3
Nu 3
2
Nrp
N reinforcement
without bond
1 2
reinforcing bar
Nrep (tension stiffening)
prestressed bar
Nrs
1
rep,p rep,r 
uncracked cracked yielding
steel
Fig. 3.2 Axial force - mean strain behaviour of an axially loaded tensile member that
consists of reinforcement only, of reinforced concrete and of prestressed
concrete, respectively
In a prestressed column, the concrete is under compression prior to loading. Therefore,

this column can be uncracked at higher load levels. As shown in fig. 3.2, the maximum
load at SLS (denoted as Nrep; a representative load) results in a strain εrep,p in the
prestressed (p) element, which is considerably smaller than the strain εrep,r in the
reinforced (r) element.
April 2018 3-2

3.2 Centrically reinforced element subjected to axial load
A centrically reinforced element is loaded by a compressive axial force F. The concrete

and steel cross-sectional area are Ac and As, respectively. The question is which part of
the axial force is carried by the concrete and which part by the steel.
The axial compressive force makes that the element shortens l (fig. 3.3). The concrete
and steel are assumed to be perfectly bonded and, as a result, they have the same strain.
For both materials it holds:
l
  c  s 
l
Fig. 3.3 Centrically reinforced tensile member loaded by an axial compressive force
The compressive forces in the concrete and steel are:
l
N c  Ac Ec (3.1a)
l
l
N s  As Es (3.1b)
l
From force equilibrium it follows:
F  N c  Ns (3.2)
From eqs. (3.1) and (3.2) the following relations for the forces are obtained:
1
Nc  F  c F (3.3a)
1   e s
 e s
Ns  F  s F (3.3b)
1   e s
April 2018 3-3

Es As
where:  e  ; s 
Ec Ac
Prestressing with pre-tensioned steel creates a similar situation (fig. 3.4).
Before cutting the tendons, the tensile force in the steel is Pmax and the concrete is free of
stresses. After the release of the force on the abutments, the force becomes a compressive
force that is taken over by the concrete element that contains prestressing steel. In
accordance with eq. (3.3), a part Pmax (1   e  p ) of the compressive force Pmax is carried
by the concrete and a part  e  p Pmax (1   e  p ) by the steel.
For the resulting forces Nc in the concrete and Pm0 in the prestressing steel it holds:
1 
Nc   Pmax 
1   e p 
   N c  Pm 0 (H  0!)
 e p 1 
Pm 0  Pmax  Pmax  Pmax
1   e p 1   e p 

The compressive force in the concrete is in equilibrium with the tensile force in the
prestressing steel. This horizontal force equilibrium follows from the requirement that,
once the tendons are cut, no external force is exerted on the concrete element.
abutment
before release of the prestressing force
after release of the prestressing force
Fig. 3.4 Prestressing with the aid of a steel bed
3.3 Calculation of the elongation required during prestressing
In section 2.3 it was already mentioned that for prestressing with post-tensioned steel,
two possibilities exist to check if the required force in the prestressing steel is reached.
The tendon force can be read from a load cell or the corresponding elongation can be
April 2018 3-4

measured (see fig. 3.5). Both measurements can also be done at the same time. The
measured elongation is then used to verify the force measurement.
Fig. 3.5 Measuring the elongation of the prestressing steel
When measuring the elongation, its components should be accounted for. To illustrate
this, the axially prestressed element from fig. 3.6 is analysed.
lp lc
sheath
fixed-end
anchorage
PFmax
po
Fig. 3.6 Elongation of post-tensioned prestressing steel
Initially, the tendon is positioned free of stress in its ducts. Then the tendon is stressed
until the prestressing force Pmax is applied. A hydraulic jack loads the prestressing steel.
The reaction force is exerted on the concrete. Afterwards, the tendon is anchored and the
duct is injected with grout.
Stressing causes an increase of the length of the tendon
Pmax l
 lp 
Ep Ap
April 2018 3-5

and a shortening of the concrete:
Pmax l
lc 
Ec Ac
The distance over which the tendon is pulled out of the concrete is the sum of both
displacements:
 l l 
l  lp  lc  Pmax  
 E A E A 
 p p c c 
When measuring the elongation, one should be aware that the tendon must be taut before
any stress can be built up (fig. 3.7). Therefore, in most cases, initially a small part of the
prestressing force is applied (pulling taut), and subsequently, the elongation is measured
for the remaining part of the force to be applied.
no force in
tendon
Fig. 3.7 Force - deformation relationship of a prestressing tendon
In the example, prestressing the member did not introduce any additional loading other
than the prestressing load itself. In practice however, often a part of the selfweight and
static load is activated during prestressing. The effect of this action should be included in
the calculation of the elongation. This will be demonstrated by an example with a
suspension column.
The structure from fig. 3.8 is designed such that before prestressing, the selfweight of the
column (Fg1) and the suspended floor slab II and its cross beam (Fg2) are carried by a
temporary structure, e.g. formwork. The prestressing force has to be designed such that
after prestressing a compressive force Fq is present in cross-section A-A. The question is
how to determine the elongation of the prestressing steel to achieve this.
During prestressing, the column shortens and the temporary supporting structure is
gradually unloaded. As soon as Pmax  Fg1  Fg2 , the prestressing steel has taken over the
total support reaction. At that moment, the elongation of the prestressing steel is:
April 2018 3-6

l1 
F g1  Fg2  l
Ep Ap
Fpo
Pmax
A A
floor I
suspension
column Fg1
floor  Fg2
formwork
auxiliary construction
Fig. 3.8 Suspension column during prestressing
The concrete stress σc in cross-section A-A now is equal to zero. To have a compressive
force Fq in cross-section A-A, the force in the prestressing steel has to be increased by Fq.
The corresponding elongation of the prestressing steel is:
Fq l
l2 
Ep Ap
whereas the shortening of the concrete is:
Fq l
l3 
Ec Ac
Note that the force in the concrete column is Fg1 + Fg2 (tension) from the selfweight of
the structure plus Fg1 + Fg2 + Fq (compression) from prestressing. The resulting force in
the concrete is Fq (compression).
The total force in the prestressing steel is: Pmax  Fg1  Fg2  Fq
The accompanying pulled-out length of the prestressing steel is:
l l
lp  l1  l2  l3  Pmax  Fq (3.4)
Ep Ap Ec Ac
April 2018 3-7

3.4 Design of a prestressed tensile member

3.4.1 Introduction
As an example of the design of a prestressed tensile member, a suspension column is

analysed. It is assumed that restrictions from traffic flows prohibit the use of a column at
ground level at the corner of a building. The problem is solved by applying a suspension
column that provides the required support (fig. 3.9).
The loads that have to be transmitted by this suspension column are:
Fg  Fg1  Fg2
selfweight of hanging column
permanent load from floor to be suspended = 500 kN
Fq  300 kN
variable load on the floor
tension side
1500
suspension column
3000
Ac
500
suspended floor
Fig. 3.9 Cantilever floor on one side suspended
The tension column is constructed from C28/35 concrete and is assumed to be prestressed
with post-tensioned steel. Prestressing takes place after the concrete has reached its 28-
day strength, so that the allowed (initial compressive) stress is σc = -0,6 · 28 =
-16,8 N/mm2 (EN 1992-1-1 cl. 5.10.2.2 (5)). The prestressing steel type is Y1030H. At
ULS the prestressing steel has to be designed to resist the forces caused by a fundamental
load combination with a load factor 1,2 for the permanent load and a load factor 1,5 for
the variable load.
April 2018 3-8

To make sure that the column remains uncracked (high stiffness to limit deformations), a
compressive stress of 2,0 N/mm2 is required when the column is subjected to the
maximum loading according to SLS.
It can be assumed that the time-dependent prestress losses are 15%. The immediately
occurring losses (namely the elastic shortening of the concrete and steel as a result of the
prestressing sequence (the stresses induced by prestressing other elements), the set at the
anchorage and the friction losses) can be neglected.
The following results are required:
 the cross-sectional area of the column;
 the deformation;
 the elongation of the prestressing steel relative to the concrete.
3.4.2 Global design
At maximum loading at SLS a (compressive) stress of 2,0 N/mm2 has to be present. The
most unfavourable situation occurs after the stress reduction caused by the time-
dependent prestressing losses, i.e. at a prestressing force of Pm(∞) = 0,85 Pm0 where Pm0 is
the initial prestressing force (after anchoring) and Pm(∞) is the working prestressing force
that is present after all time-dependent prestress losses are taken into account. Because
the immediately occurring losses (elastic shortening resulting from the stressing of
following tendons, the wedge set at the anchorage and the friction losses) are supposed to
be zero, it holds Pm0 = Pmax.
The density of the concrete is assumed to be   25 kN/m3 . The selfweight of the
column Fg2 = Ac l , where l is the column length (3 m).
Requirement:
A concrete compressive stress of at least -2 N/mm2 in SLS at full loading.
If it is assumed that full SLS loading occurs after the time-dependent prestress losses
have developed, the compressive stress requirement is:
 Pm (  )  Fg1  Ac l   Fq
c   2 N/mm 2 
Ac
0,85 Pm 0  500  Ac  3  25  300
c   2000 kN/m 2 
Ac
Ac  410 106 Pm 0  0,386 (3.5)
where Ac is the cross-sectional area of the column in m2 and Pm(∞) is the working
prestressing force in kN.
Requirement:
A not too high concrete compressive stress directly after anchoring the tendons.
April 2018 3-9

The maximum concrete compressive stress occurs when the floor is not yet loaded and
the prestressing force is still at its maximum. According to EN 1992-1-1 cl. 5.10.2.2 (5)
the initial (compressive) stress σc after anchoring has to be limited to -0,6 fck =
-16,8 N/mm2. From this it follows:
 Pm 0  Fg1  Ac l 
 ci   16,8 N/mm 2
Ac
or:
 Pm 0  500  Ac  3, 0  25
 16800 kN/m 2
Ac
Ac  59,3 106 Pm0  0, 0296 (3.6)
where Ac is in m2 and Pm0 in kN.
The eqs. (3.5) and (3.6) determine the possible combinations of Ac and Pm0 (fig. 3.10).
The minimum value of the cross-sectional area is Ac = 0,0307 m2 at a minimum initial
prestressing force Pm0 = 1016 kN (Fig. 3.10).
A column cross-section of 0, 25  0, 25 m ( Ac  0, 0625 m 2 ) is chosen.
0.3
Ac (m2)
0.2
0.1
n 3.6
equatio
0.0
5
3.
n
tio
-0.1 qua
e
-0.2
-0.3
-0.4
1000 2000
Fpi (kN)
Pm0 (kN)
Fig. 3.10 Allowable combinations of concrete cross-section and prestressing force
The corresponding minimum initial prestressing force follows from expression (3.5) and
reads: Pm0  1094 kN .
The associated cross-sectional area of the prestressing steel follows from the smallest
value of the maximum stresses allowed, see table 2.5:
 p,max  773 N / mm 2 (prestressing bars are not overstressed)
April 2018 3-10

It is assumed that losses from set at the anchor and friction can be neglected. Therefore,
σpm0 = σp,max.
It is found:
Pm0 1094 103

Ap    1415 mm 2
 pm0 773
Further, it should be checked whether the column has sufficient reserve against failure
(ULS), i.e.:
Ap 
1, 2 Fg  1,5 Fq

1, 2  500  1,5  300  103  1246 mm 2
f pd 843
Note that it is assumed that the bars reach the fpd = fp0,1k / s design strength, not the fpk / s
value, which is reached at a much higher strain. This is a conservative approach.
The SLS is governing. Four Dywidag bars Ø26 mm (see table 2.1) are selected with
Ap  4  531  2124 mm 2 . The duct dimension is Ø32/38 mm (inner/outer diameter).
The suspension column was designed in a global analysis in which the SLS condition at
t =  and ULS were checked. The results are now checked in a detailed analysis in which
characteristic consecutive moments in time are looked at. Moreover, the ducts are grouted
after anchoring. This implies that loads exerted before grouting are carried by a column
that has different properties than a column with grouted ducts.
3.4.3 More detailed calculation of the loads
The following data is used, which has been obtained from the global design method:
Prestressing steel
Total cross-sectional area: Ap  2124 mm 2
Concrete
Gross cross-sectional area: Ac  62500 mm 2
Net cross-section before grouting: Ac1  Ac  4 Aduct  62500  4  14   322  59,3 103 mm 2
Net cross-section after grouting: Ac2  Ac  Ap  62500  2124  60, 4 103 mm 2
Maximum prestressing steel stress
The stress in the prestressing steel at SLS reaches its maximum when the live load is
assumed to be activated directly after stressing and grouting. The time-dependent
prestressing losses are then still zero. According to eq. (3.3), the contribution of the steel
in carrying an additional external load applied to the column is:
April 2018 3-11

 e p
Np  Fq   p Fq
1   e p
where:
 e  Ep Ec  205 103 / 32, 0 103  6, 41 (Ec from EN 1992-1-1 table 3.1)
and
 p  Ap Ac2  2124 / 60, 4 103  0, 035 .
The remaining part of the tensile force Fq (i.e. Np – Fq) is carried by the concrete and
results in a reduction of the compressive stress from prestressing:
Fq Fq 300 103
 c   c  (1   p )  (1  0,183)   4,1 N/mm 2
Ac2 Ac2 60, 4 10 3
The increase of the steel force is:
 6, 41  0, 035 
N p   p Fq    Fq  0,183  300  54,9 kN
 1  6, 41  0, 035 
The increase of steel stress:
N p54,9 103
 p    26 N/mm 2
Ap 2124
Thus, the maximum value of the initial stress in the prestressing steel should not exceed:
 pm0  773  26  747 N/mm 2
from which the maximum initial prestressing force follows:
Pm0  Ap  pm0  2124  747  1587 103 N
This demonstrates that it might not be correct to prestress the bars in the column up to
their maximum allowable stress, because after applying the live load to the suspension
column, the stress in the bar increases and its maximum allowable stress (773 N/mm2)
might be exceeded.
The working prestressing force:
Pm( )  0,85 1587  1349 kN
April 2018 3-12

Subsequently, the occurring stresses will be checked at:

a. time t  0 ;
b. time t   .
Time t = 0
The axial compressive force in the concrete has its maximum value when the bars are
prestressed up to Pm0 and anchored. Now, only the permanent load Fg acts on the column.
In this state, with ducts that are not yet injected, the concrete (compressive) stress in
cross-section A-A of fig. 3.9 is:
 c1 
 Pm0  Fg

 1587  500  0, 0625  3, 0  25  103
 18,3 N/mm 2  16,8 N/mm 2
Ac1 59,3 103
It appears that the bars are stressed to a too high level; the bars should not be prestressed
up to the maximum stress allowed in the concrete. The maximum initial prestressing
force follows from:
 c1  16,8 N/mm 2

  Pm0  500  0, 0625  3, 0  25  103
 Pm0  1501 kN
59,3 103
with a corresponding stress in the steel:
Pm0 1501 103

 pm0    707 N/mm 2
Ap 2124
To summarise:
 The bars can be stressed to σp,max = 773 N/mm2 to meet the requirement on the
allowed initial prestressing steel stress before and directly after anchoring.
 The initial prestressing steel stress has to be reduced to 747 N/mm2 when the live load
is assumed to be present already at t = 0 (i.e. before time-dependent losses occur).
 The initial prestressing steel stress has to be further reduced to maximum 707 N/mm2
to meet the requirement on the initial concrete compressive stress caused by
prestressing.
After grouting of the prestressing ducts, the full live load is applied to the column, which
results in the previously calculated stress changes in the concrete and steel (see eq. (3.3)):
Fq Fq 300 103
 c   c  (1   p )  (1  0,183)   4,1 N/mm 2
Ac2 Ac2 60, 4 10 3
and in the steel:
Fq 300 103
 p   p  0,183   26 N/mm 2
Ap 2124
April 2018 3-13

1  e p
where:  c  and  p 
1   e p 1   e p
The stress in the concrete:
 c2  16,8  4,1  12, 7 N/mm 2  16,8 N/mm 2
And the steel stress:
 pm0  707  26  733 N/mm 2  773 N/mm 2
Time t  
It is assumed that the prestressing force reduces to 85% of its initial value, which is
caused by losses such as shrinkage, creep and relaxation (see Chapter 6). This implies
that the concrete is unloaded by a force 0,15 Pm0  0,15 1501  225 103 N , which follows
from internal force equilibrium in the column. In the unloaded state (i.e. no live load), the
concrete (compressive) force is:
0,15 Pm0 225 103

 c1  16,8   16,8   13,1 N/mm 2
Ac2 60, 4 103
After applying the live load, the concrete (compressive) stress is reduced. The stress
change was calculated before (+4,1 N/mm2). The resulting concrete stress is:
 c2   c1   c  13,1  4,1  9, 0 N/mm 2  2, 0 N/mm 2
The stress in the steel changes to 85% of its initial value plus the stress increase caused
by the external load:
Pm( ) 0,85 Pm0 F 0,85 1501 103 300 103

 p( )    p   p q   0,183  
Ap Ap Ap 2124 2124
 601  26  627 N/mm 2  773 N/mm 2
It can be concluded that during the service life of the structure, the conditions with
respect to the maximum stresses are satisfied.
It appears that the loading sequence influences the initial prestressing force allowed: The
structural engineer has to judge whether the live load can be present already directly after
grouting or will it be present after (part of) the time-dependent prestress losses have
occurred.
April 2018 3-14

3.4.4 Deformations of the column
Immediately after applying the prestressing force Pm0 and the activation of the permanent
load, the compressive strain in the concrete is:
 c1 
 Pm0  Fg

 1501  500  0, 0625  3, 0  25  103
 0,53 103  0,53 0 00
Ec Ac1 32 103  59,3 103
Applying the live load Fq causes an additional strain:
Fq 300 103
 c2   c  0,817   0,13 103  0,13 0
00
Ec Ac2 32 10  60, 4 10
3 3
During the service life, a time-dependent deformation occurs due to shrinkage and creep
of the concrete, as well as relaxation of the prestressing steel (see Chapter 6). The
elongation of the prestressing steel decreases. By approximation, it holds:
0,15 Pm0 0,15 1501 103

 p    0,52 103  0,52 0
00
Ep Ap 205 10  2124
3
3.4.5 Calculation of the elongation of the prestressing steel during tensioning
The elongation of the prestressing steel relative to the concrete at t = 0 is fully analogous
to eq. (3.4). One finds:
lp 5000
lp  Pm0   c1lc  1501 103   0,53 103  3000 
Ep Ap 205 10  2124
3
 17, 2  1, 6  18,8 mm
Note that it is assumed that the elastic shortening of the two floors can be neglected.
The example demonstrates that knowledge about the steps in the construction process and
the loading sequence are important to accurately estimate the stresses that develop in the
prestressing tendons and the concrete. A structural engineer should list possible loading
sequences and judge whether they are realistic.
In the case of the suspended column, it appeared that the time at which the live load
might be present, might be governing for the initial prestressing force allowed.
April 2018 3-15

April 2018 3-16

Chapter 4 - Bending
4. Prestressing of structures subjected to bending

4.1 Loading by prestressing
The position of the prestressing steel in concrete structures subjected to bending is mainly
determined by two factors:
1 - The concrete is not allowed to crack at all or just to a small degree, under the most
unfavourable load combinations. Therefore, limiting values are given for the maximum
tensile stress or crack width (also see Chapter 9). This usually leads to a low position of
the prestressing steel at midspan of a beam and a high position over the supports (the
position follows the concrete areas loaded in tension).
2 - It must be possible to anchor the prestressing force adequately. The anchor plates
must have prescribed minimum dimensions to transfer the high anchor forces. In
addition, a minimum mutual distance between the anchors should be applied to prevent
the development of unfavourable stress concentrations. Therefore, it is necessary to
spread out the tendons toward the beam-ends. In many cases, the anchorage issue is
governing in the choice of the type of prestressing tendon. It is often required to enlarge
the end of the concrete element in order to accommodate the required anchors ("end
block") (fig. 4.1).
Fig. 4.1 Prestressing anchors at the end of a beam (side view and longitudinal cross-
section)
As a result of these two factors, the tendons are usually draped (prestressing with post-
tensioned steel) or harped (prestressing with pre-tensioned steel), see also figs. 1.7 and
1.8.
April 2018 4-1

Chapter 4 - Bending
Because of this layout, curvature pressures or forces are generated, which load the beam.
This is illustrated in fig. 4.2.
c.a.
Pm sin  Pm sin 
Pm cos  Pm cos 
Fig. 4.2 Calculation of the curvature pressure in a beam with a draped tendon:
a. actual prestressing load on a beam
b. curvature pressure on tendon element having length R·dφ
c. relation between curvature radius R, beam length l and drape f
d. tensile force exerted on the tendon and curvature pressure exerted by
the tendon on the beam
e. schematised prestressing load on the beam
April 2018 4-2

Chapter 4 - Bending
In the example from fig. 4.2, the tendon is a circle segment with radius R and segment
height (also called the drape) f (fig. 4.2a). It is assumed that the prestressing force is
constant over the full length of the tendon. Figure 4.2b shows a small part of the curved
tendon. From the equilibrium equations it follows:
Pm
qp R d  Pm d  qp  (4.1)
R
From fig. 4.2c the following relation between the radius R and the drape f is obtained:
f  R  R 2  14 l 2
This yields:
f 2  2 f R  R 2  R 2  14 l 2  f 2  2 f R   14 l 2
In most cases R >> f, which implies that the term f 2 can be neglected. The relation then
reduces to:
l2
R (4.2)
8f
Substitution of eq. (4.1) into eq. (4.2) provides:
8 Pm f
qp  (4.3)
l2
The load generated by the tendon is shown in fig. 4.2d.
At the ends of the beam, the force Pm is introduced at an angle θ (also see fig. 4.2d). The
force Pm can be decomposed into a vertical force Pm sin  and a horizontal force
Pm cos  . In most cases, the angle θ is small. For a beam with a span of 25 m and a drape
f of 0,80 m, the angle θ is 7,5º (0,13 rad). Then it follows that Pm cos   0,991 Pm and
Pm sin   0,13 Pm . It is a good approximation to use a horizontal force introduced at the
beam-ends equal to Pm.
A similar reasoning holds for the curvature pressure qp  Pm R . The vertical component
is almost equal to Pm R . The horizontal component is relatively small. It has to
compensate for the difference between the horizontal force Pm cos  at the beam end and
the horizontal force Pm at midspan, which holds for a tendon that experiences no friction.
For an angle   7,5º the difference between both is 0, 009 Pm , which implies that the
contribution from the horizontal component of qp can be neglected.
April 2018 4-3

Chapter 4 - Bending
In the analysis of the problem, a total prestressing load as shown in fig. 4.2e has to be
taken into account.
EN 1992-1-1 (and also the Dutch code NEN 6720 (VBC 1995)) uses this approach in
which the prestressing is introduced as a load (EN 1992-1-1 cl. 5.10.1). In the literature,
this method is known as the “equivalent prestressing load method”.
An alternative for this method is the “cross-section method”.
In this method, the stresses in a cross-section are directly calculated using a cross-
sectional bending moment from prestressing equal to the prestressing force times the
tendon eccentricity in the cross-section considered. Figure 4.3a shows a case in which a
tendon has an eccentricity ep at midspan of a statically determinate beam.
The concrete stresses at the bottom and top side of the beam follow from:
Pm Pm ep
 cb    (bottom side) (4.4)
Ac Wcb
Pm Pm ep
 ct   
Ac Wct (top side) (4.5)
centroidal axis
Fig. 4.3 Statically determinate beam prestressed with draped tendons

a. calculations according to the cross-section method
b. calculations according to the equivalent prestressing load method
Exactly the same values can be found using the “equivalent prestressing load method”.
To demonstrate this, the following loads are introduced: an axial compressive force Pm
having an eccentricity ep0 at the beam-ends and an upward uniformly distributed
curvature pressure (also see fig. 4.3b):
April 2018 4-4

Chapter 4 - Bending
Pm 8 Pm f
qp  
R l2
At midspan of the beam, the bending moment comes from the eccentric axial force at the
beam-ends and the uniform upward load:
1 1 8 Pm f 2
M   Pm ep0  qp l 2   Pm ep0  l   Pm ep0  Pm f
8 8 l2
M  Pm ep
The axial compressive force Pm is also present.

The bending moment and the axial force cause the same stresses  cb and  ct as
presented in relations (4.4) and (4.5), using a cross-sectional analysis.
At first sight, the cross-section method seems to be easier to use than the equivalent
prestressing load method. However, this only holds for statically determinate beams. If
the beam is statically indeterminate, the bending moment line from the curvature
pressures no longer directly follows from the eccentricity of the tendon. This is caused by
the statically indeterminate nature of the structure: curvature pressures deform the
structure. Since the structure is statically indeterminate, these deformations cannot
develop freely. As a result, additional reaction forces and/or bending moments develop.
The actual bending moment line from the load follows from statics.
It is, therefore, strongly advised to always use the equivalent prestressing load
method, even in case of a statically determinate structure.
The equivalent prestressing load method has the following advantages:

 The method is valid for all possible prestressing, such as:
“full prestressing”, in which under SLS load no tensile stresses are allowed to occur
in the concrete,
“limited prestressing”, in which under SLS load no crack formation in the concrete is
allowed.
“partial prestressing”, in which under SLS load crack formation is accepted.
 The method gives a fast solution for statically indeterminate structures.
 The method leads to a better understanding of the behaviour of structures and is
therefore more suitable in design.
From the previous discussion, the concept “prestressing” can be summarised as follows:
“PRESTRESSING IS PRELOADING”
For the case of prestressing with post-tensioned steel (both with and without bond), the
prestressing force is the force Pm0 exerted on the beam after anchoring.
April 2018 4-5

Chapter 4 - Bending
In prestressing with pre-tensioned steel the prestressing force is the force Pmax exerted on
the abutments just before cutting (unloading) the tendons (in most cases strands). The
previously mentioned concept becomes clear when regarding the term “unloading” as
“loading” by a force having an opposite sign.
When using this approach towards the phenomenon of “prestressing” of structures, a

prestressed concrete structure is similar to a reinforced concrete structure subjected to an
external force, equal to the prestressing force. This is illustrated in figs. 4.4a and 4.4b.
yield
stress
prestressed concrete
from prestressing
reinforced concrete
Fig. 4.4 Comparison of the behaviour of a prestressed element and a reinforced

element subjected to an external axial force
Figure 4.4a shows an axially prestressed element on a prestressing bed. The tendons are
still anchored to the abutments and have a force Pmax. The stress in the prestressing steel
then is  p,max  Pmax / Ap and the stress in the concrete  c  0 .
Due the release of the prestressing force between the abutments, the force is transferred
to the concrete element. Following eq. (3.3) it holds:
 Pmax
 c 
Ac 1   e  p 
 e  p Pmax
 p 
Ap 1   e  p 
The resulting stress in the prestressing steel is:

P P  e  p Pmax Pmax
 p  max   p  max  
Ap Ap Ap 1   e  p  Ap 1   e  p 
Equilibrium of forces:
 Pmax Pmax
Ac  c  Ap  p  Ac  Ap 0
Ac 1   e  p  Ap 1   e  p 
(There is no resulting external load on the element; the internal forces are in equilibrium).
April 2018 4-6

Chapter 4 - Bending
Now, the reinforced element from fig. 4.4b is considered, in which the steel cross-section
As is equal to Ap of the prestressed element. In the unloaded state  s  0 and  c  0 .
When this element is loaded by an external force Pmax, this force is distributed over the
concrete and steel cross-sections in the same way as in the prestressed element. In both
cases, the stress increase is the same. The only difference is the initial stress  p0 in the
prestressing steel, and as a result of this, the reserve in strength relative to the yield
strength (fig.4.4c). This finding is important for the understanding of the concept of
prestressing.
4.2 The influence of the geometry of the cross-section
A statically determinate beam is prestressed using a draped tendon (fig. 4.5). The
anchorages of the tendon are in the centre of gravity (centroidal axis) of the cross-section.
Fig. 4.5 Loads on a statically determinate prestressed beam with a parabolic tendon
profile
The moment in the cross-section at midspan caused by the prestressing load (fig. 4.5a) is:
1
M p  qp l 2
8
where qp is the uniformly distributed prestressing load from the curvature pressure. Using
relation (4.3) it can be found that:
April 2018 4-7

Chapter 4 - Bending
18P f 
M p   m2  l 2  Pm f
8 l 
This result is as expected because, according to the cross-section method, as shown in fig.
4.5b, the drape f at midspan is exactly equal to the eccentricity of the tendon.
In the past it was often stated that crack formation was not permitted at SLS loading. In
that case, three boundary conditions must be met when determining the prestressing force
and the position of the centre of gravity of the prestressing in the cross-section:
 In the unloaded state (i.e. loads acting at the time of tensioning or release of
prestress), the stress in the pre-tensioned compression zone (top side of the structure
from fig. 4.5) should not exceed a specified tensile stress.
 At the same time, the stress in the pre-compressed tension zone (bottom side of the
structure from fig. 4.5) should not exceed a specified compressive stress which is
based on fck(t) (the characteristic compressive strength of the concrete at time t when
it is subjected to the prestressing force ) (EN 1992-1-1 cl. 5.10.2.2 (5)).
 At maximum load, the tensile stress in the pre-compressed tension zone should not
exceed a certain value.
A distinction is made between full, limited and partial prestressing. In case of full
prestressing, no tensile stresses should occur at all. In case of limited prestressing, small
tensile stresses are allowed, whereas partial prestressing enables crack formation in SLS.
A discussion on partial prestressing is given in Chapter 12.
zct
kt
kern area e
x kb
F
zcb
Fig. 4.6 The kern area of a cross-section
The effects of the geometry of the cross-section are now analysed. For that purpose, the
concept “kern area” has to be explained first (see fig. 4.6).
The kern area of a cross-section is the part of the cross-section where an axial
compressive force has to be applied to avoid the occurrence of tensile stresses in the
cross-section.
April 2018 4-8

Chapter 4 - Bending
The position of the bottom point of the kern area is determined by making the stress in
the top fibre of the cross-section equal to zero:
F  F e  zct
 ct    0
Ac Ic
From which it follows:
I c zct Wct
e   kb (4.6)
Ac Ac
For the position of the upper point of the kern area, the stresses at the bottom side of the
cross-section are set to zero. Solving a similar expression for the eccentricity results in:
I c zcb Wcb
e   kt (4.7)
Ac Ac
ct
F
kt
kb
F
cb
a b c
Fig. 4.7 Stress distribution from an axial compressive force for different lines of
action
The stress distribution can simply be determined as soon as the positions of the outer
points of the kern area are known. When the axial force is applied in the upper kern area
point, the stress  cb at the bottom fibre is zero (fig. 4.7a), and when the load is applied in
the lower kern area point, the stress  ct at the top side of the cross-section is zero (fig.
4.7b).
For loads applied at arbitrary positions in the cross-section, the stress distribution can be
calculated using the position of kern area points. In case of a compressive force F having
an eccentricity e relative to the centre of gravity, it is found:
 F  e  kt  F  e  kb 
 cb  ;  ct 
Wcb Wct
April 2018 4-9

Chapter 4 - Bending
In the following analysis the option “fully prestressed concrete” is considered. The
statically determinate beam from fig. 4.3 has a tendon drape ep. From the “prestressing is
preloading” approach, it followed that the midspan bending moment from prestressing is
Pmep. The axial compressive force Pm at centroidal axis (neutral axis) level is present over
full beam length. In the midspan cross-section the combination of the force Pm and the
bending moment Pmep is replaced by a force Pm that has an eccentricity ep (fig. 4.3a &
fig. 4.7).
The governing situation for the top fibre of the beam occurs when it has its permanent
load G only. Because tensile stresses are not allowed, the requirement is:
Pm M g Pm ep M Pm  ep  kb 
 ct      g  0
Ac Wct Wct Wct Wct
or:
Mg
e p  kb   M g  Pm  e p  kb  (4.8)
Pm
The governing situation for the extreme bottom fibre occurs at full loading G + Q:
Pm M g  M q Pm ep M g  M q Pm  ep  kt 
 cb       0
Ac Wcb Wcb Wcb Wcb
or:
Mg  Mq
ep  kt   M g  M q  Pm  ep  kt  (4.9)
Pm
Equations (4.8) and (4.9) provide the information to determine the zone in which the
prestressing force must be applied, see fig. 4.8.
April 2018 4-10

Chapter 4 - Bending
kt
kb
admissible area
Fig. 4.8 Area in which the line of action of the prestressing force should be situated
From this analysis, a number of interesting conclusions can be drawn with respect to the
effectiveness of the shape of the cross-section.
The minimum moment required from static loading G follows from equation (4.8):
M g,min  Pm  ep  kb 
This implies:
“The greater the distance between the prestressing steel and the lower kern area point,
the greater the bending moment capacity for static loading”.
From eqs. (4.8) and (4.9) the maximum bending moment allowed from the live load Q is
obtained:
M g  M q  Pm  ep  kt  (from equation (4.9))
Mg  Pm  ep  kb  (from equation (4.8))


Mq  Pm  kt  kb 
This implies:
“The greater the height of the kern area in the cross-section, the greater the moment
capacity for live loads”.
In fig. 4.9, a number of shapes having the same cross-sectional area are compared. The
prestressing force is kept constant.
April 2018 4-11

Chapter 4 - Bending
1 2 3 4
Ac [mm2] 384·103 384·103 384·103 384·103

kt [mm] 187 164 309 407
kb [mm] 187 282 309 407
epmax [mm] 475 592 480 780
Mg [kNm] 53,3 σcp 81,9 σcp 33,0 σcp 71,8 σcp
Mq [kNm] 71,7 σcp 62,9 σcp 118,5 σcp 156,1 σcp
Mg+q [kNm] 127,0 σcp 144,8 σcp 151,5 σcp 227,8 σcp
Fig. 4.9 Data for several cross-sections; bending moment resistance for a prestressing
force Pm = 1/2 Ac σcp
From this comparison, it can be concluded that cross-sections that contain one
compression flange and have a high permanent load (e.g. selfweight) can sustain only
relatively small variable loads. On the other hand, symmetrical cross-sections are well-
suited for resisting low permanent loads combined with high variable loads (for example
crane beams).
The most obvious conclusion from fig. 4.9 is that it is the most economic to maximise the
height of the section. However, the following aspects make that this is often not the case:
 There is a lower limit for the thickness of the web, because of the required minimum
concrete cover on the web reinforcement (e.g. stirrups and longitudinal web
reinforcement) and the required shear capacity (of the concrete struts loaded in
compression).
 The beams should have a specific minimum stiffness in the lateral direction to
prevent lateral torsional buckling to occur (See EN 1992-1-1 cl. 5.9 1(P): ‘Lateral
instability of slender beams shall be taken into account where necessary, e.g. for
precast beams during transport and erection, for beams without sufficient lateral
bracing in the finished structure etc’).
April 2018 4-12

Chapter 4 - Bending
 In beams without a bottom flange and having a small web thickness, the prestressing
elements must be arranged vertically. This raises the centre of gravity of the
prestressing, which has an unfavourable effect on the effectiveness of the
prestressing (e.g. reduction of the drape).
 In order to minimise the total construction costs, it might be desired to keep the total
structural height as small as possible.
It must be emphasised that a basic assumption in the previous analysis is a “fully

prestressed” structure, where no tensile stresses under any possible SLS load combination
are permitted.
For partially prestressed concrete different criteria hold. Therefore, for partial prestressed
concrete, especially at low levels of prestressing, some caution is required with respect to
the previous conclusions.
4.3 Prestressing with pre-tensioned steel

For pre-tensioned steel, the element shortens when the prestressing force of the
abutments is transferred to the concrete element. The results are changes in the stresses of
steel and concrete. The directly occurring stress reduction in the steel is often referred to
as “elastic prestress losses”. The order of magnitude of the elastic shortening and the
corresponding stress change are important when examining the stresses.
A beam is assumed to be prestressed using straight tendons having an eccentricity ep. A

slice from this beam is studied in detail (fig. 4.10). Note that the prestressing is assumed
to result in an axial force and bending moment (following from the “equivalent
prestressing load method”) that correspond with an eccentric force applied at the level of
the prestressing steel. As discussed before in this chapter, the eccentric force is exactly at
the level of the steel in case of a statically determinate structure.
Fig. 4.10 Prestressing with pre-tensioned steel: situation before (a) and after (b) the
tensioning
Before the prestressing force is transferred to the beam, the stress in the steel is σpmax =
Pmax / Ap. The element shortens due to the transfer of the prestressing force to the
April 2018 4-13

Chapter 4 - Bending
element. In the steel a compressive force is generated. This force is assumed to be -ΔPel
and must be added to the prestressing force Pmax. This implies that a force Pmax - ΔPel is
exerted on the net concrete cross-section. This causes the following shortening of the
concrete at the steel level:
c  P  Pel  Pmax  Pel  ep  ep  x  1 ep2  x

lc  x   max     Pmax  Pel    
Ec  Ac Ic A I E
  Ec  c c  c
The shortening of the steel is:
Pel x
lp 
Ap Ep
Because it should hold that lc  lp , the extra compressive force transferred to the steel
is:
 e p f
Pel  Pmax (4.10)
1   e p f
where:
 Ac ep2  E Ap
f  1   ; e  s ; p 
 I c  Ec Ac

This force causes a reduction of the steel stress.
The force transferred to the concrete:
1
Pmax  Pel  Pmax (4.11)
1   e p f
Example
A hollow-core slab is prestressed by seven strands Ø12,5 mm (Ap = 93 mm2/strand),
Y1860S7 (fpd = 1522 N/mm2; table 2.5). The concrete strength class aimed at is C35/45,
but is only C20/25 at the time when the prestressing is applied to the concrete.
The dimensions of the hollow-core slab are given in fig. 4.11.
April 2018 4-14

Chapter 4 - Bending
Fig. 4.11 Cross-section of a prestressed hollow-core slab (dimensions mm)
Geometrical data:
Ac  150 103 mm 2 ; Ap  7  93  651 mm 2

I c  690 106 mm 4 ; ep  60 mm
From table 2.5 it can be read that the maximum steel stress in SLS must be
σpm0 < 1395 N/mm2. During prestressing it is allowed to use σpmax = 1488 N/mm2. A
further increase to σpmax = 1590 N/mm2 is allowed by EN 1992-1-1 cl. 5.10.2.1 (2)
provided that the jack force is accurately measured (note: The NL National Annex does
not allow this increase).
The increase of the steel stress to σpmax can compensate for the stress losses caused by
wedge set when anchoring the strands, by friction and elastic shortening of the concrete.
Young’s modulus of concrete strength class C35/45 having reached a class C20/25
strength (EN 1992-1-1 table 3.1 & eq. (3.5)):
0,3 0,3
 f (t )   20  8 
Ecm (t )   cm  Ecm     34 10  29,9 10 N/mm
3 3 2
 f cm   35  8 
Es 195 103
e    6,52
Ec 29,9 103
Ap 651
p    4,34 103
Ac 150 10 3
Ac ep2 150 103  602

f  1  1  1, 78
Ic 690 106
By using eq. (4.10) ΔPel is calculated:
 e p f 6,51 4,34 103 1, 78

Pel  Pmax  Pmax  0, 048 Pmax
1   e p f 1  6,51 4,34 103 1, 78
April 2018 4-15

Chapter 4 - Bending
For Pm0 it can be derived:
Pm0  Pmax  Pel  1  0, 048  Pmax
and:
 p0  1  0, 048   pmax
The maximum initial stress is:
 p0 1395
 pmax    1465 N/mm 2
1  0, 048 1  0, 048
So, it is not possible to make full use of the stress increase to σpmax = 1488 N/mm2 since
this will cause a too high σp0 (restricted to 1395 N/mm2).
The initial prestressing force and the losses due to the elastic shortening are:
Pmax  Ap  pmax  651 1465  954 103 N

Pel  0, 048 Pmax  0, 048  954 103  45,8 103 N
 p,el  0, 048 pm ax  0, 048 1465  70,3 N/mm 2
The force transferred to the concrete:
1
Pmax  Pel  Pmax  0,952  954  908 kN
1   e p f
The maximum compressive stress in the pre-compressed tensile zone (bottom fibre,
assumed to be at a distance h / 2 from the centroidal axis):
 1 ep  12 h  3  1 60  12  200 
 cb    Pmax  Pel      908  10    6 
 13,9 N/mm 2
 150 10 690 10 
3
 Ac Ic 
When applying the prestressing force to the slab (during the release of the stress in the
strands at the abutment), the selfweight of the slab will be activated; the slab bends
upwards because of the eccentric prestressing load, and, as a result, will be supported
only at its ends. From that moment on, the bending moment caused by the selfweight is
activated.
However, at the ends of the slab the effect of the selfweight is zero, so that the previously
calculated concrete stress applies at these sections.
EN 1992-1-1 cl. 5.10.2.2 (5) requires that the initial concrete compressive stress is less
than:
 c  0, 70 f ck (t )  0, 70  20  14, 0 N/mm 2
April 2018 4-16

Chapter 4 - Bending
where fck(t) is the characteristic cylinder compressive strength at the moment the
prestressing is applied to the concrete. This requirement is just met. In most cases, the
slabs are prestressed quite quickly after casting (for example at C20/25). After continuing
hardening of the concrete, a higher strength class will be reached (for example C45/55).
4.4 Prestressing with post-tensioned steel

In the case of prestressing with post-tensioned steel, the tendons are positioned in empty
ducts that are installed before casting. After hardening of the concrete, the tendons are
applied (or were already in the ducts before casting) and stressed. The ducts are injected
with a special grout. After hardening of the grout, the cross-sections behave like
composite cross-sections: steel and concrete work together. In that stage the prestressing
steel behaves in the same way as the reinforcing steel in a non-prestressed structure.
4.4.1 The transmission of the prestressing forces
When the prestressing force is exerted on the post-tensioned steel, this force does not act
on the composite cross-section, because there is no bond yet between the prestressing
steel and the concrete. Due to the prestressing, the beam will deform, which activates the
selfweight as a load, see fig. 4.12.
centroidal axis
formwork formwork
Fig. 4.12 Prestressing with post-tensioned steel

a. Situation before applying the prestressing force
b. Situation after applying the prestressing force
When at the manometer of the prestressing jack a force Pmax is indicated, the stresses in
the concrete at midspan, at the bottom and at the top are:
Pmax M g Pmax ep
Bottom:  cb     (4.12a)
Ac Wcb Wcb
Pmax M g Pmax ep
Top:  ct     (4.12b)
Ac Wct Wct
April 2018 4-17

Chapter 4 - Bending
where Wcb and Wct are the section moduli of the concrete cross-section relative to the
bottom and top fibre respectively, and Mg is the moment caused by activating the
selfweight of the structure by upward bending.
To make sure that the intended stress in the prestressing steel is actually applied, not only
the force in the jack is registered, but also the elongation of the tendon relative to the
concrete surface is measured.
To calculate the expected elongation of the tendon, the beam is supposed to be divided
into small segments (fig. 4.13a,c). The x-axis is horizontal; the s-axis follows the tendon
profile.
1

Ec
Fig. 4.13 Calculation of the elongation of the tendon relative to the concrete surface
It should be noted that prestressing losses occur as a result of friction between the tendon
and the duct, see fig. 4.13b. The calculation of these losses is dealt with in section 4.4.3.
The elongation of the tendon relative to the concrete surface is the elongation of the
prestressing steel itself (lp > 0) and the shortening of the concrete (lc < 0) at the level
of the tendon. When introducing the (+)-sign for the elongation of the steel and the (-)-
sign for the shortening of the concrete, the relative displacement of the tendon (lp,rel) is:
lp,rel  lp  lc (  slip of the anchor) (4.13)
The elongation of the prestressing steel:

l l
Pmax ( s )
lp    p ( s ) ds   ds (4.14)
0 0
Ep Ap
April 2018 4-18

Chapter 4 - Bending
The shortening of the concrete at the level of the prestressing steel in an arbitrary
segment (fig. 4.13c), is:
 1 ep2 ( s )  1
 cp  Pmax ( s ) cos( ( s ))   
 Ac I c  Ec
In a slender beam, the inclination of the prestressing tendon is relatively small. As a

result, cos((s))  1,0. Then, x instead of s can be used. The shortening of the concrete
lc has two components, namely shortening caused by the prestressing force and an
elongation caused by activating the selfweight of the beam:
l
Pmax ( x)
l
 P ( x) ep ( x)  M g ( x)
lc    dx   max ep ( x) dx (4.15)
0
Ec Ac 0
Ec I c
The elongation of the tendon relative to the concrete:

l
Pmax ( x)
l
P ( x)
l
P ( x) ep ( x)  M g ( x)
lp,rel   dx   max dx   max ep ( x) dx (4.16)
0
Ep Ap 0
Ec Ac 0
Ec I c
In most cases the contribution of the third integral from eq. (4.16) is relatively small and
can be ignored:
 1 1 
l
 1 1 
lp,rel   
E A E A    Pmax ( x) dx   
 E A E A  mm
P l (4.17)
 p p c c  0  p p c c 
where Pmm is the mean value of Pmax over the length l.
If the elongation of the prestressing steel during prestressing is not checked, this may lead
to damage to the structure, as will be shown in the following example.
During the prestressing of a beam (fig. 4.14), flexural cracks occurred at the top, although
the manometer indicated that the intended level of prestressing was not yet reached. It
appeared that the damage was caused by a rather flexible formwork system. For that
reason the full selfweight of the structure was not activated immediately during
prestressing, but was gradually activated. This resulted in a too high upward load and, as
a result, unexpected high tensile stresses at the top of the beam, which resulted in the
observed cracks.
April 2018 4-19

Chapter 4 - Bending
Fig. 4.14 Unexpected cracking occurring during the prestressing of a beam
If one would have calculated and controlled the elongation of the prestressing tendon, this
deviation would already have been discovered during prestressing. It would be clear that
the measured elongation relative to the concrete (eq. (4.16)) did not correspond with the
magnitude of the prestressing force, as indicated by the manometer (a relatively too high
elongation). This can be explained using eq. (4.16): the last integral should have been
small, but was relatively high since a part of the selfweight was not active yet (a too small
Mg).
This example underlines the meaning of the proposition:
“Prestressing is deforming”
4.4.2 Prestressing losses caused by elastic deformation during the prestressing of a
member with more than one prestressing tendon
When applying a prestressing force, not only the beam will shorten, but also the tendons
that were tensioned before. So, before injecting the ducts, already a part of the tendons
have lost part of their stress. This is illustrated by fig. 4.15.
Fig. 4.15 Cross-sectional and side view of a prestressed beam with four prestressing
tendons
The beam shown in fig. 4.15 has a cross-sectional area Ac = 1,0·106 mm2, a modulus of
elasticity Ec = 30·103 N/mm2 and a prestressing force per tendon Pm = 875 kN (Ap =
902 mm2).
When discussing the relative elongation of the prestressing tendon during prestressing
(eq. (4.16)), it was already pointed out that the last integral can be ignored in most
practical situations. Therefore, the effect of bending of the beam will be ignored in this
April 2018 4-20

Chapter 4 - Bending
case and only the axial deformations are taken into account. If a prestressing tendon is
stressed to a force Pm, the shortening of the concrete is:
Pm l
lc  
Ec Ac
The total shortening lpel of the first prestressing tendon caused by the prestressing of the
following (n - 1) tendons is:
Pml
lpel  (n  1)
Ec Ac
The loss of prestressing force in the first tendon is:
Ep Ap
Pel  (n  1) Pm (4.18)
Ec Ac
The total loss for n tendons is calculated by summing up the losses of all individual
tendons. Note that each tendon has a different number of tendons prestressed after the
tendon considered is stressed: n - 1 ; n – 2; n – 3 ................... n - n (= 0). The mean value
of all these individual factors is (n – 1) / 2; their sum is n (n – 1) / 2. The total force loss
is:
n (n  1) Ep Ap
Pel  Pm (4.19)
2 Ec Ac
In EN 1992-1-1 cl. 5.10.5.1 & eq. (5.44) the average loss per tendon is presented:
(n  1) Ep Ap
Pel  Pm
2 Ec Ac
In this example four prestressing tendons are applied. The losses, calculated using eqs.
(4.18) and (4.19) are listed in table 4.1.
The losses can be compensated by slightly overstressing the tendons. According to EN

1992-1-1 cl. 5.10.2.1 (1) a 5% overstressing is allowed provided that the jack force is
accurately measured. In the Dutch National Annex, overstressing is not allowed. If
overstressing is not allowed or insufficient, the post-tensioning of tendons which were
tensioned before, is an option.
April 2018 4-21

Chapter 4 - Bending
Table 4.1 Losses per prestressing tendon caused by stressing of the following
tendons
tendon loss Pm losses
[kN] [%]
1 15,8 1,8
2 10,5 1,2
3 5,3 0,6
4 - -
total 31,6 0,9
During the prestressing of post-tensioned steel, especially for curved tendons, a part of
the prestressing force is lost due to friction between the tendon and the duct. Therefore,
the prestressing force is not constant along the length of the structure, see for example
fig. 4.13b.
Figure 4.16a shows a beam with a curved tendon profile. A detail is shown in fig. 4.16b.
As already described in section 4.1, the equivalent prestressing load is:
Pm ( x)
qp 
R
Using a friction coefficient  for the interface between tendon and duct, the frictional
force is:
Pm ( x)
 qp  
R
In fig. 4.16c the same part of the tendon having a length R θ is shown again. From
equilibrium considerations it follows that:
Pmax ( x)
Pμ ( x)    R 
R
From the integration
Pμ ( x)
P max ( x)
    
it follows that:
April 2018 4-22

Chapter 4 - Bending
ln( Pmax ( x))      C (4.20)
Using the boundary conditions Pmax(x) = Pmax and θ = 0 for x = 0 it is found that:
C  ln( Pmax ) (4.21)
tensioned side / active end
anchor
Pm ( s)
Pm ( s )  Pm ( s )
Pm ( s )
duct
tendon Pm ( s )  Pm ( s )
Fig. 4.16 Friction between prestressing tendon and duct

(a) Forces exerted when stressing the tendon
(b) Friction forces and curvature pressure exerted on the concrete
(c) Friction forces and curvature pressure exerted on the tendon
Substituting eq. (4.21) in eq. (4.20) and replacing Pmax(x) by Pmax(θ) results in:
Pmax ( )
ln   
Pmax
or
Pmax ( )  Pmax e    (4.22)
In this equation θ is the total angular rotation (in radians) of the tendon, calculated from
the location of the prestressing jack. The coefficient of friction  depends on the
prestressing system. For prestressing strands   0,10 – 0,24; for bars   0,33 – 0,65
(EN 1992-1-1 table 5.1).
April 2018 4-23

Chapter 4 - Bending
Measurements of the frictional losses in practice demonstrated, however, that eq. (4.22)
mostly results in too low values for the losses. This is caused by the actual angular
rotation being larger than assumed in the calculation.
Therefore, a distinction is made between the “intentional” angular rotation θ and the
“unintentional” angular rotation k, which is caused by the so-called “Wobble-effect”.
The Wobble-effect is caused by unintended curvatures of the prestressing duct, and the
deflection of a “straight” tendon between its supports, see fig. 4.17. Prestressing systems
used in practice have an angle k due to the Wobble effect that is in the range of 0,005 –
0,01 rad/m (EN 1992-1-1 cl. 5.10.5.2 (3)).
The values for  and k follow from documentation of the prestressing system used.
placing the tendon in the duct
Ø duct
deflection of the duct
tendon profile = duct profile
at prestressing:
stressing the tendon in the duct
Ø duct
deflection of the duct
tendon stretches
Fig. 4.17 Origin of the Wobble-effect by the angular rotation k according to Theile
[4.1]
In general, the following expression is used to determine the frictional losses (EN 1992-
1-1 eq. (5.45)):
Pμ ( x)  1  e   (  k x )  Pmax (4.23)
Here k is the angular rotation (in rad) per unit of length caused by the Wobble-effect, and
x is the distance from the location of the prestressing jack to the cross-section considered.
April 2018 4-24

Chapter 4 - Bending
For  (  k x)  0,30 , Pmax(x) can be linearised to
Pmax (x) = Pmax 1   (  k x ) (4.24)
Sometimes it is doubted whether combining the intended angular rotation θ and the
unintended angular rotation k is correct. From a physical point of view it is probable that
there is a certain interaction between k x and θ. For large intended angular rotations θ the
unintended angular rotation k x should be almost 0.
Therefore Walter, Utescher and Schreck [4.2] proposed not to use the sum of θ and k x
but to use only the largest one. That would imply:
  k x  Pmax ( x)  Pmax e   (4.25a)
  k x  Pmax ( x)  Pmax e  k x (4.25b)
Equation (4.25b) was found to be only valid for curved tendon profiles which meet the
condition:
1
R (4.25)
k
where R is the radius of curvature of the tendon and k is the angular rotation due to the
Wobble-effect (in rad/m).
4.4.4 Wedge set / anchorage losses
When anchoring using wedges (fig. 4.18), some wedge slip (wedge set) occurs when
releasing the jack. The wedges slip until the anchorage force is fully developed. This
results in a loss of prestress as shown in fig. 4.18. The slip of strands or wires is generally
in the range of 5 – 15 mm.
Wedge set is calculated using Hooke's law:

1
P2lw
w 2
(4.26)
Ep Ap
In this expression both ΔP2 and lw are unknown. The stress reduction in the prestressing
tendon caused by friction is assumed to be linear. Therefore, the following expression
holds:
 pμ,m
P2  2 Ap lw (4.27)
x
April 2018 4-25

Chapter 4 - Bending
where Δσpμ,m/Δx is the mean stress reduction in the tendon caused by friction over the
length lw. (Note: Units, for example, N/mm2/mm). Expression (4.26) can now be
rewritten as:
Ap  pμ,mlw2
w
Ep Ap x
tensioned side / active side
ΔP1 = force loss from friction

ΔP2 = force loss from wedge set at the
anchor
lw = length over which wedge set
influences the prestressing force
Fig. 4.18 Loss of prestress due to (wedge) slip of the anchor at the tensioned side
Now, the length over which wedge set influences the tendon force is known:
w Ep
lw 
 pμ,m
x
The force loss at the position of the anchor is calculated using expression (4.27). The
expression derived can not only be used when the friction is constant over the length lw,
but also in case of a profile with a changing radius of curvature or kinks. In the latter
case, it is mostly required to iterate to find a solution: after an initial estimation of length
lw, Δσpμ,m/Δx can be calculated and, as a result of this, lw is recalculated. If the calculated
and the estimated values are almost identical, the required distance lw is found.
4.5 The fictitious cross-section

In the case of full or limited prestressing, where the structure should remain uncracked in
SLS, a stress check is required. After injecting the prestressing ducts with grout and the
hardening of the injection grout, prestressing steel and concrete fully work together in
resisting forces. When calculating concrete stresses, the cross-sectional properties of the
composite cross-section must be used. This implies that the Young’s modulus of the
materials must be used in the expressions (“E-modulus weighed”). This holds for
properties like the cross-sectional area (A), the moment of inertia (I) and the section
April 2018 4-26

Chapter 4 - Bending
moduli (W). Its implication for practice is illustrated for the cross-section shown in fig.
4.19.
Fig. 4.19 Data of the composite cross-section (dimensions mm)
The cross-section is prestressed with tendons consisting of strands Ø12,9 mm, having a
total cross-sectional area Ap = 3600 mm2. The diameter of the prestressing ducts is
60 mm. If only the concrete in the cross-section is regarded, the following cross-sectional
properties are calculated:
Ac = 0,443 m2 zc = 0,657 m Ic = 7,8710-2 m4

Wct = 14,510-2 m3 Wcb = 12,010-2 m3
For the composite cross-section, having a concrete strength class C35/45, the following
results are obtained:
αe = Ep / Ec = 195·103 / 34,0·103 = 5,74
Ai = Ac + (αe -1) Ap = 0,443 + (5,74 – 1) · 3,60 · 10-3 = 0,460 m2
zc Ac  ( e  1) zp Ap 0, 657  0, 443  4, 74  0, 08  3, 60 103

zi    0, 636 m
Ai 0, 460
I i  I c  Ac ( zi  zc )2  ( e  1) Ap ( zi  zp ) 2  I p 
7,87 102  0, 443  0, 0212  4,97  3, 60 103  0,5562  0  8, 44 102 m 4
concrete top fibre, section modulus:

Wit  I i / (h  zi )  8, 44 102 / (1, 2  0, 636)  15, 0 102 m3
April 2018 4-27

Chapter 4 - Bending
concrete bottom fibre, section modulus:

Wib  I i / zi  8, 44 102 / 0, 636  13,3 102 m3
The SLS stresses in the concrete (if uncracked) follow from:
top fibre: ct = - M / Wit
bottom fibre: cb = +M / Wib
In such calculations mostly only the properties of the concrete section are used, not
taking into account the influence of prestressing and reinforcing steel. This is a
conservative simplification. In the case considered, the concrete stresses calculated using
the composite (fictitious) section (caused by M) are about 7% smaller than those
calculated using the concrete cross-section only.
The fictitious cross-section is not often used in design, since all dimensions must be
known in detail. It might be used in a detailed design check, using a standard computer
program.
4.6 Choice of the shape and dimensions of the cross-section

One of the tasks of the designer is to chose, for the application considered, the most
appropriate shape and dimensions of the cross-section. The choice largely depends on the
experience of the designer. However, a number of common guidelines can be given. In
section 4.2 the influence of the shape of the cross-section on the bending resistance was
dealt with. It turned out that there are differences in the efficiency of cross-sections in
different applications.
T-shaped cross-sections are mostly applied in statically determinate structures, in which

the selfweight of the structure is (considerably) higher than the variable load. Therefore,
T-shaped sections are especially suited in large span structures. In such a cross-section
the lower point of the kern area is positioned that high in the cross-section, that a large
part of the permanent load can be carried by the prestressing force, applied between the
lower kern area point and the bottom fibre.
I-shaped cross-sections are often applied for smaller spans, especially if the variable load
is significantly higher than the permanent load. As a result of the large height of the kern
area, relatively high variable loads can be sustained at a relatively small prestressing
force.
Box girder beams (fig. 4.20), which can in fact be calculated as being T- or I beams, are
mostly applied for large to very large spans. Furthermore, they are especially suited for
carrying torsional moments.
Rectangular beams and slabs are only applied in the case of small spans.
April 2018 4-28

Chapter 4 - Bending
In statically indeterminate structures, having negative (hogging) moments at the supports

and positive (sagging) moments in the spans, symmetrical cross-sections, I-shaped beams
and box girders are the most efficient options.
Other requirements and demands can, of course, also have impact on the choice of the
optimal shape of the cross-section.
Fig. 4.20 Examples of box girder bridges with one (a) and two (b) boxes and practical
dimensions
Figure 4.21a shows the cross-section of a traffic bridge. The shape of the precast beams
makes that no scaffolding/formwork has to be used when casting the concrete in-situ.
After casting, a solid slab bridge is obtained.
Figure 4.21b shows a partially precast bridge deck. For carrying the load, the lower
flanges are actually not required. Nevertheless they are applied to provide the viaduct
with sufficient resistance against transverse collision loads caused by passing vehicles,
see fig. 4.22.
The ratio between the depth of the cross-section and the span (h / l) ranges from l/20 (for a
solid cross-section) to 1/35 (for an I-beam or a box girder).
April 2018 4-29

Chapter 4 - Bending
cast-in-situ
bridge deck concrete prestressed T-girder
bridge deck formwork hollow space
b
possible
impact load
bearing block
Fig. 4.21 Prestressed T-beams for a precast bridge system
Fig. 4.22 Damage to viaducts caused by vehicle collision
4.7 Calculation example of a fully prestressed beam

4.7.1 Introduction
A bridge beam has a span of 21,5 m. Apart from the selfweight qg of the structure itself,
there is an additional permanent load qg = 2,5 kN/m’ (from the bridge deck and the
parapets) and a variable load qq = 12 kN/m’. The concrete strength class is C35/45 and
the environmental class is XD3. In this example full prestressing is applied, which
implies that in SLS no tensile stresses in the concrete should occur.
April 2018 4-30

Chapter 4 - Bending
qq = 12 kN/m
qg = 2,5 kN/m
Fig. 4.23 Scheme and loads
The structural depth of the cross-section is estimated at l / 20  1,0 m. From production

facility requirements flange width equals b = 800 mm. A lower flange (wider than the
web of the beam) is applied for the prestressing cables. The concrete cross-sectional
dimensions, and the corresponding cross-sectional properties, are given in fig. 4.24.
zct
zcb
Ac = 0,328 m2 Wcb = 0,0703 m3 zcb = 0,566 m qg = 8,2 kN/m

Ic = 0,0398 m4 Wct = 0,092 m3 zct = 0,434 m concrete C35/45
prestressing steel Y1860S7
Fig. 4.24 Cross-sectional dimensions and properties (dimensions mm)
For the SLS, the following stress limits are used:

 concrete tensile stress: c < 0 (only compression)
 concrete compressive stress, according to EN 1992-1-1 cl. 5.10.2.2:
 c  0, 6 f ck   c  0, 6  35  21 N/mm 2
Note that if the concrete compressive stress is permanently over 0,45 fck, non-linear creep
has to be accounted for.
April 2018 4-31

Chapter 4 - Bending
The maximum stresses allowed in the prestressing steel Y1860 are given in table 2.5:
 Initial stress after anchoring pm0 < 1395 N/mm2
 Stress during prestressing p,max < 1488 N/mm2 (no overstressing)
In this example it is assumed that the prestress losses caused by elastic shortening of the
concrete are compensated for by overstressing. This will, therefore, not be dealt with in
this example. Prestress losses caused by shrinkage and creep of concrete are also not
dealt with. Creep and shrinkage are discussed in Chapter 6. Tendon friction is first
neglected; it will be introduced in detail in a following section.
As a start, it is assumed that the total time-dependent losses are 20%. This implies that
the prestress is pm∞ = 0,8 · 1395 = 1116 N/mm2. The prestressing force is now
determined such that the requirement on "no concrete tensile stress" is met. This criterion
is mostly governing in the design. Afterwards, it will be checked whether the conditions
with regard to the concrete compressive stresses are met as well. Furthermore, it will be
checked whether the maximum tensile stress in the prestressing steel meets the specified
limits.
The prestressing is assumed to be applied at the centroid of the cross-section at both

supports, see fig. 4.25 (i.e. both anchors are at neutral axis level). The tendons have a
curved, parabolic profile. The drape is ep.
In this example the main item is the control of the concrete stresses in this statically
determinate beam. In the first step, prestressing is regarded as an external load, see fig.
4.5a.
Fig. 4.25 Loads as a result of prestressing a tendon that has a parabolic profile
Prestressing P as an external load introduces:

 An axial compressive force P.
P P 8 Pep
 A uniformly distributed upward load qp   2  2
R l l
8ep
The total upward force form the curvature pressure qp is balanced by the two downward
concentrated reaction forces at the anchors (Psinα ≈ Pα each).
April 2018 4-32

Chapter 4 - Bending
The load qp introduces a parabolic bending moment line, having a maximum at midspan:
M p  18 qp l 2  P ep
The concrete stresses at top (ct) and bottom (cb) are now calculated:
 Concrete stress at the top fibre of the cross-section as a result of prestressing:
Pm Pm ep
 ct   
Ac Wct
 Concrete stress at the bottom fibre as a result of prestressing:
Pm Pm ep
 cb   
Ac Wcb
The representative values of the bending moments in SLS at midspan:
selfweight:
ME,g = 1/8 · 8,2 · 21,52 = 474 kNm
static load:
ME,g = 1/8 · 2,5 · 21,52 = 144 kNm
total permanent loading:
ME,g = 474 + 144 = 618 kNm
variable load:
ME,q = 1/8 · 12,0 · 21,52 = 693 kNm
Frequent load combination:
ME,freq = ME,g + ψ1 ME,q = 618 + 0,75 · 693 = 1138 kNm
Note:
It is assumed that the variable load is caused by a crowd of people. Then, ψ1 = 0,75 (EN
1990 defines the frequent and quasi-permanent load parts of a variable load).
For the stress in the concrete at the top fibre of the cross-section, the construction stage is
governing. In that case only the selfweight of the beam and the full initial prestressing
force Pm0 (directly after prestressing, so without time-dependent prestress losses) are
present. The condition is:
April 2018 4-33

Chapter 4 - Bending
Pm0 Pm0 ep M E,g

 ct     0
Ac Wct Wct
or (units: kN, m):
Pm0 103 Pm0 10  ep 474 103

3
   0
0,328 0, 092 0, 092
This results in the condition:
474
ep   0, 28 (4.28)
Pm0
At the bottom fibre, the highest tensile stress occurs if the maximum moment is applied
on the beam, whereas at the same time, the time-dependent prestress losses have fully
developed. At a working prestressing force Pm,∞ = 0,8 Pm0 the result is:
Pm, Pm , ep M E,freq
 cb     0
Ac Wcb Wcb
or
0,8Pm0 103 0,8 Pm0 10  ep 1138 103
3
 cb     0
0,328 0, 0703 0, 0703
which results in the condition:
1423
ep   0, 21 (4.29)
Pm0
In fig. 4.26 the relation between the prestressing force Pm0 and the eccentricity ep is
shown. The shaded area shows combinations of Pm0 and ep allowed from the stress
requirements.
In fig. 4.24 it can be estimated that the maximum possible eccentricity of the prestressing
steel to the centre of gravity of the cross-section is ep,max = 0,44 m. This extra limit is also
shown in the diagram using a horizontal dashed line. This is an additional limit to
possible combinations of ep and Pm0.
April 2018 4-34

Chapter 4 - Bending
Fig. 4.26 Relation between the initial prestressing force and the eccentricity ep
From economy reasons the smallest possible prestressing force is chosen. In combination
with ep = 0,44 m, eq. (4.29) results in Pm0 ≥ 2189 kN and eq. (4.28) gives Pm0 ≤ 2963 kN.
For an initial prestress p0 = 1395 N/mm2 (Y1860) the required cross-sectional area of the
prestressing steel Ap = 2189·103 / 1395 = 1569 mm2. If four tendons are applied, each
tendon should have a cross-sectional area of at least 1569 / 4 = 392 mm2. In this example
tendons, each of them containing 5 strands Ø13,0 mm (Ap = 500 mm2) are chosen (total
Ap = 2000 mm2). The duct diameter is 45 mm.
Figure 4.27 shows the four tendons in the cross-section. The tendons are positioned such
that their centre of gravity is 125 mm from the outermost concrete fibre. Mean tendon
eccentricity is ep = 441 mm (zcb = 0,566 m).
April 2018 4-35

Chapter 4 - Bending
ep =
441
Fig. 4.27 Cross-section of beam and tendon positions (dimensions mm)
Now, the amount of prestressing steel is known and stress checks are performed. The
calculations were performed not taking into account friction losses. Tendon friction is
now dealt with in detail.
4.7.2 Calculation of the losses due to friction
Already during prestressing, the stress in a tendon varies along the length of the beam.
This is caused by the friction losses. In the calculation a parabolic profile of the
prestressing tendons was assumed. Furthermore, it was assumed that the prestressing
force is applied at the centre of gravity of the gross cross-section at the beam ends. The
drape of the tendons then is f = 0,441 m (the ep from fig. 4.27). The layout of the tendons
can be described by the relation (fig. 4.28):
4f 2
y 2
x  0, 0038 x 2 where f is the drape.
l
Fig. 4.28 Parabolic layout of the prestressing tendons
April 2018 4-36

Chapter 4 - Bending
If the beam is prestressed from one side, the prestressing force varies according to the
expression (EN 1992-1-1 eq. (5.45)):
Pmax ( x)  Pmax e   (  k x ) (4.30)
where x is the distance from the position where the jack is applied (active end) to the
cross-section considered. It is assumed that the coefficient of friction is  = 0,15 and that
the angular rotation caused by the Wobble effect is k = 0,005 rad/m’.
The inclination of the tendon can be calculated in any cross-section from its layout,
according to
 dy  8f x
   2 x (4.31)
 dx  l R
The change in inclination between two consecutive points is:
x
 
R
For a number of points along the beam the magnitude of the prestressing force has been
calculated using the equations (4.30) and (4.31) (see table 4.2)
Table 4.2 Calculated prestressing force in a number of points along the beam
x θ k Pm0(x) / Pm0(x=0)
0 0 0 1,00
1
/4 l 0,041 0,0269 0,99
1
/2 l 0,082 0,0538 0,98
3
/4 l 0,123 0,0806 0,97
l 0,164 0,1075 0,96
At an initial stress (after anchoring) p0 = 1395 N/mm2 at the position of the anchor, the
steel stress at midpan is 0,98 · 1395 = 1367 N/mm2.
So, the prestressing force at mid-span is:
Pm0 = 4 · 500 · 1367 = 2734 · 103 N = 2734 kN
This prestressing force meets the requirement found before, namely that:
2189 kN ≤ Pm0 ≤ 2963 kN.
Following, the stresses occurring in the various stages, will be checked.
April 2018 4-37

Chapter 4 - Bending
4.7.3 Situation during prestressing
The mean prestressing force is Pm0 = 2734 kN. During prestressing only the moment
caused by the selfweight ME,g = 474 kNm is activated. The compressive stresses at the
top and bottom fibre at midspan are checked:
Top:
Pm0 Pm0 ep M E,g
 ct    
Ac Wct Wct
2734 103 2734  441 103 474 106

 ct      0,38 N/mm 2  0 N/mm 2
328 103 92 106 92 106
Bottom:
Pm0 Pm0 ep M E,g

 cb    
Ac Wcb Wcb
2734 103 2734  441 103 474 106

 cb      18, 7 N/mm 2  21, 0 N/mm 2
328 103 70.3 106 70.3 106
So, in both cases the limit stresses are not exceeded.
4.7.4 Situation in the serviceability limit state (SLS)
First, the concrete stress at the outer tensile fibre at maximum load for t =  is calculated.
The most critical situation occurs when the stress losses due to shrinkage and creep have
developed. To simplify the calculation, it is assumed that the force Pm,∞ = 0,8 Pm0 =
2187 kN acts directly after prestressing. So, the cross-sectional forces due prestressing
(P) and selfweight (qg) must be calculated using the properties of the “pure concrete”
cross-section, whereas those for the additional static load (qp) and the variable load (Q)
should be calculated using the fictitious composite cross-section.
As a result, for the tensile stress at the bottom fibre of the cross-section it is found that:
Pm, Pm, ep M E,g,selfweight M E,g,static load  1M E,q

 cb     
Ac Wcb Wcb Wib
which results in:

2187 103 2187  441103 474 106 (144  0, 75  693) 106
 cb       4,9 N/mm 2  0 N/mm 2
328 10 3
70,3 10 6
70,3 10 6
76, 2 10 6
The stress is within the prescribed limit.
April 2018 4-38

Chapter 4 - Bending
Because of the small differences between the gross (pure) concrete cross-section and the
fictitious (composite) cross-section, in calculations often only the gross cross-section is
used.
Once the area within which the prestressing force has to be applied is known, the tendon
profile can be determined, see fig. 4.29. The limitations for the tendon layout usually
follow from the requirement that a certain tensile stress (here ct = 0 N/mm2) is not
exceeded.
Fig. 4.29 Area within which the tendon profile has to be to prevent concrete tensile
stresses to occur
At a section x where MEd,g(x) and MEd,freq(x) act, the following conditions (requirement:
no concrete tensile stresses) apply:
Pm0 Pm0 ep ( x) M E,g ( x)

Top fibre:    0
Ac Wct Wct
Pm, Pm, ep ( x) M E,freq ( x)

Bottom fibre:    0
Ac Wcb Wcb
This results in:
M E,g ( x) Wct
ep ( x)   (4.30)
Pm0 Ac
and
M E,freq ( x) Wcb
ep ( x)   (4.31)
Pm, Ac
As an example, with the aid of the equations (4.30) and (4.31) it is found, that at x = 1/4 l,
where ME,g = 3/4 · 474 = 356 kNm and ME,freq = 3/4 · 1138 = 854 kNm, the following
condition must be met:
April 2018 4-39

Chapter 4 - Bending
0, 25 m  ep ( 1 4 l )  0, 42 m
For x = 0 it is found that:
Wcb W
  ep (0)  ct
Ac Ac
For x = 0 this implies that the prestressing force should be applied between the most
upper and the most lower kern area points, see also eqs. (4.6) and (4.7). This refers, of
course, to the resultant of the prestressing tendons. One often denotes this as the
“fictitious” or “imaginary” tendon. In practice, the tendons are spread as much as
possible at the beam ends, in order to position their anchorages appropriately in the
structure.
Fig. 4.30 Prestressing operation for a large concrete beam (trough girder) in the
Hanzelijn (railway line) near Kampen (NL)
April 2018 4-40

Chapter 4 - Bending
4.8 Literature
4.1 Theile, V.: The influence of prestressing on the SLS in prestressed concrete
structures. Dissertation, TU Darmstadt, Germany, 1986 (in German)
4.2 Walter, R., Utescher, G., Schreck, D.: Preliminary assessment of prestress losses
due to deformation restraint. Deutscher Ausschuss für Stahlbeton, No. 282, 1977
(in German)
April 2018 4-41

Chapter 4 - Bending
April 2018 4-42

Chapter 5 – Statically indeterminate structures
5. Statically indeterminate structures

5.1 Determination of the prestressing load
When determining the prestressing load on a statically indeterminate structure, the
equivalent prestressing load method is the best method to use. The basics of the method
are illustrated in the figures 5.1 and 5.2. In fig. 5.1 the prestressing load is applied
centrically. The beam does not deform in flexure, it deforms in its axial direction only.
Deformation is not restrained in this direction. Therefore, no support reactions occur.
c
Fig. 5.1 Centrically prestressed beam
Subsequently, the tendon layout as shown in fig. 5.2 is considered. This layout shows
discrete changes of direction (kinks) at the supports, whereas in between the supports (in
the spans) the cables are straight.
Fig. 5.2 a. Prestressing tendon with discrete changes of direction

b. Forces exerted on the prestressing tendon
c. Forces exerted on the concrete (tendon forces and support reactions)
The prestressing tendon exerts forces on the beam only where its direction is changed
(kinked), or where it is anchored. This implies that, in the case considered, the
prestressing loads are directly transmitted to the supports. The support reactions
counterbalance the lateral prestressing forces. Therefore, the effect of prestressing in the
beam is the same as for the case of centrically prestressing shown in fig. 5.1: the concrete
is uniformly stressed in compression over the full height of its cross-section.
Now the beam as shown in fig. 5.3, having a curved tendon profile, is regarded. Note that
also in this case, there are kinks at the intermediate supports.
April 2018 5-1

If the layout of the tendon would have been like the one from fig. 5.2 (linear tendon parts
with kinks at the intermediate supports; dashed lines in fig. 5.3a), there would be no
bending moments in the beam as a result of prestressing.
However, the tendon profile is not linear and kinked, but curved and kinked. The loads
from the curved tendon profile parts in between the supports are uniformly distributed
equivalent prestressing loads qp:
8 Pm f
qp 
l2
The figures 5.3b and 5.3c show, qualitatively, the loads on the tendon and on the
concrete. The magnitude of the uniformly distributed prestressing loads on the beam are
shown in fig. 5.3d (points loads at anchors and kinks; uniformly distributed loads in the
spans).
Fig. 5.3 a. Beam with curved and kinked profile of the prestressing tendon
b. Forces exerted on the prestressing tendon
c. Forces exerted on the concrete
d. Practical representation of the equivalent prestressing load
In practice, it is not possible to choose the layout of the prestressing tendon such that
there are abrupt changes of direction (kinks), like the ones shown in fig. 5.2 and fig. 5.3.
For the curvature of the prestressing steel a minimum radius is prescribed, in order not to
damage the prestressing steel. This inevitable local anti-curvature of the prestressing steel
causes a downward load. This load (qp = Pm / R) acts only over a small length, but is
relatively high due to the high curvature (1 / R) of the tendon. In the design stage, the
April 2018 5-2

local downward load at the supports is often not regarded, but it cannot be ignored in an
accurate calculation.
The equivalent prestressing load, corresponding to the layout of the prestressing tendon
shown in fig. 5.4a, is represented in fig. 5.4b. Note that the actual layout is used; the
kinks at the intermediate supports are replaced by downward curved parabolas.
Fig. 5.4 Equivalent prestressing loads for upward and downward curvatures of the
tendon
The equivalent prestressing load method can also be applied in the case of non-prismatic
members. When the system line of the structure (the line through the centres of gravity of
the subsequent cross-sections) has a nonlinear layout, additional forces are generated,
resulting from the axial force in the cross-section. As a result of the angle between two
subsequent cross-sections (cross-sections  system line), the axial force has a component
perpendicular () to the system line, see fig. 5.5 (q R dφ in fig. 5.5a and 2 N sin(1/2φ) in
fig. 5.5b).
q
N
N

c.a. N
N
c.a.
d

N.d = q.R.d
q.R.d d 2.N.sin 12 
q= N
R
(a) (b)
Fig. 5.5 Load perpendicular to the system line of a non-prismatic structure, as a result
of a normal compressive force (c.a. = centroidal axis)
April 2018 5-3

At a kink in the system line, a force 2 N sin(1/2φ) (fig. 5.5b) occurs in the direction of the
bisectrice of the corner between the adjacent parts of the system line.
This force is 2 Pm sin(1/2α) in fig. 5.6a,b.
The vertical component of this force is 2 Pm sin(1/2α) cos(1/2α) = Pm sinα; the horizontal
force component is Pm (1 - cosα)
In the case of a curved system line, a load arises, which is inversely proportional to the
radius R of the system line (qp = Pm / R) (fig. 5.6c,d).
If such a structure is not calculated using a frame analysis program (which automatically
generates forces provided that the structural geometry is correctly schematised, see
section 5.2) such loads are dealt with in the same way as the equivalent prestressing load
following from the tendon curvature.
prestressing tendon
system line
system line prestressing tendon
Fig. 5.6 Superposition of the prestressing loads on a structure caused by a

discontinuous or a curved system line of the structure
As an example, the structure sketched in fig. 5.7a is now discussed. The following load
cases are distinguished:
1. Equivalent load qp as a result of the curvature of the prestressing tendon (fig. 5.7b).
April 2018 5-4

2. Two vertical forces F, one vertical force 2F and two horizontal forces Pm (1-cosα)
caused by the axial compressive force Pm as a result of a kink (angle α relative to the
horizontal axis) in the system line (fig. 5.7c).
3. Moments Pm ep as a result of the eccentric introduction of the prestressing load at the
ends of the structure (fig. 5.7d).
4. A load Pm divided in a horizontal (centric) and vertical force component (fig. 5.7e).
Up to now, the effect of tendon friction was not regarded. In a more accurate calculation
this should be regarded as well and the equivalent loads resulting from prestressing and
kinks and curvatures in the system line have to be adapted. In most cases, however, the
results are sufficiently accurate if the mean value of the prestressing force is used.
For an accurate determination of the layout of the tendons, the calculation of the friction
losses, the elongation of the tendon at prestressing and the equivalent load by
prestressing, it may be useful to express the tendon profile (layout) using a mathematical
function. To make the calculation of the tendon layout and its derivatives as simple as
possible, a parabolic cable layout is chosen, typically for beams and slabs.
R1
R1
prestressing tendon system line
Pm(1-cosα)
Pmsinβ
Pmcosβ Pmcosβ
Fig. 5.7 Influence of a kinked system line
April 2018 5-5

5.2 Input of equivalent prestressing loads in a frame-analysis

program
In order to calculate the distribution of moments and forces in a statically indeterminate
structure as a result of prestressing, a frame-analysis is often the most simple method. It
turns out, however, that in practice often errors are made with the input for such
programs. Therefore, a number of important aspects with regard to the input for frame-
analysis programs is discussed.
In section 5.1 it was shown in which way the prestressing load and the forces caused by a
curved or a kinked system line, can be applied as loads on a statically (in)determinate
structure. For the structures shown in figs. 5.6 and 5.7 it will be shown in this section
how those forces must be dealt with when a frame-analysis program is used.
The most essential difference between a frame-analysis program and a calculation by

hand, is that in the case of a frame-analysis program it is possible to implicitly take the
effects of shape and geometry of the structure into account. If this option is used, the
forces caused by a curved or kinked system line are automatically generated by the frame
analysis calculation. It would be wrong to put the load on the structure a second time.
The first example refers to the structure shown in fig. 5.6a. In this example the friction
losses are provisionally left unconsidered. In fig. 5.8a it is shown how the system line of
the structure can be schematised for a frame-analysis. Since account has been taken of the
kinked shape of the systemline, the prestressing load should only be applied at the ends of
the beam (at the anchors), see fig. 5.8c.
system line
stiff connection
Fig. 5.8 Example of a structure with a kinked system line
The second example is the structure shown in fig. 5.6c. Also in this case friction losses
are not considered. In fig. 5.9b it is shown how the system line of this structure can be
April 2018 5-6

schematised in a frame-analysis program. By subdividing the structure into a number of

straight bar-elements, the curvature of the system line can be described with sufficient
accuracy. Now the dead load and the traffic load are applied as a large number of small
loads on the structure, instead of a uniformly distributed load. However, this will hardly
have any influence on the results of the calculation. Also in this structure the prestressing
load should be introduced at the ends of the beam only (fig. 5.9c).
stiff connection
Fig. 5.9 Example of a structure having a curved system line
The third example concerns the structure shown in fig. 5.7. The friction losses will now
be considered. The schematisation of the system line is shown in fig. 5.10b, and is similar
to that of fig. 5.8b. The force Pm in the prestressing steel will be gradually reduced by the
effect of tendon friction. The development of the prestressing force Pm is represented in
fig. 5.10c. The load on the structure as a result of tendon friction and wedge set is
represented in fig. 5.10d. Figure 5.10e shows the prestressing loads at the beam ends. The
prestressing loads from the curved tendon profile are given in fig. 5.10f.
The influence of the friction (fig. 5.10d) is not only reflected by the distribution of the
equivalent prestressing load perpendicular to the structure, but also as a uniformly
distributed axial load along the structure. Also in this example it is not necessary to apply
the apparent force separately at the kinks in the system line.
April 2018 5-7

system line
distributed load in
longitudinal direction
from friction and wedge
set
Fig. 5.10 Example of a structure having a kinked system line and a curved tendon
profile; the influence of friction losses
5.3 Prestressed frames

If the beam in the frame shown in fig 5.11a is prestresssed, the columns will have the
same deformation at their top, see fig. 5.11b. Therefore, the prestressing load is not fully
introduced in the beam since part of it is carried by the columns. In this section it will be
investigated how high the losses of the prestressing force are in such a case. In fig. 5.11d
it is shown how a part of the beam and a column are separated for the calculation.
There are two compatibility conditions:
a. The displacement at the top of the column should be equal to half the shortening
of the beam.
b. The angular rotations of the column and the beam at their ends should be the same
(keeping the angle between the column axis and the beam axis 90º).
April 2018 5-8

Condition (a) can be formulated as:

2
( Pm  H )l13 Ml12 Pm e p l1 H 1 2 l2
   (5.1)
3EI1 2 EI1 2 EI1 EA2
Condition (b) results in:
Pm 3
Pm e p l1 l2
( Pm  H )l 2
Ml M 2 l2
1
 1
1
  R (5.2)
2 EI1 EI1 EI1 EI 2 24 EI 2
where I1 and I2 are the moments of inertia of the column and the beam, respectively, and
R is the radius of the curvature of the prestressing tendon.
l2
Furthermore R  2 (eq. (4.2)), where f is the drape of the tendon (fig. 5.11a).
8f
R
a l2 b
f
2
1 1
l1
c d
PFmp Pm
Fp
M M
ep ep
d2
H H
d1
Fig. 5.11 Prestressing of the beam of a frame
After combining the equations (5.1) and (5.2), a relation is obtained for the ratio between
H and Pm:
3  2 f  3ep   2
  
H 2 d2 2d 2
 (5.3)
Pm  3
   3 2
4 2
  
2 8 3 4
April 2018 5-9

where:
 = l2 / l1 (geometry of the frame)

 = d2 / d1 (ratio beam depth / column depth, fig. 5.11c)
 = d2 / l2 (slenderness of beam)
For  = 0,075, f / d2 = 0,7 and ep / d2 = 0,3 the diagram shown in fig. 5.12 is derived,
where the ratio H / Pm is presented as a function of d2 / d1 and l2 / l1.
Fig. 5.12 Ratio H / Pm for various values of d2 / d1 and l2 / l1
For d2 / d1 > 1,0, the prestressing losses in the beam due to the action of the columns is
less than 20%. However, the equivalent prestressing load, which is a result of the tendon
curvature, is not influenced. If in the case of very stiff columns only a small part of the
prestressing load would act as axial compression, whereas the equivalent prestressing
load by the tendon curvature would still be able to play a major role in carrying the loads
on the beam (fig. 5.13).
Fig. 5.13 Action of the prestressing as an equivalent load on the beam of a frame
having very stiff columns
Ignoring the effect of prestress-restraining members in a structure is, however, a regular

cause of damage. Also here the statement “Prestressing means deforming” holds
true.
April 2018 5-10

Chapter 6 – Time-dependent behaviour
6. Shrinkage, creep and relaxation

6.1 Introduction
When the durability of a structure has to be guaranteed for a long period, the long-term
behaviour of the building material concrete has to be considered as well.
“Creep” can be defined as the increase of the deformation with time under a sustained
constant load (fig. 6.1a).
“Shrinkage” is the shortening of the concrete occurring without the influence of any load
(fig. 6.1b), which is caused by the drying of the material (the chemical shrinkage
occurring during the hardening of the concrete is not considered).
“Relaxation” is closely related to creep. However, in case of relaxation, the deformation
of the material is kept constant and the initially present stresses decrease with time (fig.
6.1c).
c
 = constant
deformation
due to creep
a co
elasticdeformationby
elastic deformation by
applying
applying a load
load
t (time)
c
b
deformation due
to shrinkage
t (time)
c
decreaseofoftension
degrease compression
due to relaxation
c l = constant due to relaxation
t (time)
Fig. 6.1 Definitions of a) creep, b) shrinkage and c) relaxation
April 2018 6-1

The time-dependent behaviour cannot be disregarded in a structural design calculation, as

will be shown by a number of examples. Figure 6.2a shows a column that is loaded by an
axial force. The distribution of the load between the steel and the concrete can be
determined by means of the moduli of elasticity and the cross-sectional areas of these
materials. As a result of creep, the concrete will shorten. Then, the steel will follow and
higher compressive stresses in the steel result. So, as a matter of speaking, the concrete
backs out of the load, while the steel will be loaded more severely as time goes on.
s
t (time)
Fp
b
t (time)
instantaneous support
displacement (t = t 0)
c
t = t
d
counter-acting moment
Mp t = t0
Fig. 6.2 Some examples of the effects of the long-term behaviour of concrete
Figure 6.2b shows a beam which is prestressed without bonding. The prestressing
element can be considered as a spring.
* When shrinkage occurs, the beam will shorten.
* In the compressive zone extra shortening due to creep has to be added.
* By this shortening, the tendon will be released, by which the exerted force on the
concrete will decrease.
April 2018 6-2

Since the prestressing has to limit the magnitude of the tensile stresses, respectively the
width of the cracks in the serviceability limit state, or has to balance a fixed part of the
load, it is necessary to account for this loss in prestressing when designing the structure.
Figure 6.2c shows a beam on three supports. A deformation is forced upon the structure
by a sudden imposed displacement of the middle support. This will generate a bending
moment in the structure. However, in this case the time-dependent deformation has a
favourable effect: by relaxation of the concrete, this undesired moment will eventually be
reduced to a relatively small part of its initial value.
6.2 General concepts

6.2.1 Creep
Creep of concrete is caused by the deformation of the gel structure and the capillary
stress of the chemically non-bonded water.
Development and ultimate magnitude of creep depend on the following factors:
 The climate in which the structure is situated; especially the relative humidity and
temperature are important.
 The development of the degree of hydration1, as well as the degree of hydration under
loading. These depend on the age of the concrete under loading, the applied type of
cement and the curing conditions.
 The strength class of the concrete2.
 The dimensions of the cross-section.
 The duration of the loading.
Research has indicated that the creep consists of two components:
 A reversible part (also indicated as the visco-elastic part, the retarded elastic part, or
the reversible creep).
 A permanent part also indicated as the plastic part.
The development of these contributions as a function of time is quite different. The age of
the concrete hardly influences the reversible part εccr ( c = creep, r = reversible) and
already after a short period of time, its final level is reached. On the other hand, the
permanent contribution εccp ( c = creep, p = permanent) strongly depends on the age of
the concrete, and its ultimate value is reached just only after a long period. For the
simplification of the calculations, and because of the large scatter in the occurring creep,
a separate approach of both contributions is abandoned:
 c (t )   ce   ccr (t )   ccp (t )   ce   cc (t ) (6.1)
where:
1
The degree of hydration is the extent to which the cement is chemically bonded to water and hardened
cement paste is created.
2
The strength class of the concrete is determined to a large extent by the water-cement ratio (wcr) of the
concrete (amount of water per kg of cement).
April 2018 6-3

εc(t) is the total deformation of the concrete (elastic + creep);

εce is the elastic deformation of the concrete;
εcc(t) is the total creep deformation of the concrete (visco-elastic + plastic).
Figure 6.3 shows creep curves for two situations, each with a different stress
development as a function of time.
c c
t t
ce ce
elastic strain t t
ccr ccr
reversible strain t t
ccp ccp
plastic strain t t
c (t) c (t)
total strain t t
Fig. 6.3 Creep behaviour for two load functions
The magnitude of the creep deformation is directly proportional to the elastic

deformation. The relation between the creep deformation and the elastic deformation is
given by the creep function φ(t,t0).
It holds:
c
 cc (t )   (t , t0 ) (6.2)
Ec
where
Ec = 1,05 Ecm (EN 1992-1-1 cl. 3.1.4 (2); the Young's modulus at the origin of
the stress-strain relationship)
April 2018 6-4

According to Appendix B.1(1) of EN 1992-1-1, the creep function φ(t,t0) can be written
as:
 (t , t0 )  0  c (t , t0 ) (6.3)
where:
φ0 is the notional creep coefficient = RH  ( f cm )  (t0 )
βc(t,t0) is the factor that describes the development with time after loading
0,3
 t  t0 
=  
  H  t  t0 
16,8
β(fcm) is the factor for the effect of concrete strength = (6.4)
f cm
(fcm = fck + 8 N/mm2; EN 1992-1-1 table 3.1)
1
β(t0) is the factor for the concrete age at loading = (6.5)
0,1  t00,2
φRH is the factor for the effect of relative humidity

RH
1
 1 100 for f cm  35 N/mm 2
0,1 3 h0
 RH  (6.6)
 1  100 
 1  1   2 for f cm  35 N/mm 2
 0,1 3 h
0 
 
0,7 0,2
 35   35 
1    2   
 f cm   f cm 
2Ac
h0 is the notional size of the member in mm =
u
Ac is the cross-sectional area
u is the perimeter in contact with the atmosphere
t is the age of concrete in days
t0 is the age of concrete at loading in days
t-t0 is the non-adjusted duration of loading in days
βH is a coefficient depending on the relative humidity and the notional size
 1,5  1  (0, 012 RH)18  h0  250  1500 for f cm  35 N/mm 2
(6.7)
 1,5  1  (0, 012 RH)18  h0  250 3  1500 3 for f cm  35 N/mm 2
April 2018 6-5

0,5
 35 
3   
 f cm 
At a low relative humidity, the potential difference between the moisture content in the
structure and its environment becomes larger, which causes the structure to dry more
quickly. Now two opposite effects are of importance: a strong reduction of the moisture
content in the structure increases the creep, while a structure with a lower moisture
content creeps less. In practice, the influence of the difference in moisture content
between the structure and the environment dominates. This is demonstrated by
expressions (6.6) and (6.7). A low RH combined with a small notional size results in a
high φRH according to eq. (6.6). A high RH combined with a large notional size results in
a high βH according to eq. (6.7) and, as a result, a low βc(t,t0).
With regard to the concrete strength class two aspects play a role. First, concrete having a
high strength also has a high stiffness compared with lower strength concrete. The higher
stiffness results in a smaller creep (eq. (6.4)). Second, concrete having a higher strength is
less permeable and therefore exhibits a slower drying process. This is beneficial from the
point of view of reducing creep deformation (eqs. (6.6) and (6.7)).
The finer the cement, the faster the hydration process. Also at elevated temperatures, the
hydration process develops faster. Concrete having a high degree of hydration exhibits
less creep if loaded. The temperature effect influences the age of the concrete at loading
to be used in eq. (6.5). The actual age is corrected for the hardening temperature by using
the so-called adjusted age of concrete:

 9 
t0  t0,T   1  0,5
 2  t0,T 
1,2
where:
α = -1 for cement class S
= 0 for cement class N
= 1 for cement class R
 4000 
n   13,65 
 273T ( ti )
t0,T is the temperature adjusted concrete age   e 
ti
i 1
T(Δti) is the temperature (in ºC) during the time period Δti
Δt i is the number of days where a temperature T prevails
Instead of performing a calculation, the designer can choose to read the expected creep
coefficient from EN 1992-1-1 fig. 3.1, see fig. 6.4. The curves can be used for a
temperature range from -40ºC to +40ºC and a RH between 40% and 100%. Two curves
are available; one for inside conditions (RH = 50%) and one for outside conditions (RH =
80%).
April 2018 6-6

inside conditions – RH = 50%
outside conditions – RH = 80%
Fig. 6.4 Curves to determine the creep coefficient as a function of the concrete
strength class, the cement type, the notional size of the member and the age of
concrete at loading (EN 1992-1-1 fig. 3.1)
An example to illustrate the use of fig. 6.4:
outside conditions; h0 = 500 mm; t0 = 30 d; concrete C30/37.

Figure 6.5 shows how lines are constructed to determine the creep coefficient. A creep
coefficient of about 1,7 - 1,8 is found.
April 2018 6-7

outside conditions – RH = 80%
Fig. 6.5 Example to demonstrate the use of the curves (outside conditions; h0 =
500 mm; t0 = 30 d; concrete C30/37)
In case the concrete compressive stress at the time of loading t0 is more than 45% of the
cilinder compressive strength at that time (> 0,45 fck(t0)), non-linear creep has to be
accounted for. This might hold for precast prestressed concrete fully stressed at a
relatively low concrete strength. The creep coefficient then is (EN 1992-1-1 eq. (3.7)):
k (, t0 )   (, t0 ) e1,5 ( k σ  0,45)
where:
kσ is the stress-strength ratio σc / fcm(t0) where σc is the compressive stress and

fcm(t0) is the mean compressive strength at the time of loading t0
6.2.2 Shrinkage
Shrinkage depends on the following factors:

 The relative humidity.
 The strength class of the concrete.
 The dimensions of the cross-section.
 The age of the concrete.
According to EN 1992-1-1 cl. 3.1.4 (5), shrinkage shortening follows from:
 cs   cd   ca (6.8)
where:
εcs is total shrinkage
April 2018 6-8

εcd is drying shrinkage

εca is autogenous shrinkage
Drying shrinkage is the result of the drying of a member. Drying shrinkage can go on for
many years for structures having large dimensions.
According to EN 1992-1-1 eq. (3.9) drying shrinkage is:
 cd (t )   ds (t , t0 ) kh  cd,0 (6.9)
The final drying shrinkage strain at t → ∞:
 cd,  kh  cd,0
where
2Ac
kh is a coefficient that depends on the notional size h0 = of the cross-section,
u
see table 6.1 (EN 1992-1-1 table 3.3)
t  ts
βds(t,t0) =
t  ts  0, 04 3 h0
t0 is the age of concrete at the beginning of drying shrinkage (normally after
demoulding or at the end of curing) in days
εcd,0 is the nominal unrestrained drying shrinkage (EN 1992-1-1 table 3.2), see
table 6.2
Table 6.1 Values for kh (EN 1992-1-1 table 3.3)
h0 [mm] kh
100 1,0
200 0,85
300 0,75
> 500 0,70
April 2018 6-9

Table 6.2 Nominal unrestrained drying shrinkage εcd,0 [‰] for concrete with cement
CEM class N (EN 1992-1-1 table 3.2)
concrete relative humidity RH [%]

strength 20 40 60 80 90 100
class
C20/25 0,62 0,58 0,49 0,30 0,17 0,00
C40/50 0,48 0,46 0,38 0,24 0,13 0,00
C60/75 0,38 0,36 0,30 0,19 0,10 0,00
C80/95 0,30 0,28 0,24 0,15 0,08 0,00
C90/105 0,27 0,25 0,21 0,13 0,07 0,00
Autogenous shrinkage is caused by a lack of water during hydration, causing an

underpressure in the pore system. Since autogenous shrinkage is directly related to the
hydration process, it develops similar to the hydration process and, as a result, reaches its
final value after a relatively short period of time.
The autogenous shrinkage is taken as (EN 1992-1-1 eq. (3.11)):
 ca (t )   as (t )  ca ()
where
εca(∞) is the final autogenous shrinkage = 2,5   f ck  10  106 (EN 1992-1-1 eq.
(3.12))
βas(t) is the influence of the age of the concrete = 1  e 0,2 t
(EN 1992-1-1 eq.
(3.13))
At a low RH, the structure dries faster, which speeds up the development of shrinkage.
On the other hand, the permeability of concrete having a higher strength is lower than the
permeability of a lower strength concrete, which slows down the drying process. Both
aspects are present in table 6.2.
Just as in the case of creep, the geometry of the cross-section, in terms of the notional
size h0, plays an important role: the larger the dimensions of the structure, the longer
shrinkage will take to fully develop and the smaller the shrinkage the designer has to
account for. This is clearly shown in table 6.1.
As shown, apart from time, the dimensions influence the development of shrinkage in
time. The larger the dimensions of a structure, the smaller the theoretical final value of
shrinkage to account for and the slower the process, see eq. (6.9). This is strictly speaking
not correct since the final value of shrinkage is independent of the dimensions of the
structure. However, since shrinkage might take decades to fully develop in a large
structure, it is regarded as appropriate to use this proposed method in design.
April 2018 6-10

The final value of autogenous shrinkage strongly depends on the permeability of the
concrete and the amount of water available for the hydration process. Concrete having a
high strength has a low permeability on the one hand, but on the other hand it has a low
water-binder ratio and, as a result, exhibit a larger autogenous shrinkage. EN 1992-1-1
relates autogenous shrinkage primarily to the strength class. In practice, a low strength
concrete might also have a low permeability, for instance self-compacting concrete. The
autogenous shrinkage of such a type of concrete might be higher than the value from EN
1992-1-1. Long-term measurements (> 10 yrs) on the first large high strength concrete
bridge constructed in The Netherlands, the 2e Stichtse Brug, indicate that autogenous
shrinkage might reduce in time.
6.2.3 Calculation of the creep deformation with the principle of superposition

according to Boltzmann (1876)
For many creep calculations it is not possible to use a single load that is constant with
time. In many cases, the stress gradually increases with the progress in construction. It is
also possible that in spite of a constant dead weight, the stress even decreases due to
creep of the concrete. Because reinforcement does not creep, a part of the force in the
concrete is transferred to the reinforcement.
For an arbitrary stress development, the creep deformation is described by relation (6.2).
For a discrete stress increment Δσi applied at time ti it holds:
 i
 cc (t )   (t , ti )  (6.10a)
Ec
In case of a number of discrete stress increments:

n
   (t, t )
1
 cc (t )  i i (6.10b)
Ec i 0
Accordingly, the creep deformation Δεcc(t) at time t has to be determined separately for
each stress increment Δσi (see fig. 6.6).
April 2018 6-11

t [time]
φ(t,ti)
φ(t,t0)
φ(t,t1)
t0 t1
t [time]
Fig. 6.6 Calculation of the creep caused by a number of successive load increments
For a continuously varying stress σ(τ), equation (6.10b) can be rewritten as:
t

1 d ( )
 cc (t )   (t , ) d (6.11)
Ec d
0
The principle of superposition gives accurate results provided that the following four
conditions are met [6.4]:
 The stresses during the service life have to be smaller than about 40% of the
compressive strength.
 The total strain should not decrease due to a decrease of the load. Relaxation
however, i.e. the reduction of the stress under constant strain, is allowed.
 The variations in relative humidity should not be too large.
 No high stress increments should occur over the period after initial load application.
In this incremental calculation of the creep deformation, the full stress development as a
function of time should be accounted for. This usually results in very time-consuming
calculations, which can be performed only by a computer program. It must be kept in
mind that accurate calculations do not necessarily yield accurate results, because a
number of influencing factors (such as relative humidity and loading history) must be
estimated in advance. In order to avoid complicated calculations, approximation methods
have been developed.
April 2018 6-12

6.3 Simplified methods for the calculation of the shrinkage and creep
behaviour of structures
Several approximating methods are available for the calculation of shrinkage and creep, a
number of which are:
 The effective-modulus method (section 6.3.1).
 The method of Dischinger (section 6.3.2).
 The method of Trost (section 6.3.3).
6.3.1 The effective-modulus method
The total strain of the concrete consists of the elastic deformation εce, the creep
deformation εcc(t), and the shrinkage deformation εcs(t):
 c (t )   ce   cc (t )   cs (t )
n n
 i  i
 
i 0
Ec
 E
i 0 c
 (t , ti )   cs (t )
n
  i 
 c (t )    E
i 0 c
1   (t , ti )    cs (t )

(6.12)
For t   it holds:
n
  1   (, t )  
1
 c ( )  i i cs ( ) (6.13)
Ec i 0
n
Using φ∞ = φ(∞,t0) and      i transforms equation (6.13) into:
i 0
1
 c ( )    1      cs ()
Ec
or:

 c ( )    cs () (6.14)
Ec'
where:
Ec
Ec'  (6.15)
1  
Ec is also called the fictitious E-modulus of concrete.
April 2018 6-13

This approach makes the calculation of the deformations much easier: the fictitious E-
modulus Ec is determined using eq. (6.15) and is used in the strain calculation as if it
were the normal E-modulus Ec.
A remarkable property of equation (6.14) is that the creep deformation is fully recovered
during unloading. This implies that it is assumed that the concrete behaves as an ideal
visco-elastic material and that the creep deformation is completely reversible.
Moreover, the creep factor used corresponds to the age of the concrete at first loading.
This implies that the continuing ageing of the concrete is not taken into account (ongoing
hydration), although the creep coefficient is smaller for loads applied at a later point of
time. Therefore, the creep is overestimated when using this calculation method.
6.3.2 The method of Dischinger
Also the method developed by Dischinger uses a basic creep function, corresponding
with the time the first load is exerted on the structure, φ(∞,t0).
The influence of ageing of the concrete can nevertheless be accounted for. This is done
by adjusting the creep function for loads applied at a later point in time ti. The result is a
creep coefficient φ(ti,t0).
The creep function for a load applied at time ti is:
 (t , ti )   (t , t0 )   (ti , t0 ) (6.16)
Figure 6.7 shows the strain development as assumed by Dischinger. The basic idea
behind it is as follows: the curve that presents the strains after applying a stress increase
Δσi at time ti is parallel to the curve obtained if this stress increment Δσi were applied in
the beginning.
However, this standard creep function does not result in a reliable estimation of the
influence of concrete ageing. Contrary to the effective E-modulus method that
overestimates creep at increasing concrete age, Dischinger's method underestimates
creep.
Compared with the effective E-modulus method, there are also some differences in case
of unloading. In Dischinger's method creep deformation is irreversible, as can be derived
from fig. 6.8.
April 2018 6-14

actual behaviour
plastic from Δσi
elastic from Δσi
plastic from Δσ0
elastic from Δσ0
Fig. 6.7 Creep behaviour according to Dischinger in case of a stepwise increase of the
stress
actual behaviour
Fig. 6.8 Creep behaviour according to Dischinger in case of unloading
April 2018 6-15

6.3.3 The method of Trost
This method does not have the drawbacks of the two methods presented before.
Moreover, it results in a relatively simple calculation procedure. According to relation
(6.10b) it holds:
n
   (t, t )
1
 cc (t )  i i (6.17)
Ec i 0
For many structures, initially (t0) a relatively high stress σ0 is present, after which
relatively small stress increments Δσi(ti) follow (fig. 6.9).

i
0
t00 ti1
t
Fig. 6.9 Schematised representation of a frequently occurring loading history
Therefore, equation (6.17) can be rewritten as:

n
 
   (t, t ) 
1
 cc (t )    0  (t , t0 )  i i
Ec  i 1
Adding the elastic deformation and shrinkage results in the total strain:
n
 

1
 c (t )    0 1   (t , t0 )    i 1   (t , ti )     cs (t )

(6.18)
Ec  i 1 
According to fig. 6.9 it further holds:

n
 (t )   0   
i 1
i
Relation (6.18) can now be written as:
April 2018 6-16

n
 

1
 c (t )    0 1   (t , t0 )    (t )   0    i  (t , ti )    cs (t )

(6.19)
Ec  i 1 
In order to obtain a simple equation, which can be applied in practice, Trost proposes to
change equation (6.19) into:
1
 c (t )   0 1   (t , t0 )    (t )   0   1    (t , t0 )    cs (t ) 
Ec
1
 c (t )   0 1   (t , t0 )    (t )  1    (t , t0 )    cs (t ) (6.20)
Ec
where Δσ(t) = σ(t) - σ0 and ρ is the so-called ageing coefficient. With the aid of equations
(6.19) and (6.20) it can be derived:
n
   (t, t )
i 1
i i
 (6.21)
 (t )   0   (t , t0 )
This is a considerable improvement, since it turned out that the value of ρ is between
rather narrow boundaries. Generally, a value ρ = 0,8 appears to be a good approximation
for normally loaded concrete structures with a creep factor 1,5 < φ(∞,t0) < 4,0.
Again, relation (6.20) is considered and written more briefly as:

0  (t )
 c (t )  1     1       cs (t )
Ec Ec
or:
0  (t )   (t )
 c (t )     cs (t )  0    cs (t ) (6.22)
Ec 1    Ec 1     Ec Ec
This result shows that the total deformation due to elastic shortening and creep can be
calculated with:
 Ec  Ec 1    as fictitious modulus of elasticity for stresses that are present at t = t0.
 Ec  Ec 1     as fictitious modulus of elasticity for stresses that develop during
the creep process.
6.4 Relaxation of the concrete

Relaxation as defined in section 6.1 implies a reduction of stress at a constant
deformation. Figure 6.10 shows an example of such a type of loading. A wall cast in
between two stiff supports (e.g. rock) is prestressed. After prestressing, the gaps between
April 2018 6-17

the stiff supports and the wall are filled with a low shrinkage mortar. It will be judged
whether the prestressing force applied decreases due to relaxation.
Expressions derived in section 6.3 can be used to determine the time-dependent decrease
of the axial compressive force. When using expression (6.20) from Trost, a relaxation
function can be derived.
When using εc = σc / Ec and φ = φ(t,t0) equation (6.20) can be written as:

1
 c (t )   c0   0   (t )  1       cs (t ) (6.20a)
Ec 
wall jack
low shrinkage mortar
Fig. 6.10 a: Prestressing of a wall between two fixed points

b: Application of a low shrinkage mortar
Since εc(t) - εc0 = 0, the following expression for the change of stress Δσ(t) is derived:
 0   cs (t ) Ec
 (t ) 
1  
from which it is derived that:
    cs (t ) Ec /  0 
 (t )   0   (t )   0 1   (6.23)
 1   
The results obtained with this expression will be clarified by an example.

Input data:
σ0 = N / Ac = -10 N/mm2 φ∞ = 2,5 ρ = 0,8

Ec = 30 · 103 N/mm2 εcs(∞) = -0,25 · 10-3
April 2018 6-18

The development of shrinkage and creep over time is described by the function given in
figure 6.11.
 (t )  cs (t )
  cs,
5y
Fig. 6.11 The assumed development of shrinkage and creep over time
After 90 d (about 50% of the end values of shrinkage and creep are reached):
φ(90 d) = 0,5 · 2,5 = 1,25
εcs(90 d) = 0,5·(-0,25 · 10-3) = -0,125 · 10-3
After substituting these values in eq. (6.23), the concrete stress is found to be:
 1, 25   0,125 103   30 103 /(10) 
 c (90 d )  10, 0  1    1,9 N/mm 2
 1  0,8 1, 25 
After 360 d (about 80% of final creep and shrinkage is reached):

φ(360 d) = 0,8 · 2,5 = 2,0
εcs(360 d) = 0,8·(-0,25 · 10-3) = -0,20 · 10-3
Result from eq. (6.23) the concrete stress is found to be:

 2, 0   0, 20 103   30 103 /(10) 
 c (360 d )  10, 0  1  0
 1  0,8  2, 0 
At t = , σc (360 d) = +0,7 N/mm2
The development of the concrete stress as a function of time is presented in fig. 6.12.
April 2018 6-19

-1,9 N/mm2
360
time [d]
Fig. 6.12 The development of concrete stress as a result of relaxation
The initial compressive stress of -10 N/mm2 is vanished after 360 days. It will turn into a
tensile stress after 360 d. This example demonstrates that prestressing a structure fixed at
two points is a very inefficient way of prestressing.
6.5 Relaxation of the prestressing steel
Time-dependent deformations also occur in steel subjected to the high stress levels that
are normally present in prestressing steel. Because the time-dependent deformations of
the concrete are small compared with the strains of the prestressing steel, a constant time-
independent deformation in the steel can be assumed. Relaxation will cause the stress in
the prestressing steel to decrease under this imposed deformation, which is almost kept
constant in time.
The relaxation behaviour of the prestressing steel is determined with tests, which last
1000 to 3000 hours (fig. 6.13).
log t [h]
Fig. 6.13 Relaxation of the prestressing steel as a function of the loading time
April 2018 6-20

The relaxation of the prestressing steel appears to depend on a number of factors:

 The initial stress: the relaxation strongly increases at a higher initial stress. This is
 pi
described in EN 1992-1-1 cl. 3.3.2 (7) by the factor   where  pi   pm0 .
f pk
 The temperature: relaxation proceeds faster at higher temperatures (up to about
100 ºC). In EN 1992-1-1 cl. 10.3.2.2 (2) this is accounted for by correcting for
temperature the interval over which relaxation occurs:
1,14Tmax  20 n
teq   
  T( ti )  20  ti
Tmax  20 i 1
 The relaxation level of the steel, that depends on the manufacturing method and
possible post-treatment of the steel. EN 1992-1-1 distinguishes three classes:
 Class 1 having a relatively high stress loss, referring to cold drawn wires and
strands. Stress loss after 1000 h in a standard relaxation test is about 8%. In the
calculation according to EN 1992-1-1 cl. 3.3.2 (7) it can be assumed that ρ1000 =
8%, or the result presented on the prestressing steel certificate can be used. For
this class the following expression applies (EN 1992-1-1 eq. (3.28)):
0,75 (1  )
 pr 6,7   t 
 5,39 1000 e   105
 pi  1000 
 Class 2 having a somewhat lower stress loss and referring to heat-treated wires
and strands. Stress loss after 1000 h in a standard relaxation test is about 2,5% →
ρ1000 = 2,5%. The following expression applies (EN 1992-1-1 eq. (3.29)):
0,75 (1  )
 pr 9,1   t 
 0, 66 1000 e   105
 pi  1000 
 Class 3 which refers to bars. Stress loss after 1000 h in a standard relaxation test
is about 4,0% → ρ1000 = 4,0%. The following expression applies (EN 1992-1-1
eq. (3.30)):
0,75 (1  )
 pr 8,0   t 
 1,98 1000 e   105
 pi  1000 
The final stress loss caused by relaxation can be calculated using t = 500000 h.
Example
Strands having a low stress relaxation loss; initial stress:

 1341
 pm0   pi  1341 N/mm 2    pi   0, 72
f pk 1860
April 2018 6-21

Stress relaxation loss:

0,75 (1  )
9,1   t 
 pr  0, 66 1000 e   105  pi
 1000 
0,75 (1 0,72)
 500000 
 pr  0, 66  2,5  e 9,10,72
  105 1341  57, 2 N/mm 2
 1000 
When comparing the prestressing steel relaxation expressions from EN 1992-1-1 with the
ones used in the Dutch code NEN 6720 (VBC 1995), it might be concluded that in EN
1992-1-1, time-dependent stress losses caused by shrinkage and creep do not influence
the relaxation loss. In NEN 6720 relaxation loss is directly reduced since concrete
shrinkage and creep reduce the steel stress level, which in turn reduces the relaxation loss
since the latter depends on the stress level. However, EN 1992-1-1 accounts for this too,
but afterwards, when all time-dependent steel stress losses are added. Relaxation loss is
then reduced by 20% to 80% of the calculated value (EN 1992-1-1 eq. (5.46)).
6.6 Redistribution of stresses in structural parts loaded in

compression
6.6.1 Axially loaded, symmetrically reinforced cross-section
A column is loaded by an axial force as shown in fig. 6.14. Immediately after loading, the
force distribution between steel and concrete can be determined using equation (3.3).
It then holds:
N N  e s
N c0  ; N s0 
1   e s 1   e s
N
 c0   s0 
Ec Ac 1   e s 
N
Fig. 6.14 Symmetrically reinforced column, subjected to a long-term load
April 2018 6-22

The concrete exhibits creep and shrinkage deformation. According to Trost it holds:
 c  t    c0 1    t , t0     c  t  1     t , t0     cs
 s  t    s0   s  t 
The concrete will reduce its initial load by shrinkage and creep and transfer it to the steel.
The changes in the forces over the course of time are indicated by ΔNc(t) and ΔNs(t).
From force equilibrium it follows:
N c  t   N s  t   0
Because of the required compatibility of deformations it should hold:
 c t    s t 
Since co = so, the following should also hold for the time-dependent components of the
total concrete and steel strains:
 c  t    c0   s  t 
It can now be derived, using the changes in forces:
N c  t 
 c  t    c0   c0   t , t0  
Ec Ac
1     t , t0     cs (6.24)
N s  t 
 s  t   (6.25)
Es As
From these equations it follows:

 N c  t  
N s  t    s  t  Es As    c0  t , t0  
Ec Ac
1     t , t0     cs  Es As 
 
 1     t , t0  Ec Ac cs 
  Ec Ac c0  N c  t        t , t0  
   t , t0    t , t0   e s
 1     t , t0  Ec Ac cs 
 N c  t   N s  t        t , t0 
   t , t0    t , t0   e s
 1     t , t0     cs 
1   e s  t , t0   Ns  t    N c  t   Ec Ac   e s  t , t0 
   t , t0      t , t0  
  cs   e s  t , t0 
N c  t   N s  t     N c  t   Ec Ac  (6.26)
   t , t0   1   e s 1     t , t0  
April 2018 6-23

The effects of shrinkage and creep on the development of the stresses are demonstrated
by an example. For the column as shown in fig. 6.13, the following data are assumed:
Ac  2502  62500 mm 2   2,5

As  8 Ø   mm 2  cs ()  0, 25 103
Ec  30 103 N/mm 2  cs (t ),  (t ) according to fig. 6.11
Es  205 103 N/mm 2   0,8
e  Es Ec  6,83 s  As Ac  0, 04
The initial load at the cross-section is N0 = -640 kN. Directly after applying this load it
holds (also see eq. (3.3)):
1 1
N c0  N0    640   503 kN
1   e s 1  6,83  0, 04
 e s
N s0  N 0  137 kN
1   e s
 c0  N c Ac  8, 0 N/mm 2
 s0   e  c0  54, 6 N/mm 2
90 days after loading:
 (90 days)  0,5  1, 25

 cs (90 days)  0,5  cs ()  0,125 103
From equation (6.26) it can now be deduced:
 0,125 103  6,83  0, 04 1, 25

N c  t     503 10 3
 30 103  2502    152 103 N
 1, 25  1  6,83  0, 04  1  0,8 1, 25 
 152 kN
N s  t   N c  t   152 kN
 c (90 days)   503  152  103 2502  5, 6 N/mm 2

 s (90 days)   137  152  103 2512  115 N/mm 2
In a similar way it can be found:

360 days after loading :  c (360 days)  4,5 N/mm2 ;  s (360 days) = 142 N/mm2
5 years after loading :  c (5 years)  3,9 N/mm2 ;  s (5 years) = 158 N/mm2
After 5 years, the concrete stress is reduced to 49% of its initial value. In fig. 6.15, the
stress reduction as a function of time is presented for several reinforcement ratios. In all
cases, an initial concrete compressive stress of -8,0 N/mm2 is assumed.
April 2018 6-24

ρs = 0%
ρs = 2%
ρs = 4%
ρs = 6%
ρs = 8%
t [years]
Fig. 6.15 Stress reduction in the concrete due to shrinkage and creep of a symmetrically
reinforced cross-section loaded in compression
When detailing structures, creep should be considered as well. Figure 6.16 shows an
example of a column clad with natural stone tiles. When creep is not taken into account
when cladding the column (for example by applying wide joints), the tiles may spall off
after some time.
Fig. 6.16 Spalling of a natural stone cladding due to creep of the column
6.6.2 Eccentrically loaded, asymmetrical cross-section
In section 6.6.1, the specific case of an axially loaded symmetrical cross-section is

discussed. In many cases however, the cross-section is asymmetrical and subjected to an
eccentrically applied load. This more general case will be discussed in this section. For
that purpose, the cross-section shown in fig. 6.17 is considered.
By prestressing and partially loading of the structure at time t = t0, the stress distribution
as shown in fig. 6.17b occurs. The cross-section contains prestressing steel Ap and
April 2018 6-25

reinforcing steel As. The centre of gravity of the joined steel cross-section is at a distance
eps from the centroidal axis of the total cross-section. The change in steel stress due to
shrinkage and creep of the concrete is calculated.
concrete stress
at t = 0
Fig. 6.17 Redistribution of stresses in an eccentrically loaded, asymmetrical cross-

section
The initial strain of the concrete (c) at the centre of gravity of the steel (ps) is:
 cps (0)
 cps (0)  (6.27)
Ec
where Ec is the modulus of elasticity of the concrete at the time of loading.
If the concrete can deform without any restraint from the steel, the following additional
deformation from creep (c) and shrinkage (s) (= c + s) is expected in the concrete at the
level of the prestressing and reinforcing steel:
 cps (0)
 cps,c+s  t    (t , 0)   cs (t ) (6.28)
Ec
where εcs(t) is the absolute value of the shrinkage developed at time t.
In reality, the concrete and the steel are joined together, and the steel will react with a
force ΔF(t) to restore compatibility (fig. 6.17c). This force results in a strain change
Δεps(t) in the steel (Ap + As):
F  t 
 ps  t   (6.29)
( As  Ap ) Eps
where Eps is the modulus of elasticity of both prestressing and reinforcing steel (for the
sake of convenience, no distinction between the two is made).
April 2018 6-26

From equilibrium it follows that the same force is exerted to the concrete in opposite
direction. The force develops gradually since creep and shrinkage develop in time. This
implies that the deformation of the concrete caused by the force ΔF(t) is also influenced
by creep. Since the force is not constant in time, but gradually increases, the coefficient ρ
(from Trost) is used.
The result is the following change in concrete strain at the centre of gravity of the steel:
 cps  t     cps from F (t )  (1   (t , 0))
where the relation between ΔF(t) and ∆cps is:
 F  t  F  t  eps2  1
 cps    
 A Ic
 c  Ec
 F  t  F  t  eps2  1
The result is:  cps  t       (6.30)
 A
 c Ic  Ec /(1   (t , 0))
The compatibility requirement is:
 cps (0)
 cps,c+s  t    (t , 0)   cs (t )   cps  t    ps  t 
Ec
 cps (0)  F  t  F  t  eps2  1 F  t 
 (t , 0)   cs (t )       (6.31)
Ec  Ac Ic  Ec /(1   (t , 0)) ( As  Ap ) Eps
 cps (0)  (t , 0)   cs (t )
F  t  
 1 eps2   1   (t , 0)  1
    
 Ac I c   Ec  ( As  Ap ) Eps
April 2018 6-27

The stress change in the steel caused by creep and shrinkage of the concrete is:
 
 
F  t  1   cps (0)  (t , 0)   cs (t ) 
 ps,c+s  t     
( As  Ap ) ( As  Ap )  1 eps   1   (t , 0) 
2
1
    

  Ac I c    
Ec  ( As  Ap ) Eps (6.32)
Eps   cps (0)  (t , 0)   cs (t ) 
 ps,c+s  t  
 Ac eps2 
( s   p ) 1 
  1   (t , 0)   e  1
 I c 
Equation (6.32) can be extended by adding the influence of the relaxation of the
prestressing steel. Relaxation of steel not only implies a reduction of the steel stress but
also of the prestressing force exerted on the concrete. As a result, the time-dependent
response of the concrete is also influenced by the relaxation of steel. The interaction
between creep and shrinkage of concrete and relaxation of steel is taken into account by
reducing relaxation with a factor 0,8 (EN 1992-1-1 cl. 5.10.6 (1)).
Equation (6.32) then reads:
Eps   cps (0)  (t , 0)   cs (t )   0,8 pr  t 

 ps,c+s+r  t   (6.33)
 A e2 
( s   p ) 1  c ps  1   (t , 0)   e  1
 Ic 
which results in the following expression:
 cps (0)  e  (t , 0)  0,8 pr  t    cs  t  Eps

 ps,c+s+r  t   (6.34)
1   e   p  s  f 1    (t , 0) 
In this expression the cross-section factor f is used (see also section 4.3):
Ac eps2
f  1
Ic
When ρp = 0 and f = 1 (only reinforcing steel and a symmetrical cross-section) are
substituted in this general expression, the earlier derived relation (6.26) remains.
Equation (6.34) is presented in EN 1992-1-1 as eq. (5.46). EN 1992-1-1, however, is

based on calculating the absolute value of the stress loss. Therefore, in the nominator of
eq. (5.46) it is assumed that:
 shortening from concrete shrinkage has a positive value;
April 2018 6-28

 stress loss from prestressing steel relaxation has a positive value (see EN 1992-1-1
eqs. (3.28)-(3.30));
 the concrete stress at the level of the prestressing steel caused by quasi-permanent
loads has a positive value in case of compression.
It is noted that the derivation is primarily based on the behaviour of a statically

determinate structure.
Example
A prestressed hollow-core slab is simply supported at the two ends. It has a span l =
10 m. The selfweight is 4 kN/m and the static load is 1 kN/m. Further, the following data
hold:
Ac  165,8 103 mm 2   2,5   0,8

I c  1427 106 mm 4  cs ()  0, 25 103 e p  85 mm
Ap  7  93  651 mm 2
Y1860S7 f pk  1860 N/mm 2
Ec  36 103 N/mm 2 (C45 / 55)  po  1395 N/mm 2 (table 2.4)  p,max  1488 N/mm 2
Fig. 6.18 Calculation of the stress redistribution due to shrinkage and creep in a
prestressed hollow-core slab (dimensions mm (top) and m (bottom))
April 2018 6-29

For this cross-section it can be derived:
Ac ep2 165,8 103  852

f  1  1  1,84
Ic 1427 106
 e  Ep Ec  195 103 36 103  5, 42
 p  Ap Ac  0, 0039
The following part of the prestressing force Pmax is transferred to the concrete (see
equation (4.11)):
1 1
Nc  Pmax   6511488  932 103 N
1   e p f 1  5, 42  0, 0039 1,84
from which for σp0 it follows:
932 103
 p0   1432 N/mm 2  1395 N/mm 2
651
Conclusion: a too high stress after anchoring.
The prestressing force Pmax has to be reduced to:
Pm0  6511395  908 103 N
The maximum force to be applied during stressing:
1
Pmax  908 103 N  Pmax  943 103 N
1  5, 42  0, 0039 1,84
As a result of the permanent loads, the midspan bending moment is:
M G  18  (4  1) 10, 02  62,5 kNm
At the beginning of the shrinkage and creep process, the concrete stress at the level of the
prestressing steel is:
2
N N e M e
 cp (0)   c  c ps  G ps
Ac Ic Ic
908 103 908 103  852 62,5 106  85
    6, 4 N/mm 2
165,8 10 3
1427 10 6
1427 10 6
The loss caused by relaxation is calculated using:
April 2018 6-30

 pi 1395
 pi   p0  1395 N/mm 2      0, 75
f pk 1860
For low relaxation strands (class 2):

0,75 (1  )
 9,1   t  
 pr   0, 66 1000 e   105   pi 
  1000  
 
 9,10,75  500000 
0,75 (1 0,75)
5

  0, 66  2,5  e    10  1395  67,9 N/mm
2
  1000 
 
Note that eq. (6.34) uses shrinkage in absolute value to comply with EN 1992-1-1 eq.
(5.46). Initial compression in concrete at the level of the steel is negative in expression
(6.34). When following expression (6.34) the result is:
6, 4  5, 42  2,5  0,8  67,9  0, 25 103 195 103

 p,c+s+r   170 N/mm 2
1  5, 42  0, 0039 1,84  1  0,8  2,5 
which is 12,1% of the initial prestressing stress σp0.

The contribution of the relaxation of the prestressing steel is about 32% of the total loss.
Remark
The previous calculations may create the impression that time-dependent deformations
can be calculated very accurately. This is not the case since many factors can only be
estimated roughly: the relative humidity and temperature for example are not exactly
known, while in practice also the magnitude of the permanent load and the quasi-
permanent part of the variable load are only known by approximation.
When the influence of the steel on the shrinkage and creep process is neglected, a simpler
expression results, which provides a more conservative (i.e. too large) estimation of the
prestress losses.
Using ρp = 0, equation (6.34) results in:
 p,c+s+r   cp (0)  e  0,8 pr   cs Ep
or:
 p,c+s+r    cp (0)    cs  Ep  0,8  pr (6.35)
In the example presented, the prestress loss now is -190 N/mm2 instead of -170 N/mm2.
The difference in this case is 10%. The accuracy depends on the shape of the cross-
section, the modulus of elasticity of the concrete and the amount of steel.
April 2018 6-31

6.7 Redistribution of forces in statically indeterminate structures

The redistribution of forces in statically determinate structures can be determined with
the methods discussed in section 6.6. However, also for statically indeterminate
structures, the method of Trost can be used to analyse the behaviour of the structure over
the course of time.
In section 6.3.3 it was already pointed out that a simple method is obtained if the
deformations caused by the elastic shortening of the concrete are calculated using:
Ec
Ec'  (6.36)
1 
for the stresses acting from t = 0, and with
Ec
Ec''  (6.37)
1  
with  = 0,8, for stresses which develop over the course of time.
How the redistribution of forces in statically indeterminate structures can be determined

will be shown for a number of representative examples.
6.7.1 Redistribution of forces and moments in statically indeterminate structures
As an example a beam on three supports is considered, see fig. 6.19. The beam is
subjected to an equivalent load caused by prestressing and by a constant uniformly
distributed load qconst. Only creep is considered. It is assumed that the structure is
uncracked in the SLS.
Fig. 6.19 Statically indeterminate beam
For the calculation it is assumed that the mid support is removed. A uniformly distributed
unity load on the structure, now only supported at A and C, results in a displacement 10
(fig. 6.19). The value k10 represents the spring flexibility of the beam AC:
April 2018 6-32

5  (2l ) 4
k10 
384 EI c
The deflection caused by the permanent load qg and the prestressing load qp (upward
prestressing load Pm / R) then is (qg - qp)·k10. Furthermore, it is assumed that a deflection
11 occurs as a result of an upward force RB (fig. 6.19c).
The spring flexibility for the load RB:
(2l )3
k11 
48 EI c
The deflection of the beam at point B should be 0, so:
 B (t )  (qg  qp ) k10  RB k11  0 (6.38)
Over the course of time the equivalent prestressing load is reduced due to time-dependent
prestress losses. This then also holds for the reaction force RB:
 B (t )  (qg (t )  qp (t )) k10  RB (t ) k11  0
where:
qp (t )  qp0  qp
RB (t )  RB0  RB
Since k10 and k11 both depend on the time-dependent value of Ec, in analogy with eq.
(6.18) the following relation must hold:
 B (t )  (qg  qp ) k10 (1   (t , 0))  RB0 k11 (1   (t , 0))  qp k10 (1   (t , 0)) 

(6.39)
 RB k11 (1   (t , 0))  0
in which the deformations caused by the loads present already from t = 0 are multiplied
with a factor (1+(t,0)), and the deformations caused by loads which develop over time
are multiplied with a factor (1+(t,0)).
It holds at t = 0: (t,0) = (0,0) = 0, qp = 0 and RB = 0
As a result:  B (t  0)  (qg  qp ) k10  RB0 k11  0
When combining this equation and eq. (6.39), it is found that:
April 2018 6-33

qp k10 (1   (t , 0))  RB k11 (1   (t , 0))  0

k10 k 8 f Pm f
RB  (qp )  10  2
 10 Pm
k11 k11 l l
This shows that, in this case, the change in the reaction force RB is only caused by the
loss of prestress, and not to redistributions caused by creep of the concrete. So, as a
general statement:
In statically indeterminate structures no redistribution of forces and moments occurs due
to creep, apart from the effect of prestress losses.
6.7.2 Redistribution of moments caused by creep in the case of a change in the

structural system
A change in the structural system occurs when individual structural concrete members,
which are loaded already, are coupled. As a first example the coupling of two precast
beams to one continuous, statically indeterminate, beam is regarded, see fig. 6.20. Before
coupling, the individual beams were already loaded by their permanent load qg and by the
equivalent prestressing load qp. The angular rotation of the beams at the intermediate
support B, before coupling, at time t = t0, is:
(qg  qp ) l 3
 B (t0 )  (6.40)
24 Ec I c
qG + q p
Fig. 6.20 Coupling of two single statically determinate precast beams to one continuous
(statically indeterminate) beam
Directly after coupling at time t0, the moment at the intermediate support MB(0) = 0.
However, as time elapses, a bending moment MB(t) develops because the following
angular rotation due to creep at the beam ends at B is restrained.
If the loads (qg - qp) and the moment MB(t) could develop independent of each other, the
angular rotation B would become, after Trost:
(qg  qp ) l 3
 B (t ) 
24 Ec' I c
or, with Ec’ = Ec / (1 + ):
April 2018 6-34

(qg  qp ) l 3
 B1 (t )  (1   ) (6.41)
24 Ec I c
However, there would also be an angular rotation caused by the moment MB(t) that
develops over time:
M B (t ) l
 B2 (t )  (1   ) (6.42)
3 Ec I c
But, since the beams are connected from t = t0, from then on no angular rotation can
occur anymore. This means that:
 B1 (t )   B2 (t )   B (t0 )  0
Substitution of eq. (6.40)-(6.42) into this equation results in:
(qg  qp ) l 2 
M B (t )  
8 1  
This equation leads to a remarkable observation. The part -(qg - qp)·l2/8 is exactly equal
to the moment introduced by (qg - qp) if the structural system is continuous from the early
beginning, so if the system were built as a “monolithic” system immediately. Therefore:

M B (t )  M B,mono 
1  
The development of the support moment MB over time is shown in table 6.3. The
calculation is based on the creep function shown in fig. 6.21 (the influence of prestress
losses, as discussed in section 6.7.1, has not been considered).
Table 6.3 Development of the support bending moment
56 days 180 days 1 year 5 years
 (t) 1,00 1,75 2,00 2,50
MB(t)/MB,mono 0,56 0,73 0,77 0,83
April 2018 6-35

 (t )

Fig. 6.21 Development of support moment as a function of time
For loads applied after the coupling of the beams, the monolithic structural system
applies. According to section 6.7.1, no redistribution occurs because of this load.
For the general case of redistribution due to a change in the structural system, Trost
derived the following equation:
  
M (t )  M 0  ( M mono  M 0 )  
 1   
or:
   
M (t )  M 0 1    M mono (6.43)
 1    1  
where:
M(t) is the moment at time t caused by a load that was applied before the coupling;
M0 is the moment at time t = t0, before coupling;
Mmono is the moment that would have occurred if the system were monolithic from
the early beginning.
Figure 6.22 shows the case of a bridge, built according to the cantilevering erection
method. In this case the bridge is built from two sides. As soon as the cantilevering parts
have reached the mid of the span, they are coupled, which leads to a change of the
structural system.
April 2018 6-36

bending moment line

before coupling
Fig. 6.22 Redistribution of moments caused by creep of the concrete in a bridge built
by the cantilevering erection method
In the middle of the span Mspan(0) = 0 (just before coupling at time 0), so it follows:

M span (t )  M span,mono
1  
At the support there is already a moment Msupport,0 before the coupling is realized, so here
the result is:
     
M support (t )  M support,0 1    M support,mono  
 1     1   
There are also construction methods where the structural system changes a number of
times. An example of this is a segmental bridge, which is built in one direction, see fig.
6.23.
April 2018 6-37

3
2
construction part 1
M1
M2
M3
Mmono
Mmono
M
M
Fig. 6.23 Redistribution of moments caused by creep in a structure built in phases
In any cross-section of the structure the following equation holds:
n 
  (t , ti )    (t , ti )  
M (t )    M i 1    M mono  
i 0   1   (t , ti )   1   (t , ti )  
where:
Mi is the moment in the cross-section considered due to a load applied in phase i;

Mmono is the moment in the cross-section considered that would have occurred due to
the total load in the construction phase, if the structure would have been
monolithic from the early beginning.
6.7.3 Redistribution of moments due to creep in the case of a sudden support

settlement
Due to a sudden support settlement B (fig. 6.24) a redistribution of forces and moments
occurs. The supports A and C are more heavily loaded, whereas the support reaction at B
is reduced. Due to the settlement at B, moments occur which have to be superimposed on
the moments that already exist. The development in time of those moments caused by a
sudden support settlement, will be analysed.
April 2018 6-38

Fig. 6.24 Imposed deformations due to a sudden support settlement
As a result of a unit force RB a displacement 11 occurs. In that case:
RB011   B (6.44)
In the course of time no further settlement occurs, in spite of the change of the reaction
force RB in time:
RB (t )  RB0  RB
According to Trost the following relation holds:
RB0 11   RB 11 (1    )  0 (6.45)
This leads to:

RB   RB0
1  
For RB(t) this means:
  
RB (t )  RB0  RB  RB0 1  
 1   
The effect of a sudden support settlement is significantly reduced in time: with  = 2,5
and  = 0,8 one finds:
RB ()  0,17 RB0
It was assumed that the structure remains uncracked. If the structure would crack, its
bending stiffness would be reduced. As a result, RB() would be reduced even more.
April 2018 6-39

6.7.4 Effect of a slow support settlement
Now the structure shown in fig. 6.25 is regarded, which undergoes a slowly increasing
support settlement.
Fig. 6.25 Imposed deformations caused by a slowly increasing support settlement
It is assumed that the development of the support settlement in time is similar to that of
the creep function:
 (t )
 B (t )   ( )
 B
Since at t = t0 there is no deformation yet, the reactions caused by the imposed

deformation are still 0, so:
RB(t) = RB0 +RB = RB
For reasons of compatibility the following relation holds:

RB 11 (1   )   ( )
 B
or:
 B ( )  
RB  .  RB,fic 
11  (1   )  (1   )
where RB,fic is the fictitious support reaction in B if B, would have occurred
immediately at t = t0. Because of the condition that RB(t) = RB the following relation is
obtained:
RB (t ) 

RB,fic  (1   )
The results of this function, with the time-dependent development of the creep according
to fig. 6.21, is given in table 6.4.
April 2018 6-40

Table 6.4 Development of support reaction RB(t)
t (t) RB(t) / RB,fic
56 days 1,00 0,22

180 days 1,75 0,29
1 year 2,00 0,31
5 years 2,50 0,33
6.7.5 Summary of the calculation of the redistribution of moments and forces

according to Trost
The analysis of the redistribution of forces due to creep leads to a number of important
conclusions, which are summarised here:
1) The forces and moments which arise in a structure due to a sudden deformation
(support settlement) can be calculated by multiplying the corresponding elastic
distribution of forces and moments with a factor:

kφ1  1  with  = 0,8 (see section 6.7.3)
1  
2) The forces and moments that occur in a structure due to a gradually developing
deformation (slow support settlement, slow temperature change, shrinkage) can be
determined by multiplying the corresponding distribution of forces and moments by a
factor:
1
kφ2  (see section 6.7.4)
1  
3) In the case of a changing structural system, the finally occurring forces and moments
can be calculated as the sum of the following contributions:
 The linear elastic distribution of forces in any phase, immediately after
application of the load (selfweight, prestressing), multiplied with:

kφ1  1 
1  
 The linear elastic distribution of forces and moments, which would occur in the
structure if the monolithic structural system would have been present from the
early beginning, multiplied with:

kφ3  (see section 6.7.2)
1  
April 2018 6-41

 Here only the loads should be considered that are already present before the final
structural system is activated. The loads applied afterwards act on the monolithic
system and do not result in redistribution of forces.
In a statically indeterminate structure which is monolithic from the beginning, the forces
and moments only change due to prestress losses (section 6.7.1.)
6.8 Literature
6.1 Trost, H., Mainz, B., Wolff, H.J.: “Calculation of prestressed concrete structures
in the serviceability limit state under consideration of the time-dependent
behaviour of concrete”, Beton- und Stahlbetonbau, 1971, Nos. 9 & 10 (in
German)
6.2 Menn, C.: ”Long term processes in concrete structures”, Lecture Notes, ETH-
Zürich (in German)
6.3 König, G., Gerhardt, H.C.: ”Redistribution of internal forces and moments due to
creep and shrinkage of concrete in reinforced and prestressed concrete structures”,
Mitteilungen aus dem Institut für Massivbau der TU Darmstadt, Nr. 34, Ernst &
Sohn (in German).
6.4 Bazant, Z.P.: “Mathematical modeling of Creep and Shrinkage of Concrete”, J.

Wiley and Sons, Chichester.
April 2018 6-42

Chapter 7 – Bending moment resistance
7. Bending moment resistance

7.1 Design philosophy
After the structure is designed, it must be shown that SLS requirements concerning
maximum initial concrete compressive stress (EN 1992-1-1 cl. 5.10.2.2) and concrete
tensile stress or crack width (EN 1992-1-1 cl. 7.3) are met and that structural resistance
meets the ULS requirements. With regard to ULS, one of the resistances to check is the
bending moment resistance of the structure.
7.1.1 Minimum reinforcement
To prevent brittle failure of a structure, it should be provided with a sufficient amount of

reinforcing and/or prestressing steel. From this point of view an uncracked cross-section
should not have a higher bending moment resistance than the cracked cross-section
provided with steel. If this were not the case, the cross-section might be uncracked up to a
load that can not be resisted once the cross-section actually cracks, resulting in brittle
failure. A basis of design is to apply an amount of steel that is at least able to resist the
cracking moment of the structure. This is denoted as minimum reinforcement. On the
other hand, there is a very small probability that the cracking moment is actually greater
than the design bending moment including an extra safety factor. Therefore, EN 1992-1-1
cl. 9.2.1.1 states that minimum reinforcement has not to be greater than the amount
required to resist 1,2 MEd. In the Dutch National Annex, this value is raised to 1,25 MEd to
comply with NEN 6720 (VBC 1995).
The minimum amount of reinforcement required to resist the cracking moment follows
from the cracking moment of a cross-section. Figure 7.1 shows a statically determinate
beam prestressed by a straight tendon having an eccentricity ep relative to the centroidal
axis (c.a.) and a working prestressing force Pm,.
April 2018 7-1

c.a
Fig. 7.1 Calculation of the cracking moment
A cross-section at midspan is considered. The prestressing introduces an axial

compressive force Pm, and a bending moment Pm, ep. If a load qg + qq is applied (for
instance caused by selfweight and a variable load), the bending moment in the cross-
section at midspan is Mg+q = 1/8 (qg + qq ) l2. Cracking occurs if:
M g+q Pm, ep Pm,

   f cr (7.1)
Wcb Wcb Ac
where fcr is the concrete flexural tensile strength (r = rupture) (EN 1992-1-1 eq. (3.23)).
The bending moment (ME,r) at which the cross-section cracks, follows from:
 P 
M g+q  Pm, ep   f cr  m,  Wcb (7.2)
 Ac 
To prevent brittle failure to occur at the onset of first cracking, the bending moment
resistance MRd must be greater than the cracking moment from eq. (7.2). However, there
is a limit: MRd has not to be greater than 1,25 MEd (EN 1992-1-1 cl. 9.2.1.1 (1) & NL
National Annex) where MEd is the design bending moment.
April 2018 7-2

Reinforced concrete
The minimum reinforcement in a reinforced concrete beam with a rectangular cross-

section (no prestressing) follows from the cracking bending moment of the beam.
Cracking is assumed to occur when the concrete reaches its flexural tensile strength (EN
1992-1-1 eq. (3.23)). In case of a rectangular cross-section, the cracking moment is:
M E,r  M cr  16 bt h 2 f ctm,fl
The bending moment resistance of an amount of reinforcement As,min is approximately
M Rd  0,9d f yd As,min
Requirement to prevent brittle failure:
M Rd  M E,r
0,9d f yd As,min  16 bt h 2 f ctm,fl
Using fyd = fyk/1,15 and d ≈ 0,9 h results in EN 1992-1-1 eq. (9.1N):
f ctm,fl
As,min  0, 26 bt d
f yk (7.3)
As,min  0, 0013 bt d
Prestressed concrete
Note that eq. (7.3) does not hold for non-rectangular cross-sections, nor for prestressed
concrete. When following the same approach for prestressed concrete, the cracking
moment is calculated using the following expression:
 P   Pm, 
M cr   f cr  m,  Wcb   f ctm,fl   Wcb
 Ac   Ac 
Note that the minimum amount of reinforcement has not to be based on a bending
moment greater than 1,25 MEd (EN 1992-1-1 cl. 9.2.1.1 (1) & NL National Annex).
Also note that in the case of prestressed concrete the cracking moment includes the axial
compressive stress introduced by the prestressing. Prestressing might be regarded as a
fictitious increase of the concrete flexural tensile strength.
The Dutch National Annex to EN 1992-1-1 takes into account the influence of an axial
compressive or tensile force on the cracking moment. Moreover, in this annex equation
April 2018 7-3

(9.1N) from EN 1992-1-1 is deleted and presented in a more general way, making it
applicable to any cross-section. The result is a set of expressions that is similar to the
expression for Mcr that holds for prestressed concrete.
EN 1992-2 ‘Concrete bridges – Design and detailing rules’ follows a different approach
to prevent brittle failure to occur. Any of the following methods may be used (EN 1992-2
cl. 6.1 (109)):
 Calculate (1) the applied bending moment from the frequent combination of actions
and (2) the reduced area of prestress that results in a tensile stress fctm at the extreme
tension fibre in the cross-section considered. Now calculate the ultimate flexural
capacity of the cross-section, adding reinforcing steel to resist the moment due to
the frequent load combination. (Note: Reduced partial material safety factors are
used).
 Calculate the cracking bending moment of a cross-section, using an appropriate
concrete tensile strength at the extreme tension fibre, ignoring any effect of
prestressing. Provide a minimum reinforcing steel area such that the cracking
moment calculated can be resisted by the reinforcing steel only. Note that this
calculation method results in the use of reinforcing steel, whereas in some
pretensioned members, e.g. hollow core slabs, only prestressing steel is applied.
The code then allows the use of an alternative approach in which the pretensioned
tendons can be effective in As,min.
 Agree on an appropriate inspection regime (Note: Not allowed in The Netherlands).
7.1.2 Rotational capacity
The rotational capacity of a member is its capacity to deform. Rotation is the product of
length and curvature. Therefore, the rotational capacity depends directly on the
compression zone height in a cross-section which is a measure for the curvature. When is
rotational capacity required or beneficial for the behaviour of a structure? Rotational
capacity is required in case plastic hinges are assumed in a calculation based on the
theory of plasticity or when bending moment redistribution is applied. In case of
statically indeterminate structures, sufficient rotational capacity enables the structure to
resist, up to a certain limit, differential settlements and imposed deformations, without
them having substantial impact on structural resistance. Figure 7.2 contains an example.
If the beam shown in fig. 7.2 is reinforced according to the theory of elasticity, yielding
of the steel at the span and support cross-sections occurs at the same moment. A
differential settlement (fig. 7.2b), however, introduces an imposed support bending
moment which makes that the yield moment is reached at the support first. The rotational
capacity of the plastic hinge at the support (fig. 7.2c) makes that the yield moment can
now also be developed in the span, before failure occurs at the support (caused by
reaching the ultimate strain of the concrete, e.g. εcu3 in EN 1992-1-1 fig. 3.4). Sufficient
rotational capacity makes that, in general, the designer does not have to take into account
the effect of imposed deformations on structural resistance (EN 1992-1-1 cl. 2.3.1.2 (2) &
cl. 2.3.1.3 (2)).
April 2018 7-4

Fig. 7.2 Example of a differential settlement of a support and the development of the
full plastic moment of resistance
According to EN 1992-1-1 cl. 5.5 the bending moments from linear elastic analysis in a
structure primarily loaded in bending can be redistributed to a limited extent. The extent
to which redistribution is allowed, depends on the concrete strength class, the height of
the concrete compression zone in ULS after redistribution on the basis of the
reinforcement applied, and the type of reinforcing steel used. According to EN 1992-1-1
cl. 5.5 (eqs. (5.10a) & (5.10b)) the following holds:
 0, 0014  xu
  0, 44  1, 25   0, 6   for fck < 50 N/mm2
  cu2  d
 0, 0014  xu
  0,54  1, 25   0, 6   for fck > 50 N/mm2
  cu2  d
  0, 70 in case of reinforcement class B and C

  0,80 in case of reinforcement class A
where:
δ is the ratio of the redistributed moment and the moment from an elastic analysis;
xu is the concrete compression zone height (= depth of neutral axis) after
redistribution;
d is the effective depth of the cross-section.
It is not allowed to apply redistribution in case the rotational capacity cannot be defined
with confidence, as is the case for columns (EN 1992-1-1 cl. 5.5 (5) & 5.5 (6)).
The expressions use εcu2 as ultimate strain of concrete. It has the same value as εcu3 from
EN 1992-1-1 fig. 3.4 (see EN 1992-1-1 table 3.1).
April 2018 7-5

EN 1992-1-1 makes it possible to carry out a plastic analysis, provided that additional
requirements are met. One of these is a more strict reduction of the concrete compression
zone height compared to the one that holds in case of limited redistribution. Since
prestressed concrete structures are usually not checked using plastic analysis, this item is
not further addressed.
It is noted that the NL National Annex to EN 1992-1-1 changes the expressions:
f x
  u for fck < 50 N/mm2
500  f d
7f x
  u for fck > 50 N/mm2
 cu 10  7 f d
6
where
 f pk 
   pm,  Ap  f yd As

f  s 
Ap  As
Note that f is in N/mm2 and must not be mistaken for the drape of a tendon.
The reasoning behind the introduction of f is that the expressions from EN 1992-1-1 seem
to be applicable in case of reinforced concrete structures only. Therefore, the equations
were extended to account for possible prestressing steel too. Additionally, the equations
obtained for reinforced concrete only (Ap = 0) were tuned to give the same results as NEN
6720 (VBC 1995). The modifications can be relevant in design: in case of reinforced
concrete and no redistribution (δ = 1) EN 1992-1-1 results in xu / d < 0,448 for < C50/60;
the Dutch National Annex results in xu / d < 0,535 for < C50/60 and steel B500 (fyd =
435 N/mm2).
In case the relative compression zone height exceeds its limit value, the designer has
several options. For instance, it can be decided to increase the height of the cross-section.
The increase of internal lever arms will reduce the amount of steel required. It’s also an
option to apply reinforcement in the compression zone. The compression zone height is
reduced since the part of the compressive force now carried by reinforcing steel has not
to be carried by concrete. It is noted that this option is not preferred in The Netherlands.
The Dutch National Annex to EN 1992-1-1 provides another option, namely performing
the check taking into account a fictitious amount of (prestressing and reinforcing) steel.
This is the amount of steel required to resist the design bending moment in the cross-
section. The reasoning behind this is as follows: The amount of prestressing steel
required in a concrete structure often follows from SLS requirements (e.g. no tensile
stresses are allowed or crack widths must be limited), not from ULS requirements. A
ULS cross-sectional analysis then often demonstrates that the bending moment resistance
is (much) greater than the design bending moment (MRd >> MEd). From a ULS point of
April 2018 7-6

view, the structure is provided with (much) more steel than required. This enhances
structural safety, but has a negative impact on the rotational capacity. To compensate for
the latter, the designer then is allowed to fictitiously remove steel from the cross-section
when calculating the compression zone height. This of course provided that the modified
(fictitious) cross-section can resist the design bending moment (MRd,fictitious > MEd).
7.1.3 Principles of moment resistance design
The bending moment resistance of a cross-section with prestressing steel is determined in

a similar way as for a reinforced concrete cross-section.
Plane sections remain plane (Bernouilli’s law): there is a linear relationship between the
distance of a fiber to the neutral axis and the strain in the fiber. It is noted that this
relationship does not hold for deep beams and walls (EN 1992-1-1 cl. 5.3.1 (3)).
A cross-section is in equilibrium: ΣH = 0 and ΣM = 0.
The stress-strain relationships of concrete, reinforcing steel and prestressing steel are
presented in figs. 7.3-7.5.
Fig. 7.3 Stress-strain relationship of concrete in compression in ULS
April 2018 7-7

The shape of the stress-strain relationship of concrete is described by the surface factor α
and the centre of gravity factor β:
concrete strength class α β

< C50/60 0,75 0,39
C53/65 0,72 0,38
C55/67 0,71 0,37
C60/75 0,67 0,36
C70/85 0,62 0,35
C80/95 0,58 0,34
C90/105 0,56 0,34
Fig. 7.4 Stress-strain relationship of reinforcing steel in ULS
Fig. 7.5 Stress-strain relationship of prestressing steel in ULS, including significant

design points for steel Y1860
April 2018 7-8

The bending moment resistance of a cross-section is based on the equivalent prestressing

load method, as discussed in Chapter 5. The influence of the prestressing is taken into
account as external loads which act on the structure. Bending moment, axial force and
shear force diagrams can be constructed and used as input in cross-sectional analysis.
This method holds for both statically determinate and indeterminate structures.
7.2 Moment resistance according to the equilibrium method

This design method has the same calculation procedure as the design method used for a
reinforced concrete cross-section. The differences are:
1. For reinforced concrete, the stresses in both the steel and the concrete are zero at the
beginning of loading. For a structure with both reinforcing and prestressing steel, the
initial stresses differ from zero because the prestressing steel is pre-tensioned. In
order to meet the compatibility condition with respect to the deformation, the
difference in strain between reinforcing steel and prestressing steel has to be
accounted for.
2. The prestressing loads (Mp and Np = Pm,) are already incorporated in the externally
applied loads (MEd and NEd). Therefore, it is not allowed to use the full capacity of the
prestressing steel when determining the moment resistance; the capacity should be
reduced by the working prestress (σpu - σp).
Just as for reinforced concrete, the following conditions must be taken into account when
determining the bending moment resistance of a prestressed cross-section:
 The concrete reaches the ultimate compressive strain εcu2 if a parabolic stress-strain
relationship is used; εcu3 for a bi-linear stress-strain relationship (fig. 7.3).
 At least a part1 of the reinforcing steel in the cross-section yields. This implies that εs
> fyd / Es. The ultimate strain εud of the reinforcing steel is only of importance when
using the upward sloped part of the curve (fig. 7.4). When the horizontal branch is
used, the steel strain has not to be checked (EN 1992-1-1 cl. 3.2.7).
 The strain in the prestressing steel is higher than the “computational value” of the
0,1% proof-stress fpd = fp0,1k / s. The strain limit εud (EN 1992-1-1 cl. 3.3.6) of the
prestressing steel only has to be accounted for when the sloped branch of the curve is
used; not if the horizontal part is used (fig. 7.5).
Note that the ultimate strain of concrete is usually indicated by per mille ( 0 00 ), whereas
the ultimate strain of both reinforcing and prestressing steel is often expressed in percent
(%)!
Figure 7.6 shows a beam loaded by its selfweight (qg), a variable load (qq) and
prestressing (qp). In the midspan cross-section, the following moment is present from the
design loads:
1
In case more than one layer of reinforcement is applied, it might be possible that only the layer having the
most eccentric position with respect to the neutral axis yields.
April 2018 7-9

1
M Ed 
8
  g qg   q qq   p qp  l 2   p Pm, epo
where g = 1,2 for permanent loads, g = 1,5 for variable loading and p = 1,0 for the
prestressing load.
The calculation of the bending moment resistance MRd is discussed step by step. It is
assumed that (MEd, NEd) > (Mcr, Ncr) where Mcr and Ncr are the moment and force for
which cracking occurs, respectively. Structural safety requires that MRd > MEd.
c.a.
Fig. 7.6 Beam subjected to permanent load, variable load and prestressing load
1) First, the height of the compressive zone is estimated. For that purpose the
compressive forces are determined. When the reinforcing steel yields (using the
horizontal branch from fig. 7.4 in the calculation), the total force in this steel is As fyd.
When the prestressing steel is in the plastic state (using the horizontal branch from
fig. 7.5), its total force is Ap fpd. Since equilibrium of horizontal forces is required, it
should hold (see fig. 7.7):
N c  Pm,  N p  N s  Ap  p,  Ap  f pd   p,   As f yd
N c   b xu f cd (in case of a rectangular compressive zone cross-section) (7.4)

For a rectangular compressive zone cross-section, a concrete strength class < C50/60
and a bi-linear stress-strain relationship (EN 1992-1-1 fig. 3.4) α = 0,75.
From this first approximation the height xu of the compressive zone is obtained. In
case of a rectangular concrete compressive zone cross-section it reads:
As f yd  Ap  f pd   p,   Pm,
xu  (7.5)
 b f cd
April 2018 7-10

(in case of a
compression zone
having a constant width)
M Rd
c.a.
N p  Ap  f pd   p, 
Fig. 7.7 Equilibrium between external and internal forces
2) Check whether the height xu of the compressive zone meets the requirement with
regard to the maximum height of the compressive zone according to the Dutch
National Annex to EN 1992-1-1 cl. 5.5:
f x
500  f d
where:
 f pk 
   pm,  Ap  f yd As

f  s 
Ap  As
and where δ is the ratio of the redistributed moment to the elastic bending moment
(according to the linear theory of elasticity).
3) With the obtained height xu of the compressive zone, the concrete and prestressing
steel strains can be determined (fig. 7.8).
April 2018 7-11

c.a.
Fig. 7.8 Determination of the strains in concrete and prestressing steel for a certain
height of the compressive zone xu
The strain εs in the reinforcing steel follows from:
 cu  cu   s
 
xu ds
 ds 
 s   cu   1 (7.6)
 xu 
The increase of the strain in the prestressing steel Δεp can be determined in a similar
way:
 cu  cu   p
 
xu dp
 dp 
 p   cu   1 (7.7)
 xu 
The total strain in the prestressing steel is εp = εp, + Δεp.
4) With these strains εs and Δεp, the stresses in the reinforcing and prestressing steel σsu
and σpu can be calculated. From these strains the force in the reinforcing steel and the
increase of the force in the prestressing steel are obtained:
N  N s  N p  As  su  Ap  pu   p,  (7.8)
It is noted that the stress in the reinforcing steel often will be σsu = fyd, but of course
this has to be checked.
April 2018 7-12

5) Next it should be checked whether the following condition is satisfied:
N  Pm,  N c (  b xu f cd )  N  N c  Pm, (7.9)
In principle three solutions are possible:
a. N  N c  Pm,  The height xu of the concrete compressive zone is assumed too

large.
b. N  N c  Pm,  The height xu of the concrete compressive zone is assumed
correctly.
c. N  N c  Pm,  The height xu of the concrete compressive zone is assumed too
small.
If the height xu of the concrete compressive zone appears to be assumed (almost)

correctly, the calculation can be continued with step 6, which is the calculation of the
resistance (or failure / ultimate) moment of the cross-section. In all other cases, a new
height of the concrete compressive zone that is somewhat larger or smaller should be
assumed, after which the steps from step 2 in the calculation must be repeated. After a
number of iterations the correct height of the concrete compressive zone is found.
Fig. 7.9 Graphical representation of the iteration process for the determination of the
height of the concrete compressive zone xu
Figure 7.9 shows a graphical representation of the iteration process. For the first
assumed compressive zone height xu, the magnitude of the concrete compressive
force Nc from eq. (7.4) and the steel force N + Pm, (with N from eq. (7.8)) are
presented in horizontal direction, using the solid vertical line from the strain diagram
as axis. If N + Pm, is larger than Nc, as is the case in the first iteration step shown in
fig. 7.9, the compression zone height is too small and the calculation has to be
repeated using a larger xu. So, with graphical support as shown in the figure, the
correct value for xu can be found for which axial equilibrium is satisfied.
April 2018 7-13

6) With the correct height xu of the concrete compressive zone, the magnitude of the
bending moment resistance MRd can be determined. This follows from (also see fig.
7.10):
M Rd  As  su  ds  y   Ap  pu   p,  d p  y   Pm,  zc  y  (7.10)
where:
y is the position of the resulting compressive force;
zc is the distance from the controidal axis to the concrete fiber that has the
highest compressive strain.
MRd relative to S
centroidal axis
Fig. 7.10 Determination of the bending moment resistance MRd
For a rectangular cross-section y = β xu; for concrete strength class < C50/60 β = 7/18 =
0,39.
Example
As an example, the bending moment resistance of the cross-section shown in fig. 7.11 is
determined. The data relevant for this calculation are:
Ac = 0,36 m2 ep = es = 0,442 m (relative to centroidal axis)
dp = ds = 0,820 m (effective depth)
zcb = 0,622 m zct = 0,378 m

concrete C35/45 Ec = 34 · 103 N/mm2
fcd = 23,3 N/mm2
prestressing steel Y1860S7 (7 wires per strand); total: Ap = 2000 mm2

Ep = 195 · 103 N/mm2 (strands)
σp, = 1080 N/mm2
April 2018 7-14

friction loss between end and midspan section of beam is 2%:

σp, = 0,98 · 1080 = 1058 N/mm2
fpd = 1522 N/mm2
reinforcing steel B500

Es = 200 · 103 N/mm2
As = 905 mm2 (8 bars Ø 12 mm)
MRd
10,7 N/mm2
As fyd
Fig. 7.11 Example of the calculation of the moment resistance MRd
Figure 7.11a shows the midspan cross-section of the beam. The equilibrium between the
internal and external loads is visualised in fig. 7.11d. At the left hand side of the element
the following loads act: the centrical prestressing force Pm, and the flexural bending
moment MRd.
In the case considered, MEd is:
M Ed   g M g   q M q   p M p (7.11)
where:
April 2018 7-15

Mg is the bending moment from permanent loads (selfweight and static loads);
Mq is the bending moment from variable loads;
Mp is the bending moment from prestressing (only caused by an upward tendon
pressure since in this case the tendon is anchored in the centroidal axis at both beam
ends).
Now a stepwise determination of the bending moment resistance is discussed.
Step 1
The stress in the prestressing steel at the moment of failure is estimated to be σpu = fpd =
1522 N/mm2. Further, it is assumed that the reinforcing steel yields. The following
equilibrium equation holds in case of a rectangular cross-section:
N c  Pm,  As f yd  Ap  f pd   p,  
 b xu f cd  Ap p,  As f yd  Ap  f pd   p,   (7.10)
Ap f pd  As f yd 2000 1522  905  435

xu    246 mm
 b f cd 0, 75  800  23,3
Note that the working prestressing force is incorporated in the load (eq. (7.10)). The part
σp, of the tensile strength of the prestressing steel is used already. That is why the
component Ap (fpd – σp,) is in the right-hand side part of eq. (7.10).
The value found for xu is larger than the height of the flange (200 mm). This implies that
the force in the compressive zone is overestimated, since part of it has a width equal to
the web width (250 mm), not the flange width (800 mm). The actual compressive zone
height must be greater to resist the reinforcing and prestressing steel force assumed. If the
800 mm flange width would be available over the full compressive zone height, the
concrete compressive force is 3444 kN. The compressive zone height must be increased
to xu = 258 mm to carry the same force (3444 kN) when taking into account that part of
the compression zone width is reduced to web width. Increasing the compressive zone
height reduces the internal lever arms. In this case, however, its impact is small since the
part of the web loaded in compression has a minor contribution to the compressive force
(76 kN; about 2%).
Step 2
Now the height of the compressive zone is checked according to the Dutch National
Annex to EN 1992-1-1 cl. 5.5:
f x
500  f d
where:
April 2018 7-16

 f pk 
   pm,  Ap  f yd As

f  s 
Ap  As
and where δ is the ratio of the redistributed moment to the elastic bending moment
(according to the linear theory of elasticity).
No redistribution is applied, so δ = 1,0.
 f pk 
  pm,  Ap  f yd As  
1860
  1058   2000  435  905
 1,1
f  s     571 N/mm 2
Ap  As 2000  905
The following condition is found:
xu f 500 500
 1    0, 47  xu  0, 47  820  385 mm
d 500  f 500  f 500  571
This condition is by far met.
Step 3
For the assumed height of 258 mm for the compressive zone, the strain in the reinforcing
steel is (also see fig. 7.11f):
 ds   820 
 s   cu   1  3,5 103    1  7, 6 103
 xu   258 
The centers of gravity of the reinforcing steel and the prestressing steel coincide. The
strain in the prestressing steel directly follows from:
1058
 p   p   pw  0, 0076   13, 0 103
195 10 3
Step 4
The stress in the prestressing steel assumed in the analysis is fpd = 1522 N/mm2. This
stress is reached at a strain fpd / Ep = 1522 / 195 · 103 = 7,81 · 10-3 (the strain at the kink
in the stress-strain relationship from fig. 7.5). The actual strain is higher (13,0 · 10-3) and,
as a result, the actual stress will be higher than the 1522 N/mm2 assumed.
The strain at which εud is reached depends on the type of prestressing steel. For strands,
εuk = 35 · 10-3 can be assumed.
April 2018 7-17

The prestressing steel stress at the actual 13,0 · 10-3 strain then is:
 13, 0 103  7,81103 

3  
 pu  1522   3
 1691  1522   1522  32  1554 N/mm 2
 35 10  7,8110 
The force in the reinforcing steel is:
N s  As f yd  905  435  394 103 N
and the increase of the force in the prestressing steel is:
N p  Ap  pu   p,   2000  1554  1058   992 103 N
Summation yields:
N  Ns  N p  1396 kN
Step 5
The strain and the concrete compressive stress at the bottom of the compression flange
are:
 258  200 
 c,flange  3,5 103     0,8 10
3
 258 
 c,flange 0,8 10 3
 c,flange  f cd    23,3  10, 7 N/mm 2
 c3 1, 75 10 3
The stress distribution over the concrete compressive zone is presented in fig. 7.11g.
The concrete compressive force is:
N c  800  12  258  23,3   200  12  258   800  12   23,3  10, 7   12  250   258  200  10, 7
  2404, 6  965, 6  77, 6  103 N  3448 kN
Now it will be checked if the requirement from horizontal force equilibrium is satisfied.
In this case it should hold:
N  Pm,  N c
Substitution of these values using Pm, = 2000 · 1058 = 2116 · 103 N results in:
1396  2116  3512  3367 kN
It can be concluded that the height of the compressive zone is somewhat underestimated.
Therefore, the calculation will be repeated using a higher value of xu.
April 2018 7-18

After some trial and error it is found that xu = 265 mm.

Reinforcing steel strain:
 ds   820 
 s   cu   1  3,5 103    1  7,3 103
x
 u   265 
Prestressing steel strain:
1058
 p   p   pw  7,3 103   12, 7 103
195 10 3
Prestressing steel stress:
 12, 7 103  7,81103 

3  
 p  1522   3
 1691  1522   1522  30  1552 N/mm 2
 35 10  7,8110 
The strain and the concrete compressive stress at the bottom of the compression flange
are:
 265  200 
 c,flange  3,5 103     0,86 10
3
 265 
 c,flange  11, 4 N/mm 2
The bending moment resistance of the cross-section is calculated relative to the centroidal
axis. It is first assumed that the cross-section has a width equal to the flange width over
its full height. Then, the actual web width is taken into account by extracting the part of
the cross-section that’s actually not present:
 x h 
M Rd   bflange xu f cd  zct   xu   0,5  xu  hflange  bflange  bweb   c,flange  zct  hflange  u flange  
 3 
 As f yd  ds  zct   Ap  p  d p  zct  
 0, 75  800  265  23,3  (378  0,39  265) 
 265  200 
 0,5  (265  200)  (800  250) 11, 4   378  200  
 3 
 905  435  (820  378)  2000  (1557  1058)  (820  378)  1603 106 Nmm
(7.12)
Forces in the cross-section:

concrete compression zone: 3705 - 205 = 3500 kN
reinforcing steel 394 kN
prestressing steel 2116 + 988 = 3104 kN
April 2018 7-19

Resulting force: 3500 - 394 - 3104 = +2 kN; OK
The design bending moment in ULS:
1
M Ed 
8
  g qg   q qq   p qp  l 2 (7.13)
The selfweight of 0,36 · 25 = 9,0 kN/m and a static load of 1,5 kN/m, result in qg =
10,5 kN/m.
For a working prestressing force Pm, = 2116 kN and a drape f = 0,62 – 0,18 = 0,44 m
(fig. 7.11b), it is found that:
qp  8 Pm, f / l 2  8  2116  0, 44 252  11,9 kN/m
(in upward direction; see minus sign in eq. (7.13)).
With g = 1,2; q = 1,5 and p = 1,0, eq. (7.12) and (7.13) demonstrate that a live load qq =
13,2 kN/m can be resisted by the structure with sufficient safety against failure:
1
1603   1, 2 10,5  1,5 qq  1, 0 11,9   252 (7.13)
8
7.3 Determination of the resistance of statically indeterminate

structures
When representing prestressing forces using the equivalent prestressing load method, the
control of the resistance of statically indeterminate structures is no longer a problem. A
simple equilibrium condition is sufficient. This is illustrated on the basis of fig. 7.14. It is
asked to control the moment resistance at the mid support.
April 2018 7-20

moment from Mp1

upward loading
0,5 Pm, e0
Pm, e0 Mp2
c.a.
Fig. 7.12 Control of bending moment resistance at an intermediate support in a

statically indeterminate structure.
In this case the moment introduced by the equivalent prestressing load (by tendon
curvature) qp is:
8 P f 
M p1  18 qp l 2  18  m,2  l 2  Pm, f
 l 
However, there is also a second component due to prestressing. This is caused by the
eccentric introduction of the prestressing force at both ends of the structure, which also
causes a moment at the intermediate support. This moment is:
M p2   12 Pm, eo
Furthermore, in the centre of gravity of the cross-section an axial compressive force Pm,
acts.
April 2018 7-21

The bending moments Mp1 and Mp2 and the axial force Pm, do not change when the
structure is loaded to failure. For all loads originating from prestressing, the load factor is
p = 1,0, so:
M p   p ( M p1  M p2 )  Pm, ( f  12 eo )
and
N p   p Pm,  Pm,
Because the internal (resistance) moment and the external (design) moment at the
intermediate support must be in equilibrium, the next condition should be satisfied:
zs As  su  zp Ap ( pu   p, )   g M g   q M q  M p  Pm, ec (7.14)
where:
zp distance from the concrete compressive force Nc to the centre of gravity of the
prestressing steel
zs distance from the concrete compressive force Nc to the centre of gravity of the
reinforcing steel
ec distance from the concrete compressive force Nc to the line of action of the axial
prestressing force Pm, introduced at the ends of the structure.
The left hand side of eq. (7.14) represents the moment resistance MRd. The right hand side
is the design moment MEd. If, at the control, it turns out that MRd < MEd, then reinforcing
steel As should be added until MRd ≥ MEd. It is of course also an option to increase the
prestressing.
April 2018 7-22

Chapter 8 - Shear
8. Shear
8.1 Introduction
In this chapter the shear serviceability and ultimate limit state analysis of prestressed
elements is discussed. As an introduction a recapitulation is given on the shear analysis of
non-prestressed structural elements containing longitudinal reinforcement only (section
8.2.1) and also provided with shear reinforcement (section 8.2.2). In section 8.3 the shear-
stress analysis of prestressed beams is discussed. SLS is discussed in section 8.3 and ULS in
section 8.4. In the last section (8.5) a special item, namely vertical prestressing of the web, is
discussed.
8.2 Shear behaviour of elements not containing prestressing

8.2.1 Reinforced elements without shear reinforcement
To better understand the influence of prestressing on the shear behaviour, a summary is given
of the shear behaviour of an “ordinary” reinforced concrete structure. Figure 8.1 shows a
beam subjected to two symmetrically positioned point loads. Between the two external loads
the beam is loaded in pure bending.
Fig. 8.1 Crack formation due to shear and bending in a reinforced beam without shear
reinforcement
At increasing load, one of the flexural cracks (caused by bending) will develop into a shear
crack that propagates over a large distance. As a consequence, internal equilibrium is lost and
failure occurs. This case, where a critical shear crack develops out of a flexural crack, is
denoted as flexural-shear failure. The area of concrete in compression above the advancing
inclined crack is that much reduced, that it can no longer support the compressive force
caused by flexure. Since the behaviour of this type of failure is quite complicated, the
analysis is based on empirically derived expressions. A good prediction of the 5% lower limit
of the shear capacity of beams without shear reinforcement is obtained by the following
expression (8.1):
1
 d 3 1
VRk,c  0,15  3   100 l f ck  3 bw d (8.1)
 a
April 2018 8-1

Chapter 8 - Shear
where:
VRk,c is the characteristic shear force capacity;

a is the distance from the concentrated load to the support;
200
ξ is a size-effect coefficient;   1   2, 0 where d is in mm;
d
ρl is the reinforcement ratio based on the web width and the effective depth d of the
cross-section; ρl = Asl / (bwd)
bw is the web width;
d is the effective depth;
fck is the characteristic cylinder compressive strength of the concrete.
Expression (8.1) has a coefficient of variation of 12,5% (for the mean value VRm the coefficient
0,15 has to be replaced by 0,18 and the characteristic value of the cylinder compressive strength
has to be replaced by its mean value fcm). A design value is obtained by dividing VRm from eq.
(8.1) by c = 1,5.
EN 1992-1-1 cl. 6.2.2 (1) presents an expression that is similar to eq. (8.1):
 1

VRd,c  CRd,c k 100 l f ck  3  k1 cp  bw d
  (8.2)
VRd,c  vmin  k1 cp  bw d
where:
VRd,c is the design value of the shear force capacity;

200
k is a size-effect coefficient; k  1   2, 0 where d is in mm;
d
ρl is the reinforcement ratio of the longitudinal reinforcement based on the web width
A
and the effective depth d of the cross-section; l  sl  0, 02 ;
bw d
bw is the smallest width of the cross-section in the tensile area;
N
σcp is the axial stress caused by loading or prestressing;  cp  Ed  0, 2 f cd ;
Ac
NEd is the axial force (NEd > 0 for compression);
Ac is the cross-sectional area of the concrete;
fcd is the design cylinder compressive strength of the concrete;
fck is the characteristic cylinder compressive strength of the concrete.
3 1
vmin  0, 035k 2 f ck2
April 2018 8-2

Chapter 8 - Shear
When an element is subjected to an axial compressive force, the shear capacity of that
element increases since the cracks are partially closed and crack growth is reduced. An axial
tensile force has the opposite effect. The Dutch Annex of EN 1992-1-1 uses the
recommended values: CRd,c = 0,12 and k1 = 0,15.
8.2.2 Reinforced elements with shear reinforcement
If the design shear force is higher than the shear resistance of a member that contains no
shear reinforcement, shear reinforcement has to be provided to increase the resistance. As a
result of this provision, the behaviour significantly changes: the formation of an inclined
crack does not anymore result in failure and a new load transfer mechanism that can be
described by a truss model develops, see fig. 8.2. The load transfer of a truss model is
discussed first.
Tensile ties in the truss model
It is assumed that the truss has to carry a design shear force VEd. At first, the required amount
of shear reinforcement is calculated. It is assumed that closed stirrups (two legs active per
strirrup; total cross-sectional area Asw for each stirrup) are spaced at a distance s. In the truss
model a number of stirrups are represented by one tensile tie.
A tensile tie is the equivalent for the stirrups in the beam over a distance z (cot + cot), see
fig. 8.2, where z is the distance from the tensile reinforcement to the resulting concrete
compressive force. The tensile bar has an equivalent steel cross-sectional area of Aequi:
Asw
Aequi  z (cot   cot  ) (8.3)
s
where θ is the angle of the compression struts and α is the inclination of the tensile ties.
On the basis of equilibrium it follows that, in case of a full truss model, the tensile force NT
in a tensile tie in ULS is:
VRd,s
NT  (8.4)
sin 
where VRd,s is the shear force that is carried by the shear reinforcement, see fig. 8.2.
In case the stress in the shear reinforcement reaches the design yield stress fywd, it follows
from eqs. (8.3) and (8.4) that
Asw V
z (cot   cot  ) f ywd  Rd,s
s sin 
April 2018 8-3

Chapter 8 - Shear
internal
lever arm
Fig. 8.2 Truss model with compression struts at an angle  and inclined tensile ties at an
angle 
The shear force that can be resisted follows from the applied amount of shear reinforcement
per unit length of the beam (EN 1992-1-1 eq. (6.13)):
Asw
VRd,s  z f ywd (cot   cot  ) sin  (8.5)
s
In case of vertical stirrups ( = 90º) and compression struts at an angle  = 45º, eq. (8.5)
becomes:
Asw
VRd,s  z f ywd (8.6)
s
Tests showed that designing shear reinforcement using eq. (8.5) is conservative. This is
caused by the following aspects:
 the pure truss is a simple representation of the actual behaviour: the connections between
the truss bars are not perfect hinges;
 redundancy from dowel action of the longitudinal reinforcement;
April 2018 8-4

Chapter 8 - Shear
 frictional forces in the cracks (aggregate interlock);

 the uncracked concrete compression zone carries a part of the shear force;
 direct load transfer to the support.
From series of tests it turned out that the extra resistance (often denoted as “concrete”
capacity) is equal to the bearing resistance of the member not provided with shear
reinforcement (eq. (8.2)). It was, therefore, often assumed that the shear resistance is the sum
of the concrete and steel resistances from eqs. (8.2) and (8.6). This approach was followed in
the Dutch code NEN 6720.
Additional research demonstrated that this approach can be modified. Results from shear
resistance experiments demonstrated that, at failure, the first initial bending/shear-cracks are
crossed by a second crack pattern that has a different angle to the beam axis than the first
pattern. The difference between both crack pattern angles depends on the amount of shear
reinforcement applied: the smaller the amount of shear reinforcement, the larger the
difference. The failure mode seems to adapt itself to the shear reinforcement applied: the
smaller the amount of shear reinforcement, the smaller the angle to the beam axis of the
second crack pattern and the larger the number of shear reinforcement bars crossed and
activated by an inclined shear crack. This is denoted as rotation of the compression
struts/diagonals. When applying this reduced crack angle in eq. (8.5), the shear resistance is
accurately predicted. Figure 8.3 shows an example in which the resistance of the truss not
taking into account an additional “concrete” part is shown. This line is denoted as Mörsch’s
line.
Fig. 8.3 Shear force test: stress in shear reinforcement as a function of the shear force
from a test and for truss models with different compression strut angle 
The angle  can be reduced to 18,4º. EN 1992-1-1 assumes a lower limit value  = 21,8º
(cot 21,8º = 2,5).
April 2018 8-5

Chapter 8 - Shear
In case vertical shear reinforcement is applied ( = 90º), eq. (8.5) reduces to EN 1992-1-1 eq.
(6.8):
Asw
VRd,s  z f ywd cot  (8.7)
s
Compression struts in the truss model
The width of a strut in the plane of a beam is indicated in fig. 8.4. The width is equal to the
distance bD between the struts, which is:
bD  z (cot   cot  ) sin 
So, the cross-sectional area of a strut is:
AD  bw bD  bw z (cot   cot  ) sin  (8.8)
Fig. 8.4 Calculation of the dimensions of the compression struts from the geometry of the
truss model
From force equilibrium it follows that the force in the compression strut is:
VEd
ND  (8.9)
sin 
where VEd is the design value of the shear force. When using eqs. (8.8) and (8.9), the concrete
compressive stress σcD in the strut follows from:
VEd 1 VEd 1  cot 2 

 cD     (8.10)
bw z (cot   cot  ) sin 2  bw z cot   cot 
The resistance of the concrete compressive strut can be calculated if the allowed concrete
compressive stress is known. This stress is called σcd. The shear force resistance of the
concrete struts is denoted as VRd,max. From eq. (8.10) it follows that (EN 1992-1-1 eq. (6.14)):
cot   cot 
VRd,max  bw z (cot   cot  ) sin 2   cd  bw z  cd
1  cot 2 
April 2018 8-6

Chapter 8 - Shear
The compressive struts cannot be loaded up to the uni-axial concrete compressive strength
because the stirrups that cross the concrete diagonal struts in the web are loaded in tension.
The tensile forces perpendicular to the struts’ direction are transferred by bond. Their impact
is that strut strength is reduced, see fig. 8.5a. Figure 8.5b shows a failure envelope of
concrete, loaded in two directions. In the struts a combination of a tensile and a compressive
stress occurs. The dashed line from fig. 8.5b shows that the stress combination results in a
reduction of the ultimate compressive stress relative to the cylinder compressive strength fck.
Furthermore, from experiments it turns out that the maximum stress does not proportionally
increase with the concrete strength class. The following expression is derived (EN 1992-1-1
cl. 6.2.3 (3)):
 cd   cw  1 f cd (8.11)
where:
cw is a coefficient taking into account the state of stress in the compression strut;
1 is a strength reduction factor for concrete cracked in shear.
If the design stress of the shear reinforcement is below 80% of the characteristic yield stress
fyk (EN 1992-1-1 eq. (6.10aN) & (6.10bN)):
1 = 0,6 for fck < 60 N/mm2

f ck
 1  0,9   0,5 for fck > 60 N/mm2
200
where fck is the characteristic cylinder compressive strength.
Otherwise (EN 1992-1-1 eq. (6.6N)):

 f 
 1  0, 6 1  ck 
 250 
tension
b
fct fc
compression tension
f 'c c
fc
f 'c
fc
compression
Fig. 8.5 Reduction of the concrete compressive strength due to transverse tension from
bond forces introduced by the shear reinforcement
April 2018 8-7

Chapter 8 - Shear
An axial compressive stress in the cross-section introduced by loading or prestressing

influences the resistance of the concrete struts loaded in compression. A stress below 0,6 fcd
has a positive impact, whereas a higher value has a negative impact. EN 1992-1-1 cl.
6.2.3 (3) gives the following expressions:
 cw  1, 0 for non-prestressed structures

 cp
 cw  1, 0  for 0 < σcp < 0,25 fcd
f cd
 cw  1, 25 for 0,25 fcd < σcp < 0,50 fcd
  cp 
 cw  2,5  1, 0   for 0,50 fcd < σcp < 1,0 fcd
 f cd 
where σcp is the mean compressive stress (compression is positive) in the concrete. Figure 8.6
shows cw as a function of σcp.
Fig. 8.6 Coefficient for the concrete strut compressive strength as a function of the axial
compressive stress
The designer is free to choose an angle  between 21,8º and 45,0º. A small angle has a
positive effect on the shear resistance of the shear reinforcement (eq. (8.7)): The smaller the
angle, the more reinforcement is crossed by the crack. As a result, the shear resistance
increases. However, the smaller the angle, the higher the compressive stress in the inclined
concrete struts (eq. (8.10) and, as a result, the lower the shear resistance. Designers often
assume  = 21,8º and increase this value only if the concrete strut compressive stress is
governing.
The shift rule
As a result of the shear resistance mechanism with its inclined crack, the force in the
longitudinal reinforcement is larger than follows from a cross-sectional equilibrium analysis.
This is illustrated in fig. 8.7. Note that z is the distance from the tensile reinforcement to the
April 2018 8-8

Chapter 8 - Shear
concrete compression zone, which is often assumed to be almost equal to the distance to the
resulting concrete compressive force (see fig. 8.2, 8.4 and EN 1992-1-1 fig. 6.5).
al
I
VEdI M Ed
Aswfywd
z
Fig. 8.7 Equilibrium analysis to explain the bending moment shift rule
It is assumed that the “concrete” part VRd,c of the shear force capacity can be totally attributed
to the shear capacity of the uncracked compression zone. In case all stirrups yield, the
resulting force VRd,s is at a distance ½ z cot from section I-I. The resultant R of the forces
VRd,s and VRd,c is through point S. The distance al from point S to section I-I follows from:
al VRd  12 z cot  VRd,s  z cot  VRd,c
where VRd = VRd,c + VRd,s.
 V V   VRd,c 
al  z cot   12 Rd,s  Rd,c   12 z cot  1   (8.13)
 VRd VRd   VRd 
From the bending moment equilibrium around point S, the force NsI in the steel is found.
Here one finds:
I
M Ed VI a
N sI   Ed l (8.14)
z z
where al is the horizontal distance from section I-I to the point S.
April 2018 8-9

Chapter 8 - Shear
Equation (8.14) is written as:

II
M Ed
N sI 
z
So, the steel force NsI in section I-I is obtained from the bending moment in section II-II, by
dividing it by the internal lever arm.
In practice this is solved by shifting the moment line over a distance (bending moment line
shift rule):
 V 
al  1 2 z cot  1  Rd,c  (8.15)
 VRd 
Fig. 8.8 Shift (B) of original bending moment curve (A) according to EN 1992-1-1 fig.
9.2
Here, two extremes can be distinguished:

 If the concrete part VRd,c = 0, as assumed in EN 1992-1-1, the shear force should be
totally carried by the shear reinforcement. Equation (8.15) then reads al = ½ z cot . EN
1992-1-1 also takes into account the angle of the stirrups, which results in:
al = ½ z (cot  - cot ).
 If the compression zone would carry the total shear force (as is the case in members
without shear reinforcement) then al = z cot . For  = 45º and z ≈ d it follows that al = d
(EN 1992-1-1 cl. 9.2.1.3 (2)).
To summarise, EN 1992-1-1 presents (EN 1992-1-1 cl. 9.2.1.3):

 Shifting the bending moment line over al = d for members without shear reinforcement.
April 2018 8-10

Chapter 8 - Shear
 Shifting the bending moment line over al = ½ z (cot  - cot ) for members with shear
reinforcement. Alternative: shift over al = d.
 Instead of shifting: add an additional tensile force to the longitudinal reinforcement force
(EN 1992-1-1 cl. 6.2.3 (7)).
8.3 Prestressed elements without shear reinforcement, serviceability

limit state
In fully or limited prestressed concrete, crack formation caused by flexure and/or shear is not
permitted in the serviceability limit state. Figure 8.9 shows a prestressed beam in which the
stresses caused by flexure and shear are determined in an arbitrarily chosen cross-section I.
The stress σx in longitudinal direction caused by bending, in a point at distance y from the
centroidal axis:
M y Pm Pm ep y
x    (8.16)
Ic Ac Ic
c.a.
Fig. 8.9 Prestressed beam under bending and shear loading.
The shear stress in this point is:

V Sy
 xy  (8.17)
I c by
where Sy is the static moment for the considered height y and by is the corresponding width at
that level. With Mohr’s circle and relations (8.16) and (8.17), the principal stresses and their
directions can be determined (see also fig. 8.10). It holds:
x  x2
I     xy
2
(8.18)
2 4
x  x2
 II     xy
2
(8.19)
2 4
 xy
tan  2 I   (8.20)
1
2x
April 2018 8-11

Chapter 8 - Shear
where σx (a compressive stress) has a negative sign.
compression tension
Fig. 8.10 Determination of the principal stresses in the serviceability limit state.
If the principal tensile stress appears to be too high, the stress can be reduced by increasing
the thickness of the web or by prestressing in vertical direction as shown in fig. 8.11.
vertical prestressing
compr. tension compr. tension
with vertical prestressing without vertical prestressing
Fig. 8.11 Influence on the principal stresses of a vertical prestress in the web
April 2018 8-12

Chapter 8 - Shear
With Mohr’s circle, it can be shown that the principal tensile stress can be compensated for
completely by applying prestressing in vertical direction (fig. 8.11, σI < 0). For this case, the
principal stresses are:
  y 
2
x y x
I     xy
2
(8.21)
2 4
  y 
2
x y x
 II     xy
2
(8.22)
2 4
The direction follows from:

 xy
tan  2 I   (8.23)
1
2  x  y 
in which σx and σy have a negative sign in case of compression.
If cracks in SLS are allowed, their width should be well controlled. The stress in the stirrups
at SLS can be determined using the truss analogy (eq. (8.7) with s instead of fywd). With the
tensile member model, which will be discussed in Chapter 9, the crack width can be
calculated.
Example
Figure 8.12 shows the beam from section 4.7. Stresses in the cross-section which is 2 m from
a beam end are analysed. Beam length l = 21,5 m.
At midspan, the centroidal axis of the four anchors is 125 mm from the bottom fibre. THe
tendon profile is a parabola. The anchors are at neutral axis level.
The cross-section is now modelled using three rectangles, see fig. 8.12.
Characteristics of the concrete cross-section:

Ac = 328·103 mm2
zcb = 548 mm
zct = 452 mm
Ic = 39,5·109 mm4
Wcb = 72,1·106 mm3
Wct = 87,2·106 mm3
Tendon drape at midspan: f = 548 – 125 = 423 mm.
At x = 2 m from the support, the tendon is at 0,5·x (l – x) f / (0,125l2) = 0,337f = 143 mm

from the neutral axis.
April 2018 8-13

Chapter 8 - Shear
Radius of curvature: R = l2 / (8f) = 21,52 / (8·0,423) = 136,6 m
qq = 12 kN/m
qg = 2,5 kN/m
qselfweight
zct
zcb
Fig. 8.12 Example of the calculation of the principal stresses in SLS (dimensions mm,
stresses N/mm2).
Loads in SLS:
Concrete selfweight plus additional static load: qG = 8,2 + 2,5 = 10,7 kN/m
75% of the variable load (“representative load”) qQ,rep = ψ qQ = 0,75·12,0 = 9,0 kN/m
Working prestressing force P∞ = 2187 kN qP = -2187/136,6 = -16,0 kN/m
Total uniformly distributed SLS load: 19,7 – 16,0 = 3,7 kN/m
Bending moment and shear force from qG and qQ,rep at x = 2 m:
ME,G+Qrep = (10,75·19,7·x) – (19,7·x·x/2) = 384 kNm

VE,G+Qrep = (10,75 – x)·19,7 = 172 kN
Bending moment and shear force from qP at x = 2 m:
ME,P = (10,75·-16,0·x) – (-16,0·x·x/2) = -312 kNm

VE,P = (10,75 – x)·-16,0 = 140 kN
Resulting bending moment an shear force at x = 2 m:
ME = 384 - 312 = 72 kNm
April 2018 8-14

Chapter 8 - Shear
VE = 172 – 140 =·32 kN
Concrete compressive stress from the axial prestressing force:
 2187 10 3
c  3
 6,67 N/mm 2
328 10
Concrete longitudinal stresses from bending:
 2187 103 72 10 6

 c, top  3
 6
 7,49 N/mm 2
328 10 87,2 10
 2187 103 72 10 6
 c, bottom  3
 6
 5,67 N/mm 2
328 10 72,110
Concrete shear stress at neutral axis level:
S c  800 150  452  75  160  302 151  52,6 10 6 mm 3

VE S c 32 103  52,6 10 6
 c, xy   9
 0,27 N/mm 2
b Ic 160  39,5 10
The following figures contain the bending stress and shear stress over beam height. The
principal stress that follow from these two stresses are also included, as well as the angle θ
between the longitudinal axis of the beam and the principal stress direction. The shear stress
is relatively small compared to the shear stress; the minimum principal stress (compression)
is almost equal to the bending stress and the angle θ is almost zero.
Bending stress (including prestressing)

1000
-7.4862
900
800
700
600
z [mm]
500
400
300
200
100
0
-5.6771
-7.6 -7.4 -7.2 -7 -6.8 -6.6 -6.4 -6.2 -6 -5.8 -5.6
2
x [N/mm ]
April 2018 8-15

Chapter 8 - Shear
8.4 Prestressed elements loaded in shear, ultimate limit state
8.4.1 Elements without shear reinforcement
In section 8.2.1 it is demonstrated that the shear capacity of an element without shear
reinforcement depends on several influencing factors (eqs. (8.1) and (8.2)). The shear
capacity also depends on the development of (bending) tensile cracks since they may connect
and develop into shear cracks.
The development of tensile cracks in the element is postponed by applying prestressing. In

the case shown in fig. 8.13, the tensile zone is prestressed, which introduces an additional
compressive stress in the outer tensile fibre:
Pm Pm ep
 cb   
Ac Wcb
This stress is reduced to zero by applying a bending moment (fig. 8.13b):
April 2018 8-16

Chapter 8 - Shear
W 
M 0   cb Wcb  Pm  cb  ep   Pm  1
6 h  ep 
 Ac 
The moment M0 is denoted as the compensating moment.

The shear force at which the stress σc at the bottom side of the beam becomes zero, is:
M0
Vn  (8.24)
a
Fig. 8.13 Influence of the prestressing force on the shear capacity of a structural element
without shear reinforcement
When loading is continued from the situation shown in fig. 8.13b, the same situation is
obtained as when starting to load a non-prestressed beam. This implies that the shear capacity
of the prestressed beam is increased by Vn = M0 / a relative to a reinforced beam.
For the distance a it can also be written:
M  Md 
a  
V  Vd 
where M and V are the moment and shear force at the position of the load, respectively. Thus,
equation (8.24) can be rewritten as:
M0
Vn  Vd (8.25)
Md
which is an expression that is generally valid.
April 2018 8-17

Chapter 8 - Shear
For general use, EN 1992-1-1 uses eq. (8.24) as a starting point. Assuming that the beam has
a rectangular cross-section with d = 0,85 h and ep = 0,35 h (as often used in experiments), the
following results are obtained.
The concrete stress at the bottom fibre from prestressing:
Pm Pm  0,35h  0,5h Pm
 cb    3
 3,1
bh 12 bh
1
bh
The concrete stress is zero in case of an additional bending moment:
M   cb 16 bh 2  0,52 h Pm
This bending moment must be introduced by the support reaction Vn at a distance a from the
cross-section considered:
M  0,52 h Pm  Vn a
From which it follows:
h Pm d Pm P
Vn  0,52  0,52  0, 61 m
a 0,85 a a/d
Many tests have been carried out for a ratio a / d between 2,5 and 4,0. The following results
apply:
a / d = 2,5 : Vn = 0,24 Pm
a / d = 4,0 : Vn = 0,15 Pm
Research indicated that 0,15 Pm is a conservative lower bound value. This value is used in
EN 1992-1-1; see the coefficient k1 in eq. (8.2). This increase of the shear resistance can only
be applied in case the member has not to be provided with shear reinforcement.
A second aspect is that the prestressing results in a part of the pre-compressed tensile zone
remaining uncracked, even in the ultimate limit state (fig. 8.14). In these parts, flexural
tensile stresses are compensated for by the prestressing (area a in fig. 8.14). Flexural cracks
will occur at bottom side parts of the beam where the flexural tensile stress exceeds the
tensile strength of the concrete (area b in fig. 8.14).
April 2018 8-18

Chapter 8 - Shear
area a area b area a
due to prestressing
tensile strength concrete

due to bending
Fig. 8.14 Crack pattern in a prestressed beam, and the stresses in the bottom part
In area a, where the bottom part remains uncracked, even in the ultimate limit state, shear
cracks may still develop. These cracks don’t start at the bottom fibre but origin in the web at
the position where the principal tensile stress in the concrete reaches the concrete tensile
strength. From eq. (8.18) it follows that this occurs in case:
x  x2
I     xy
2
 f ctd
2 4
where:
VEd Sc
 xy 
bw I c
It can be found that an inclined crack occurs in case (EN 1992-1-1 eq. (6.4)):
bw I
VRd,c  f ctd2   x f ctd (8.26)
S
In case of compression σx has a negative sign.
Note that eq. (8.26) contains no αl since it is assumed that the full prestressing force is
transferred at the beam end. The influence of the transmission length is discussed later.
When no shear reinforcement is applied in area a, the development of an inclined crack under
the load from eq. (8.26), will generally lead to failure. This type of failure is called tensile
splitting shear failure. EN 1992-1-1 cl. 6.2.2 (2) states that it is only allowed to use eq. (8.26)
at locations where the tensile bending stress in the outer fibre of the cross-section is less than
fctk,0,05 / γc.
Example: Calculation of the shear capacity in area a

A special application of the previous theory is the analysis of the behaviour of a prestressed
hollow-core slab.
April 2018 8-19

Chapter 8 - Shear
The cross-section of the slab considered is shown in fig. 8.15.
40
185
265
40
40 185 45 185 45 185 45 185 45 185 40
1195
Fig. 8.15 Cross-section of the hollow-core slab considered (dimensions mm)
The slab is prestressed with 14 Y1860S7 strands of 9,3 mm (3/8''; 52 mm2). To avoid
splitting of the concrete, the two outer strands are stressed to half their capacity, which
implies that, with regard to prestressing, effectively 13 strands are present. The total effective
steel area is Ap = 13 · 52 = 676 mm2. The following calculations are based on 13 effective
strands, not taking into account that 14 strands are present in the cross-section.
Material properties:
concrete strength class C45/55
f ctd  1,8 N/mm 2 ,
I c  1,55 109 mm 4 ,
Ac  1, 78 105 mm 2 ,
ecb  132 mm ; ect  133 mm
Wcb  11, 7 106 mm3 (concrete bottom fibre),
Sc  7, 66 106 mm3 ,
bw  260 mm (governing cross-section at the level of maximum hollow core width),
d  d p  225 mm ,
 p,  1100 N/mm 2 .
The working prestressing force Pm,  676 1100  744 103 N . The bending moment
introduced by prestressing is constant over the slab length:
M p,   Pm, ep  744 103  132  40  =  68,4 106 Nmm

Pm, 744 103
Mean concrete compressive stress from prestressing:  cpm,   = 4,18 N/mm 2
Ac 178 103
The strands are anchored by bond. A transmission length lpt2 = 500 mm is assumed, see fig.
8.16 (EN 1992-1-1 cl. 8.10.2.2).
A maximum strut angle of 45º is assumed. The first crack that causes failure (dashed line in
fig. 8.16b), then develops at a distance 70 + h / 2 = 202 mm from the end (70 mm support
length assumed) at a height h / 2 = 132 mm. A this position αl = 202 / 500 = 0,404 (EN 1992-
1-1 eq. (6.4)) which implies that about 40% of the prestressing steel force is transferred to the
April 2018 8-20

Chapter 8 - Shear
concrete at the location considered. The steel stress is 0,404 · 1100 = 445 N/mm2. The shear
capacity of the concrete (tensile splitting shear failure) is (eq. (8.26)):
bw I 1,55 109  260

VRd,c  f ctd2   l  cp f ctd  1,802  0, 404  4,18 1,80  132 103 N
S 7, 66 10 6
Note that the concrete is assumed to crack at the design value of its tensile strength, fctd =
fctk,0,05/γc (EN 1992-1-1 cl. 6.2.2 (2)).
Figure 8.16a shows this 132 kN starting point of the tensile splitting shear resistance curve.
The curve shows that this failure mechanism first shows a gradual increase of the resistance.
This comes from the gradual introduction of the prestressing force in the slab. The maximum
resistance is reached at lpt2 = 500 mm from the end of the slab. At that position, the full
prestressing force is in the slab which implies that σcp = 4,18 N/mm2 (compression) and
VRd,c,max = 173 kN.
It is now assumed that the prestressing steel at midspan reaches a stress pu = fpd =
1522 N/mm2 in ULS (at the kink in the stress-strain diagram, which is conservative). With Ap
= 676 mm2, p, = 1100 N/mm2 and an assumed internal lever arm z  200 mm, the bending
moment resistance is (see fig. 8.17):
MRd = 676 · (1522 – 1100) · 200 + 676 · 1100 · (132 – 40) = (57,1 + 68,4) · 106
MRd = 135,5 · 106 Nmm
If the slab is simply supported with a span of 10,0 m and subjected to a uniformly distributed
load, the design value of the load that can be resisted in bending is:
1 2 8  (135,5  68, 4) 106
M Ed  M Rd  qd l  M p,  135,5 10  qd 
6
 16,3 N/mm  16,3 kN/m
8 10, 02
At the cross-section where the prestressing force is fully transferred to the concrete, the
concrete stress in the bottom fibre is:
Pm M p P Pm ep
 cb      m   4, 2  5,9  10,1 N/mm 2
Ac Wcb Ac Wcb
This stress is reduced to zero (fig. 8.13b) by applying a bending moment:
W 
M o   cb Wcb  Pm  cb  ep   Pm  1
6 h  ep 
 Ac 
April 2018 8-21

Chapter 8 - Shear
VRd,c,max
governing width
Fig. 8.16 Calculation of the tensile splitting shear capacity in area a:

a. graph of ultimate shear force according to equation (8.26) [kN];
b. determination of the critical position;
c. stress in prestressing steel.
Since the concrete is assumed to crack when the design value of its tensile strength fctd =
fctk,0,05/γc is reached (EN 1992-1-1 cl. 6.2.2 (2)), the design value of the flexural cracking
bending moment is:
P 
M cr  Wcb  m  f ctd   11, 7 106 (4, 2  1,80)  70 106 Nmm  70 kNm
 Ac 
When the bending moment resistance MRd = 135,5 kNm (at a design load qd = 16,3 kN/m and
Mp, = -68,4 kNm) is reached, the length of area a, assuming a parabolic distribution of the
bending moment, follows from:
April 2018 8-22

Chapter 8 - Shear
1
2 x (10  x) 16,3  68, 4  70 kNm 
81,5  81,52  4  8,15 138, 4
x  5, 00  2,83  2,17 m or 7,83 m
2  8,15
This result indicates that the slab is uncracked up to a distance of 2,17 m from the supports.
The remaining part (10,0 – 2,17 – 2,17 = 5,66 m in length) is cracked in ULS and, as a result,
will demonstrate flexural-shear failure, whereas the two uncracked parts exhibit tensile
splitting shear failure.
The shear capacity VRd,c of the cracked area b is now calculated, using EN 1992-1-1 cl.
6.2.2 (1). From equation (8.2) (EN 1992-1-1 eq. (6.2a) & (6.2b)) it follows that:
VRd,c  CRd,c k 100 l f ck   k1  cp  bw d

1/ 3
 
VRd,c  0,12 1,94  100  0, 012  45   0,15  4,18  260  225  88, 2 103 N
 1/ 3
 
where
200 200
k  1  2, 0  k  1   1,94
d 265  40
A 676
l  sl   0, 012
bw d 260  (265  40)
The shear resistance has a minimum value:
VRd,c  vmin  k1  cp  bw d
VRd,c  0, 035 1,943/ 2  451/ 2  0,15  4,18  260  225  73,8 103 N
The shear resistance VRd,c = 88,2 kN from flexural-shear failure is shown in fig. 8.17,
together with the shear resistance that follows from tensile splitting shear failure. The figure
also shows the linear shear force curve that holds for uniform loading.
Shear failure occurs if the design shear force curve crosses one of the shear resistance curves.
The design shear force is zero at midspan. An increase of the uniform load on the slab makes
that the straight VEd-line rotates around the point where VEd = 0 at midspan. The governing
load and shear force are found when the rotating line first crosses the resistance lines VRd,c. It
then follows that tensile splitting shear failure occurs first. Failure is at 0,2 m from the end
face of the slab. It then holds:
1  132
qd  l  0, 2   132 kN  qd   27,5 kN/m
2  5, 0  0, 2
Since the slab width is 1,2 m, it also holds:
April 2018 8-23

Chapter 8 - Shear
27,5
qd   22,9 kN/m 2
1, 2
At a selfweight of:
Ac  25 0,178  25
qg    3, 7 kN/m 2
1, 2 1, 2
it can be written:
 q qq   g qg  22,9  1,5 qq  1, 2  3, 7  22,9
So that for the maximum variable load it follows:
qq  12,3 kN/m 2
Mcr = 70 kNm
2,17 m MEd = 137 kNm 2,17 m

uncracked concrete cracked concrete
failure
88,2
shear force [kN]
Fig. 8.17 Determination load-bearing capacity of a prestressed hollow-core slab
Bending moment failure occurs at a design value of the load qd = 16,3 kN/m (MEd =
135,5 kNm). Shear failure, namely tensile splitting shear failure, occurs at qd = 27,5 kN/m.
These results demonstrate that the slab fails in bending, not in shear.
Fig. 8.18 shows a typical tensile splitting shear fracture in a laboratory test. Note that the line
load is close to the line support to have a relatively high shear force and a small bending
moment.
April 2018 8-24

Chapter 8 - Shear
Fig. 8.18 Tensile splitting shear fracture in a beam that is prestressed by pre-tensioned steel
8.4.2 Structures with shear reinforcement, prestressed with straight prestressing

tendons
In the case of structures prestressed with straight tendons, the effect of prestressing is as
follows:
In area b, cracked in bending, the calculation of the required shear reinforcement is almost
the same as for reinforced concrete. The only difference is that in eq. (8.2) the axial
compressive stress σcp results in an increase of shear resistance.
In area a, uncracked in bending, the uncracked lower flange has a positive effect on shear
resistance. This can be explained by cutting a section along a shear crack (fig. 8.19).
a b
Fig. 8.19 Transfer of a shear force in the areas a and b
In area a both the uncracked compression zone and the uncracked tensile zone contribute to
shear resistance. Because the cracks do not proceed to the outer fibres of the beam, they
hardly open. Therefore, the crack width is small and the shear reinforcement is only lightly
stressed. Figure 8.20 shows the results of a shear test on a prestressed beam, carried out by
Leonhardt, Koch and Rostásy [8.2].
April 2018 8-25

Chapter 8 - Shear
Fig. 8.20 Stresses in the shear reinforcement of a prestressed beam in the areas uncracked
in bending (a) and cracked in bending (b) [8.2]
Because of the constant shear force in the areas between the load and the supports, the beam
was provided with a constant shear reinforcement. Measuring the strains in the stirrups
demonstrated that in area a the stresses were only about 1/3 of those in area b.
If the shear force is higher than the force that causes tensile splitting shear failure, which
implies that shear reinforcement has to be applied, the contribution of the "concrete" (which
is the component between Mörsch's line and the stirrup stress measured, see fig. 8.3) is
somewhat higher than in case of shear bending failure. EN 1992-1-1 could take this positive
effect into account by allowing for a larger rotation of the compressive diagonal concrete
struts. This is not accounted for in the code: it is prescribed that the calculation of the
required amount of shear reinforcement should be carried out following the same procedure
as for area b.
With regard to the resistance of the concrete compressive struts other considerations apply.
An axial compressive force has basically two effects. On one hand it postpones cracking and
crack propagation in the web of a beam. On the other hand, the axial load generates inclined
compressive stresses in the struts as well, so that a part of their capacity is consumed. An
evaluation of tests showed, that a small normal force improves the bearing capacity of the
struts, but a large force reduces it, see fig. 8.6 [8.9].
8.4.3 Structures with shear reinforcement, prestressed with draped tendons
Figure 8.21a again shows the total load exerted on a beam prestressed using draped tendons.
The beam is analysed at ULS. Apart from the load components qg (permanent load;
selfweight and static loads) with load factor g = 1,2 and qq (variable load) with a load factor
q = 1,5, a prestressing load acts on the structure, consisting of:
April 2018 8-26

Chapter 8 - Shear
 the equivalent prestressing load qp  8 Pm, f / l 2 ;

 a downward directed force Pm,∞ sin  at the anchorages. Because the equivalent
prestressing load is in equilibrium with the anchorage forces Pm,∞ sin  = ½ qpl;
 an axial prestressing force Pm,∞ (Pm,∞ cosα at an anchor and an additional force form the
horizontal component of the curvature pressure, namely Pm,∞ (1 - cosα), over ½ l);
 in the case of an eccentric position of the anchorages, a bending moment Pm,∞ epo at the
beam ends.
loads
shear force line
“passive” steel
Ap  pu   p, 
Fig. 8.21 a. Loads on a beam with draped prestressing tendons.

b. Shear force diagram.
c. Loads on the “rest-system”.
Since the prestressing load does not change when the structure is loaded to failure, the load
factor p = 1,0.
The stress in the prestressing steel is p,∞ which implies that the part pu - p,∞ is "not used"
yet. This part can be regarded as being "reinforcing steel" (passive prestressing steel) present
in the beam and will from now on be denoted as the ”rest-system”. This rest-system can be
provided with additional reinforcing steel to have the required resistance in bending and
shear. The rest-system is dealt with as if it was a normal reinforced concrete beam. It should
be checked whether it has sufficient resistance.
April 2018 8-27

Chapter 8 - Shear
In the beam shown in fig. 8.22, in section I, the passive part of the prestressing steel and the
active force Pm,∞ introduced at the ends, provide a bending resistance:
M Rd,I  zpI AP   pu   p,   z0 Pm,
If this were not sufficient, reinforcing steel could be added, so that the bending moment
resistance is increased to:
M Rd,I  zpI AP   pu   p,   zsI As f yd  z0 Pm,
where zpI and zsI are the internal lever arms of the prestressing steel and the reinforcing steel
respectively, and where it is assumed that the reinforcement reaches the design yield stress
fyd. If this might not be the case, the more general variable fs needs to be used.
al
MII MI
MRd II
Asfyd
MRd,I
Asfyd
Fig. 8.22 Design for bending
April 2018 8-28

Chapter 8 - Shear
In cross-section II, which is in the area subjected to shear, the shift rule should be applied. To
be able to design the reinforcement in cross-section II, at first the bending moment curve is
shifted over a distance al, according to eq. (8.13).
The bending moment resistance around the concrete compressive force in section II is:
M Rd,II  zpII AP   pu   p,   z0 Pm,
Also here the bending moment resistance can be increased to the required level by providing
reinforcing steel at the bottom of the beam, so that:
M Rd,II  zpII AP   pu   p,   zsII As f yd  z0 Pm,
This shows that it is always favourable to have tensile reinforcement at the bottom of the
beam. If the beam is prestressed with more than one prestressing tendon, it is advised to have
at least one tendon along the bottom side of the beam, up to the beam end. It is advised to, if
possible, add reinforcing steel since this is not only favourable with regard to bending
moment resistance, but also for shear resistance and crack width control.
Fig. 8.23 Determination of shear resistance ( is the inclination of the shear crack)
Subsequently, shear resistance is regarded. Also here, like in the case of reinforced concrete,
an inclination of the struts between 21,8º and 45º can be chosen. The shear force to be carried
by the section considered, see fig. 8.23, is:
VEd  qd  12 l  x 
where qd   g qg   q qq  qp
April 2018 8-29

Chapter 8 - Shear
In case the shear resistance is assumed to be the design value of the shear force, vertical
equilibrium results in the following requirement:
qd  12 l  x   max VRd,c ,VRd,s  (8.27)
where:
VRd,c  CRd,c k 100 l f ck   k1  cp  bw d

1/ 3
(EN 1992-1-1 eq. (6.2.a))
 
Asw
VRd,s  z f ywd  cot   cot   sin  (EN 1992-1-1 eq. (6.13))
s
z is the internal lever arm between the force in the compression zone and the centre of
gravity of the longitudinal steel.
d is the effective depth of the cross-section calculated from the centre of gravity of the
longitudinal steel
α is the angle of the shear reinforcement relative to the horizontal axis.
Also here it is demonstrated that providing a tensile tie at the bottom of the beam is
favourable, since both z and d increase, so the shear resistance increases, even if the same
amount of shear reinforcement is applied.
For statically indeterminate structures, the approach is the same. Figure 8.24 shows the loads
for a statically indeterminate beam. Also here the design load is composed of a number of
contributions:
qd   g qg   q qq  qp
In the area close to the intermediate support, the equivalent prestressing load has a downward
direction, so that the load is:
qd   g qg   q qq  qp
In the shear force diagram this effect is directly recognised (a kink close to the intermediate
support).
April 2018 8-30

Chapter 8 - Shear
bending moment
shear force
axial force
detailing “passive” part of Ap
Fig. 8.24 Design for shear with the equivalent prestressing load method
8.4.4 Influence of ducts in the cross-section
Special attention has to be given to the control of the strength of the concrete compression
struts. The ducts for the prestressing tendons create a discontinuity that should be regarded
when checking the stress in the struts (fig. 8.25).
For non-grouted ducts (as well as for grouted plastic ducts and unbonded tendons), the net
width of the compression struts, available for resisting the inclined compressive forces, is
(EN 1992-1-1 eq. (6.17)):
b  bw  1, 2 Ø (8.28)
April 2018 8-31

Chapter 8 - Shear
where Ø is the diameter of the prestressing ducts. The reduction is 1,2 Ø, which is more
than Ø, since it can happen that not the compressive strength of the concrete, but the
splitting tensile strength of the concrete (because of transverse tension) is governing (fig.
8.26) (EN 1992-1-1 cl. 6.2.3 (6)).
duct
Fig. 8.25 Strength reduction of compression struts by prestressing ducts
Fig. 8.26 Splitting tensile effect in the case of large ducts
Especially in the construction stage this restriction is important, because it happens

frequently that a part of the load is already applied before the tendons are grouted.
For grouted ducts the resistance reduction is smaller, but also here some strength loss has to
be taken into account, because of the relatively low stiffness of the injecting grout. In this
case the expression is (EN 1992-1-1 eq. (6.16)):
b  bw  12 Ø (8.29)
In case the widths of metal grouted ducts Ø < bw / 8, it can be assumed that bw,nom = bw.
It is noted that the influence of the ducts also has to be accounted for when checking tensile
splitting shear failure.
Figure 8.27 shows the crushing of the concrete compression struts due to strength loss in the
web caused by a duct. The figure shows the left span of a two span statically indeterminate
beam. The tendon profile consists of straight parts and curved parts (at midspan A and
April 2018 8-32

Chapter 8 - Shear
intermediate support B). The beam is loaded by point loads, one at each span. The crack
pattern at failure is shown. Concrete crushing in the beam close to the intermediate support is
indicated. For an extended numerical analysis of the strength of the concrete struts, reference
is made to [8.8].
tendon profile
stirrups
section A (midpan) section B (intermediate support)
web crushing area intermediate support
Fig. 8.27 Shear test on a beam with inclined prestressing tendon [8.6]:
a. Side view (dimensions mm).
b. Cross-sections (dimensions mm).
c. Failure of concrete struts.
In EN 1992-1-1 the control of the strength of the compression struts is in the check of VRd,max,
see eq. (8.12) (EN 1992-1-1 eq. (6.9)). The impact of the ducts is accounted for by
introducing bw,nom according to eq. (8.28) or (8.29).
As a result of the equivalent prestressing load, a substantial part of the load can be
“balanced”. Figue 8.28 shows the inclined tendons of a box girder bridge close to an
intermediate support.
April 2018 8-33

Chapter 8 - Shear
Fig. 8.28 Inclined prestressing tendons near an intermediate support of a box girder bridge
8.5 Vertical prestressing of the web

Already in fig. 8.11 it was shown that webs highly loaded in shear can be provided with
vertical prestressing to prevent the formation of shear cracks at SLS and to act as shear
reinforcement at ULS.
Applying vertical prestressing has the following advantages:

 Reduction of the principal tensile stresses, and consequently, the possibility of reducing
the web thickness if it is required that the web is uncracked at SLS.
 Reduction of shear reinforcement and less labour-intensive installation.
 Reduction of cost of the reinforcement in high beams
The last two advantages mentioned are a result of the much higher yield stress (or 0,1%
strain limit) of the prestressing steel than the yield stress of reinforcing steel, so that the
cross-sectional area of the shear reinforcement can be much smaller. Applying vertical
prestressing is only economic in the case of high webs because of the high costs of the
anchorages.
An example of the application of vertical prestressing is the Rhine Bridge near to Bensdorf
(Germany, 1965), with a span of 208 m (fig. 8.29). The depth of the concrete cross-section
April 2018 8-34

Chapter 8 - Shear
near to the supports is 10,45 m. In the direction of midspan the depth of the cross-section is
reduced such that the nominal shear stress is approximately constant over a large distance.
Therefore, also the width of the web could remain constant at 300 mm over almost the entire
bridge. Only close to the supports the web width had to be increased to 370 mm. The webs
are prestressed with bars Ø 32 mm, steel type Y1030H, inclined at an angle of 45º [8.5].
longitudinal section
hinge
prestressing in box girder web in the main span
prestressing in the web

cross-section A-A Ø 32 mm Y1030H
cross-section at the pier
Fig. 8.29 Example of a structure with prestressed shear reinforcement: the Rhine bridge
near to Bensdorf in Germany (dimensions m) [8.5]
With regard to the shear resistance of girders with vertical prestressing, the following two
questions can be raised:
1) From eq. (8.23) it turns out that, due to vertical prestressing, the inclination of the shear
cracks increases, so that less shear reinforcement is intersected by a crack. Should this be
considered in design?
April 2018 8-35

Chapter 8 - Shear
2) The difference between the design strength (ULS) and the working stress (SLS) of the
prestressing can be substantial (order of magnitude 500-600 N/mm2). Fully utilising this
reserve capacity requires a considerable strain in the prestressing steel. A large strain
implies a large deformation of the truss. Does the corresponding rotation of the struts
introduce such high secondary compressive stresses in the concrete, that strut resistance
is reduced?
To answer these questions, Kupfer and Ruhnau [8.5] tested a beam that was at one side
prestressed with inclined and at the other side with vertical prestressing tendons, see fig.
8.30. In both parts of the beam additional shear reinforcement was applied. The
reinforcement in the webs was designed such that the prestressed shear reinforcement carries
70% of the shear force; the passive reinforcement 30%. Both parts of the beam are designed
to have the same shear force resistance.
2 stirrups Ø14 2 stirrups Ø14
stirrup Ø10 in the flanges

stirrup Ø8
Fig. 8.30 Test on a beam with two types of vertical prestressing (dimensions mm) [8.5]
In fig. 8.31 the stresses measured in the prestressing steel and the reinforcing steel are
presented.
April 2018 8-36

Chapter 8 - Shear
prestressing bar n° 2 4 6 8 10
1 3 5 7 9 11
stirrup n° 1 3 5 7 9 11
2 4 6 8 10 12
[N/mm2] 1500
load step 10
1400 V = 822 kN
load step 9
1300
V = 740 kN
1200 load step 8
V = 658 kN
1100
1000
load step 1
(prestress)
900
prestressing bar n° 1 3 5 6 8 10
2 4 7 9
stress in the vertical direction prestressing bars
[N/mm2] 500
load step 10
400 V = 822 kN
300 load step 9

V = 740 kN
200
load step 8
100 V = 658 kN
0
1 3 5 7 9 11
stirrup n° 2 4 6 8 10 12
stress in the stirrups
Fig. 8.31 Stresses in the prestressing and reinforcing steel measured at different load levels
[8.5]
The stress in the inclined prestressing tendons is higher than in the vertical tendons. This can
be explained by the truss analogy.
It was concluded that the tendon stress reserve of 580 N/mm2 between SLS and ULS can be
fully utilised. On the basis of this investigation, it was proposed to put a limit of 600 N/mm2
to the prestressing steel stress reserve that can be activated. The design failure strength of the
prestressing steel then is:
 pu   p,  600 N/mm 2  f pd
Furthermore, it is advised not to use a strut inclination smaller than  = 450.
April 2018 8-37

Chapter 8 - Shear
8.6 Literature
8.1 CEB-FIP Model Code for Structural Concrete, 1990.
8.2 Leonhardt, F., Koch, R., Rostasy, F.,”Schubversuche an Spannbetonträgern”,

Deutscher Ausschuss für Stahlbeton, Heft 227, Berlin, 1973.
8.3 Walraven, J.C., Mercx, W.P., “The bearing capacity of prestressed hollow core slabs”
Heron, Vol. 28, 1983, Nr. 3.
8.4 Walraven, J.C., “Reinforced concrete” Lecture Notes TU Delft, nov. 1991 (in Dutch).
8.5 Kupfer, H., Ruhnau, J., “Vorgespannte Schubbewerung”, Deutscher Ausschuss für
Stahlbeton, Heft 359, Berlin 1985.
8.6 Trinh, J., “Effort tranchant en beton armé et beton précontraint: influence de la
précontrainte”, Annales de l’Institut Technique du Batiments et des Travaux Publics”,
No. 360, April 1978, pp. 140-144.
8.7 Anderson, A.R., Anderson, R.G. "An assurance criterion for flexural bond in
pretensioned hollow core units", ACI-Journal, Vol. 73, Aug. 1976, pp. 457-464.
8.8 Netzel, H., “Schubtragfähigkeit von durch Hüllrohre geschwächten Trägerstegen”,

MSc-Thesis, TU Delft, July 1994.
8.9 Keller, K., N.V. Tue & M. Zink, “Influence of prestressing forces on the shear
capacity- part 2: Beams with shear reinforcement", Lacer no. 7, 2002.
April 2018 8-38

Chapter 9 – Crack width
9. Crack width control

9.1 Introduction
The tensile strength of concrete is about ten times smaller than its compressive strength.
Therefore, the principal idea of designing in reinforced concrete is to make the concrete
resist the compressive forces and the reinforcing steel the tensile forces. The fact that a
reinforced concrete structure demonstrates controlled cracking under service loads is
therefore more a proof for the correctness of the design than a reason for worrying.
Furthermore, from research and practice it is known that the durability of a concrete
structure is not endangered when:
1. the crack widths are small (for instance w < 0,4 mm in a non-aggressive
environment);
2. the concrete cover on the reinforcement is sufficient;
3. the concrete meets requirements with regard to strength and durability (mixture
composition).
Fig. 9.1 Crack in a reinforced concrete beam with crack width w > 0,4 mm (not a
calculation error but damage during transport) [9.1]
In general, crack width limitation does not play a dominant role in the design of a
concrete structure. Only in the case of a low reinforcement ratio, or when due to
aggressive environmental conditions strict limits apply (e.g. w < 0,1-0,2 mm), or when
particular liquid tightness requirements must be met, the crack width criterion can be
governing.
For prestressed concrete structures other arguments apply: prestressing steel is more
prone to corrosion than reinforcing steel. The first prestressed concrete structures were
therefore designed such that in serviceability limit state conditions no cracks would
April 2018 9-1

occur. Gradually, however, it was experienced that it is hardly possible to design a

structure that stays uncracked: to guarantee that no cracks at all will occur, three-
dimensional prestressing is required. For practical reasons, however, prestressing is
mostly limited to one or two directions.
Moreover, the loads to which the structure is exposed during its life time must be exactly
known if one aims at preventing all possible cracking. However, it is often impossible to
accurately predict all loading situations that might occur during the service life,
especially if temperature effects or differential settlements are concerned.
Furthermore, creep and shrinkage gradients over a cross-section might be of importance.

In Chapter 6 it turns out, that in parts of a structure having small cross-sectional
dimensions (small notional size), shrinkage and creep deformations develop faster than in
larger parts. Therefore, a redistribution of stresses occurs in the cross-section. Ignoring
this effect has in the past resulted in unexpected cracking, even in regions that should be
in compression according to the design calculations.
This has resulted in the conviction that the best solution is to design and detail a
prestressed concrete structure in such a way, that if cracking might occur for some
reason, the crack width is limited/controlled (well-distributed small cracks do not
endanger the performance of a structure). This can be achieved by combining
prestressing steel with reinforcing steel: reinforcing steel has better bond properties than
prestressing steel and is therefore better suited to limit crack widths.
April 2018 9-2

Figure 9.2 shows the results of bending tests on two prestressed concrete beams, both
prestressed with one bar Ø 26 mm. The first one is, in the lower flange, additionally
reinforced with 2 reinforcing bars Ø 6 mm, whereas the second one has 2 reinforcing bars
Ø 16 mm. Figure. 9.2 shows the crack patterns for those two beams, tested at the
Technical University of Aachen in Germany. The first beam demonstrated one large
crack (width w = 0,85 mm), whereas the second beam showed a large number of cracks
having small widths (w = 0,3 mm).
Fig. 9.2 Crack pattern in a prestressed beam having different ratios of reinforcing to
prestressing steel (dimensions mm) [9.2]
The awareness that effective crack width control can be obtained by using a combination
of reinforcing and prestressing steel is an important basic consideration for the
application of partially prestressed concrete. When using this mixed type of
reinforcement, an optimum solution is possible for any practical case. This offers the
designer various advantages, see references [9.3] and [9.4].
The requirements with regard to the allowable crack width wmax are based on the durability
of the structure and aesthetics. Requirements from EN 1992-1-1 table 7.1N are given in
table 9.1.
April 2018 9-3

Table 9.1 Recommended values of wmax [mm] (EN 1992-1-1 table 7.1N)
It is noted that the Dutch National Annex to EN 1992-1-1 prescribes other crack widths
limits and load combinations. The following table presents some data.
reinforcing steel and/or not- combination of reinforcement and

exposure class bonded prestressing steel bonded prestresing steel
frequent load combination frequent load combination
X0, XC1 wmax < 0,4 mm wmax < 0,3 mm
XC2, XC3, XC4 wmax < 0,3 mm wmax < 0,2 mm
XD1, XD2, XD3, wmax < 0,2 mm wmax < 0,1 mm
XS1, XS2, XS3
According to the Dutch National Annex to EN 1992-1-1, the crack width limit values
from the table can be multiplied by a factor
capplied
kx   2, 0
cnom
if the applied concrete cover on the reinforcement is larger than the nominal concrete
cover that follows from durability requirements.
Crack formation has been studied since decades. Test series have been carried out on,
among others, axially loaded reinforced concrete tensile bars, see fig. 9.3. By varying the
concrete strength class, the reinforcement ratio, the bar diameter and/or the number of
bars, a good impression was obtained of the basics of the cracking mechanisms.
In modern methods for crack width control in structures, the structural behaviour is
mostly modelled by defining an axially loaded, centrically reinforced concrete tensile bar
as shown in fig. 9.3. Therefore, this basic case is dealt with in detail first. At first the case
April 2018 9-4

of a concrete tensile bar containing reinforcing steel only is discussed. Subsequently the
case of combined reinforcing and prestressing steel is dealt with.
1 2 3 4 5 6 7 8 9 10 11 12
ρs =s 0,0113
= 1.13% ρs =
s0,0201
= 2.01% ρs =0,0314
s = 3.14% ρs =0,0452
s = 4.52%
Fig. 9.3 Reinforced concrete tensile members subjected to axial tension (Stevin
laboratory, TU Delft, 1976)
9.2 Bond between steel and concrete

When concrete cracks, the reinforcement has to carry the tensile force. The reinforcing bars
are stressed. As a result, the reinforcing bars activate bond stresses to transfer force to the
surrounding concrete. The bond stresses are caused by slip of the bars relative to the
concrete. By means of the bond between concrete and steel, the steel force is gradually
transferred to the concrete. For this load transfer mechanism the ribs on the bars are of
importance.
Based on the analysis of numerous test results it turned out that crack widths can be
calculated when assuming a constant bond stress between concrete and steel [9.5]. This
(mean) bond stress, that is independent of the magnitude of the slip between steel and
concrete, is about two times the mean tensile strength fctm of the concrete: τbm = 2 fctm (fig.
9.4).
Fig. 9.4 Simplified bond stress-slip relationship for short term static loading
April 2018 9-5

9.3 Crack formation in a reinforced concrete tensile member

As a basic tool for the calculation of crack widths in a concrete structure, the concrete
tensile member subjected to axial tension is used, see fig. 9.5. The behaviour of the
member will be studied in detail, assuming that the imposed strain gradually increases.
In the stage before the occurrence of the first crack, the strains of the steel and the concrete
are the same. The contributions of the steel and the concrete in carrying an external force
Ntot are:
N s  Es As 
N c  Ec Ac 
The total force:
N tot  N s  N c  Es As   Ec Ac   Ec Ac 1   e   
For ε = εs = εc:
N tot  Ec Ac 1   e    c (9.1)
where:
Es
e = is the ratio between the Young's moduli of steel and concrete;
Ecm
As
= is the reinforcement ratio.
Ac
When the concrete strain (c) increases, at a certain moment the tensile stress in the
concrete will reach the tensile strength of the concrete. Since the tensile strength of the
concrete over the length of the member is subjected to scatter, the first crack will appear at
a location where the tensile strength is the lowest, see figure 9.5a.
At the location where the concrete is cracked, the concrete tensile stress ct = 0. In a crack
only the steel carries the tensile force. As a result of the bond stresses between the steel and
the concrete, acting at both sides of the crack, the concrete is active again in carrying the
tensile force (fig. 9.5b). At a certain distance lt from the crack (the transfer length), the
concrete carries its original part of the tensile force N. Outside the transfer lengths, the
strains of concrete and steel are again equal, so that the undisturbed situation (as before
cracking) is present.
The distance required to again introduce a part of the cracking force into the concrete
depends on, among others, the bond strength bm. For the calculation of this distance, the
basics as shown in fig. 9.5 are used. From the assumption that the bond stress is constant
(fig. 9.5e), it follows that the course of the steel stress and the concrete stress along the
transfer length lt are linear, see figs. 9.5c and d.
April 2018 9-6

When the member is subjected to a strain (displacement-controlled testing), the force N

decreases as soon as a crack is formed. This behaviour can be explained from the
formation of a crack, which causes a reduction of the stiffness of the tensile member.
Since the total strain imposed on the member is the same before and after cracking, the
force drops from Ncr,1 (r = rupture) to N0, see fig. 9.6.
When further increasing the strain, the force increases again. However, the force cannot
be larger than Ncr,2, because then a new crack appears. The stage in which again and
again new cracks are formed at increasing imposed deformation is denoted as the “crack
formation stage”. In the crack formation stage the stress in the reinforcing steel s in a
crack reaches a maximum just before, at another location, a new crack arises. At that very
moment the stress is s = sr , see fig. 9.5c.
σse = e fctm
σc = fctm
Fig. 9.5 Tensile member model

a. Prismatic reinforced concrete member with crack
b. Force transfer/transmission (“disturbed”) area
c. Steel stress
d. Concrete tensile stress
e. Bond stress bm
April 2018 9-7

N
N cr,1
Nr,1
Ncr,2
Nr,2
t 2.t 2. t 2. t t Ncr Nr
N N
N0
disturbed area
Fig. 9.6 Crack formation stage
If, as a simplification, the calculations are based on the mean concrete tensile strength
fctm, the steel stress directly after the formation of a new crack is:
N cr f ctm
 s   sr   1   e   (9.2)
As 
In the undisturbed areas, see fig. 9.6, the steel stress is directly proportional to the
concrete stress:
 se =  e f ctm (9.3)
At a further increase of the imposed strain, the force increases again. However, the force
cannot exceed Ncr,2, since then a new crack appears.
The concrete tensile stress ct in a crack is zero, whereas at the end of the transfer length
the concrete stress is fctm (fig. 9.5d). This implies that the force transmitted over the
transfer length is:
N  Ac f ctm (9.4)
This force is transmitted by bond stresses along the perimeters of the bars over a distance
lt. It then holds:
N =  bm lt m  Ø (9.5)
where m is the number of reinforcing bars and Ø is their diameter. Combining eqs. (9.4)
and (9.5) (and using ρ = As / Ac and As = m 1/4 π Ø2) gives an expression for the transfer
length lt:
1 f ctm Ø
lt = (9.6)
4 bm ρ
The maximum crack width is equal to the difference between the elongation of the steel
and the elongation of the concrete over the length 2lt, so:
wmax  2 lt  sm  cm  (9.7)
April 2018 9-8

where sm and cm are the mean steel strain and concrete strain, respectively, along the
transfer length lt. The course of the stresses at both sides of a crack is shown in fig. 9.7.
The strains can be calculated from the stresses.
sr
se
steel stress
t t
fctm
concrete stress
Fig. 9.7 Course of steel and concrete stresses at both sides of a crack
The mean steel strain is:

1
εsm   σsr  σse  (9.8)
2 Es
Substitution of se from eq. (9.3) into this equation results in:
1
εsm   σsr  αe f ctm  (9.9)
2 Es
The mean concrete strain over lt is:
f ctm
εcm  2  1 α f (9.10)
e ctm
Ec 2 Es
Substituting sm and cm (from eqs. (9.9) and (9.10)) and eq. (9.6) in eq. (9.7) yields:
1 f ctm Ø 1
wmax = σ sr (9.11)
4 bm ρ Es
where sr follows from eq. (9.2).
When the strain is further increased, more and more cracks occur. The cracking process
continues until the tensile bar consists of “disturbed regions” only. When a certain
number of cracks are formed, the disturbed regions start to overlap each other. The
smallest spacing between two cracks is found, where at the end of a disturbed region
(which is at a distance lt from an already existing crack), a new crack has occurred. The
April 2018 9-9

largest spacing between two cracks is found, where a new crack has occurred at a
distance just smaller than 2lt from an already existing crack. The length of the tensile
member part in between the two cracks then is just too short for the bond stresses to make
that the concrete reaches its tensile strength again. As a result of this, the final crack
spacing varies between lt and 2lt. When, finally, the reinforced member consists of
disturbed regions only, the crack formation stage is finished. Although during further
increase of the strain the external force increases, no new cracks are formed. The stage
after the crack formation stage is denoted as the “stabilised cracking stage”. In this stage,
no new cracks occur and existing cracks widen,.
At a further increase of the strain, and as a result also an increase of the force N (fig. 9.6),
the steel stress in the crack s exceeds sr (eq. (9.2)). Because the force transmitted from
the steel to the concrete does not increase (the bond stress is constant), the concrete strain
between the cracks does not increase. As a result, the increase of the crack width follows
from the additional elongation of the steel only.
Figure 9.8 shows a crack where at both sides the maximum crack spacing is 2 ℓt. In the
crack formation stage the maximum stress in the steel was sr (eq. (9.2)). After
completion of the crack formation stage, the steel stress further increases because of the
increasing external tensile force N. The increase of the steel stress is s = s - sr. All
corresponding elongation of the steel over the distance 2 ℓt results in an increase of the
crack width:
( s   sr )  2 t
w  (9.12)
Es
Fig. 9.8 Determination of the maximum crack width in the stabilised cracking stage
April 2018 9-10

The total crack width in the stabilised cracking stage is obtained by adding w from eq.
(9.12) to wmax from eq. (9.11). In combination with ℓt from eq. (9.6), the following
expression is obtained:
1 f ctm Ø 1
wmax =  σs  0,5 σsr  (9.13)
2 bm ρ Es
Eqs. (9.11) and (9.13) are continuous since if in eq. (9.13) s is replaced by sr, this
expression reduces to eq. (9.11) (crack formation stage). Equation (9.13) is the general
expression for the calculation of the maximum crack width in both the crack formation
and the stabilised cracking stage. To calculate the crack width, it is only necessary to
determine the transition point (strain) between the crack formation stage and the
stabilised cracking stage. This is discussed in section 9.5.
9.4 Long term effects

During the service life of a concrete structure, shrinkage of the concrete occurs.
Furthermore, structures or parts of structures can be subjected to long term constant loads
or dynamic loads. The influence of those effects is dealt with in the following.
The effect of shrinkage in the crack formation stage differs from that in the stabilised
cracking stage. When in the crack formation stage shrinkage occurs while,
simultaneously, the external imposed strain is kept constant, the external force tends to
increase. Since in the crack formation stage the external force cannot exceed the cracking
load Ncr, this implies that the existing crack widths will not increase. The result is that
additional cracks will develop.
In the stabilised cracking stage the shrinkage influences the crack width. In this stage no
new cracks can be formed. The shortening of the concrete then can only result in
widening of the already existing cracks. The influence of the shrinkage on the crack
width is explained on the basis of the behaviour of a reinforced concrete element having a
length 2 ℓt and restrained at both ends, see fig. 9.9(a). As a result of previous loading, the
element exhibits one crack in the centre and it is assumed that the stabilised cracking
stage is reached. On behalf of symmetry, only one half of the element is regarded, see fig.
9.9(b).
April 2018 9-11

lt
Fig. 9.9 Influence of shrinkage on the crack width (stabilised cracking stage)
The concrete tends to shrink, but this is counteracted by the steel. If the concrete would
be able to shrink freely (assuming that there is no bond between the steel and concrete),
the concrete strain would be cs, see fig. 9.9(c).
Note: Shrinkage implies a shortening of the concrete, so εcs < 0. In the following
expressions it will be assumed that the absolute value of the shrinkage is used, so εcs > 0.
To restore compatibility between steel and concrete, in a first step, a compressive force N
is applied to the steel. The length of the steel is made equal to the length of the concrete,
see fig. 9.9(d). The steel stress increases with Δs = cs Es whereas the stress in the
concrete c remains constant. This situation is regarded as the initial situation. In the next
step the steel and the concrete are bonded and the same force N is applied to the steel, but
now in the opposite direction. Now concrete and steel can act together to carry the tensile
force N. However, slip occurs between steel and concrete. Compared with the initial
situation, the final result is an increase of the steel stress s = cs Es and the
corresponding increase of the crack width follows directly from eq. (9.13).
The total crack width (load + shrinkage) in the stabilised cracking stage is:
1 f ctm Ø 1
wmax =  σs  0,5 σsr  εcs Es  (9.14)
2 bm ρ Es
Now the effect of a long term constant load and/or a varying load will be discussed.
It was stated before that assuming a bond stress τbm = 2,0 fctm between steel and concrete
gives good results for ribbed bars. Under a long term or dynamic load, the bond stress
decreases. Tests have shown that a value τbm,∞ = 1,6 fctm is realistic.
April 2018 9-12

In the crack formation stage, the reduced bond stress results in an increase of the transfer
length of 25%, and, as a result, a similar increase of the crack width. This follows directly
from eq. (9.11), when instead of τbm = 2,0 fctm the value τbm = 1,6 fctm is used.
For the stabilised cracking stage, the situation is different. In most cases, the load has
been applied over a short period of time. This implies that the value bm = 2,0 fctm holds
for the transfer length, and, as a result, also for the crack spacing. The influence of the
concrete in between the cracks depends on the bond stress developed. As a result, this
influence decreases when the bond stress decreases. It can be assumed that this reduction
is about 40%. This can be taken into account by replacing the coefficient 0,5 from eq.
(9.14) by 0,3.
When taking these effects into account, the following more general expression for the
crack width is obtained:
1 f ctm Ø 1
wmax =  σs   σsr   εcs Es  (9.15)
2 bm ρ Es
where
s is the steel stress in a crack under external tensile load;

sr is the maximum steel stress in a crack in the crack formation stage
f
= ctm 1   e   ;

cs is the shrinkage of the concrete (assumed > 0);
 is the reinforcement ratio As / Ac;
fctm is the mean tensile strength of the concrete.
The values for τbm,  and  are given in table 9.2.

Through recalculations these values for  (namely 0,5 and 0,3) have been slightly
modified and have become 0,6 and 0,4, respectively, in EN 1992-1-1. In table 9.2 these
values are enclosed by brackets.
Table 9.2 Values for τbm,  and  from eq. (9.15) for various conditions. The values
for  between brackets are the recalibrated values as applied in EN 1992-
1-1 by means of the coefficient kt (EN 1992-1-1 eq. (7.9))
crack formation stage stabilised cracking stage
short term  = 0,5 (0,6)  = 0,5 (0,6)

loading  =0  =0
τbm = 2,0 fctm τbm = 2,0 fctm
long term or  = 0,5 (0,6)  = 0,3 (0,4)

dynamic loading  =0  =1
April 2018 9-13

An expression for the maximum crack width wmax that agrees with the EN 1992-1-1
expression is obtained by substituting ℓt (eq. (9.6)) and sr (eq. (9.2)) in eq. (9.13). The
result is:
f ctm
σs -  ( 1 + αe ρ)
ρ
wmax = 2 lt
Es
where
α = 0,5 in both the crack formation stage and the stabilised cracking stage in case of
short term loading.
The maximum crack spacing:

sr,max  2 lt
The expression for wmax is (EN 1992-1-1 eq. (7.9)):
f ct,eff
 s  kt (1   e  p,eff )
 p,eff
wmax  sr,max
Es
where:
fct,eff is the mean value of the concrete tensile strength at the time when the crack may
first be expected to occur;
fct,eff = fctm or lower (fctm(t), if cracking is expected earlier than at a concrete age of 28
days);
kt is the influence of load duration: 0,6 (short-term loading) or 0,4 (long-term
loading);
ρp.eff is the reinforcement ratio of the effective tension area (see EN 1992-1-1 fig. 7.1),
which is a “hidden tensile member”.
Note that σs = σsr in the crack formation stage.
When taking into account the influence of shrinkage, the equation becomes:
f ct,eff
 s  kt (1   e  p,eff )   cs Es
 p,eff
wmax  sr,max
Es
April 2018 9-14

EN 1992-1-1 uses the following expression for the maximum crack spacing (EN 1992-1-
1 eq. (7.11)):
Ø
sr,max  k3 c  k1k2 k4
 p,eff
Since for k3 and k4 usually the recommended values 3,4 and 0,425 are used, sr,max reads:
Ø
sr,max  3, 4 c  0 , 425 k1k2
 p,eff
where:
c is the concrete cover to the longitudinal reinforcement;

k1 is a coefficient which accounts for the bond stress of the reinforcement:
= 0,8 for high bond (ribbed) bars;
= 1,6 for bars with an effectively plain surface;
k2 is a coefficient which accounts for the distribution of the strain over the height of
the concrete area considered (EN 1992-1-1 eq. (7.13)):
= 0,5 for bending;
= 1,0 for pure tension.
ρp,eff is the reinforcement ratio of the tensile member.
Note that the EN 1992-1-1 expression for maximum crack spacing is similar to eq. (9.6)
if it is assumed that the mean bond stress is directly proportional to the concrete tensile
strength. EN 1992-1-1 uses a minimum value for the crack spacing (k3 c) since this
demonstrated to give better results, especially for members having a high reinforcement
ratio.
Note:
The EN 1992-1-1 wmax expression (EN 1992-1-1 eq. (7.9)) has ρp,eff, the
reinforcement ratio of the “hidden” tensile member (see section 9.6) to calculate
tension stiffening. This implies that the steel stress σs follows from a cross-sectional
analysis, whereas σsr, the steel stress directly after cracking, follows from the
cracking force of the hidden tensile member. This approach might give unexpected
results, especially for a member loaded in bending to a level that is not much higher
than the cracking moment Mcr. Therefore, in this textbook, the steel stress directly
after cracking follows from a cross-sectional analysis, using the cracking force
(bending moment) as input.
9.5 The transition point between the crack formation stage and the
stabilised cracking stage
Figure 9.10 shows the schematised behaviour of the tensile member.
April 2018 9-15

In the first linear branch (1) the concrete is uncracked.
When the cracking load Ncr (r = rupture) is reached, crack formation starts (2). At
increasing deformation the load N can not exceed Ncr.
After the completion of crack formation, the force N increases. The dashed line shows the
N - Δℓ / ℓ relation of the steel reinforcement only. The line representing the behaviour of
the reinforcement with the surrounding, cracked concrete (3) is assumed to be parallel to
the line of the steel only (4). To calculate the position of line (3), it is assumed that the
mean crack spacing is 1,5 ℓt. The resulting representative zone in between two cracks is
shown in fig. 9.11.
1 = uncracked stage
2 = crack formation stage
3 = stabilised cracking stage
Ncr 4 = steel bar(s) only (unbonded)
5 = yielding of reinforcement
Fig. 9.10 Deformation of a reinforced concrete tensile member
April 2018 9-16

1,5. t = mean crack distance
s.As s.As
sx s
sm
0,75. t 0,75. t
Fig. 9.11 Variation of stresses along a representative part of the member having a
length equal to the mean crack spacing and located in between two cracks
The stress in the steel at a distance x = 0,75 ℓt is:

0, 75 lt τ bm  Ø
σ sx =σ s - (9.16)
4Ø
1 2
where:
1 f ctm Ø
 bm  2, 0 f ctm and lt =  (eq. (9.6)).
4 bm ρ
Equation (9.16) then becomes:
0, 75 f ctm
σ sx =σ s - (9.17)
ρ
The mean steel stress:
0,375 f ctm
σ sm =σ s - (9.18)
ρ
The mean steel strain:
σ s 0,375 f ctm f
sm = -  s - 0, 4 ctm (9.19)
Es Es ρ Es ρ
With the aid of fig. 9.10 it is now possible to determine the strain for which the cracking
pattern can be regarded to be complete (the end of the crack formation stage).
The horizontal branch (2) is defined by the cracking force:
N cr  Ac f ctm 1   e   (9.20)
April 2018 9-17

whereas the following branch (3) is described by:
 f 
N  Es As  s  0, 4 ctm  (9.21)
 Es ρ 
When substituting N = Ncr in eq. (9.21) and by using eq. (9.20), the intersection point of
branches (2) and (3) is found:
f ctm  0, 6  αe ρ  0, 6 f ctm
s   (9.22)
Es ρ Es ρ
If the imposed strain of the reinforced concrete tensile member is smaller than the value
resulting from eq. (9.22), the member is in the crack formation stage. For a higher value
of the strain the stabilised cracking stage is reached. In most practical applications where
imposed deformations apply, for instance in case of a temperature drop or concrete
shrinkage at fixed boundary conditions, the imposed strain is mostly smaller than the
value given by eq. (9.22). The crack pattern then is not stabilised and the structure is in
the crack formation stage (2).
If, on the contrary, the member is subjected to a tensile load N > Ncr, fig. 9.10 shows that
the member is in the stabilised cracking stage (3).
Note:
In order to simplify the calculations, fig. 9.10 is a schematised representation of the
actual behaviour. A horizontal plateau (2) will in reality not occur, since the cracking
force gradually increases: the first crack is formed at the weakest spot and each following
crack occurs at a location where the tensile strength of the concrete is slightly higher. The
most realistic description might therefore be to use the lower bound 5% characteristic
concrete tensile strength for the first crack and to end with the mean tensile strength for
the last crack. However, in the crack formation phase, the maximum crack width is found
at the highest cracking force. This implies that the designer should focus on the
maximum cracking force. By using a constant cracking force based on the mean concrete
tensile strength fctm, this is incorporated in the model.
9.6 The effective tensile area around the reinforcement

In the previous derivations, it was assumed that the reinforcing steel is uniformly
distributed over the concrete cross-section. As a result, the forces transferred by bond
from the reinforcing steel to the concrete, do not have to spread over a large concrete area
to load all the concrete in tension. If, however, the bond forces must spread, not only the
bond properties play a role, but also the geometry of the element and its reinforcement.
When the expression for the crack width was derived, it was implicitly assumed that the
transfer length lt is large compared with the dimensions of the cross-section. At the
location where the tensile strength of the concrete is reached, the tensile stresses are then
April 2018 9-18

almost uniformly distributed over the concrete cross-section. As a result, the cross-
section will fully crack through once the concrete tensile strength is reached.
If one dimension of the cross-section is much larger than the other, the behaviour is
different. At the position where the tensile strength of the concrete is reached, cracking
starts. The distribution of the tensile stresses that spread into the concrete is not uniform,
see fig. 9.12a. The crack now does not proceed over the full width of the tensile member.
Only if the force, introduced by the reinforcement, is more or less uniformly spread over
the width of the element, the full cross-section cracks. Close to the reinforcing steel, the
behaviour is identical to that of the reinforced tensile member discussed in the previous
sections. In the wide member only a few cracks reach the outer surface of the concrete,
fig. 9.12b.
Because several internal cracks join, continuous cracks are formed. A continuous crack
often has a disproportionally large width since, at the outside of the concrete, the
deformation is concentrated in a small number of cracks. If the reinforcement is
concentrated at the outside of the cross-section, see fig. 9.12c, the outer surface
demonstrates many cracks having small widths, whereas wide internal cracks occur.
These findings demonstrate that there is a so-called "effective concrete area" around the
reinforcement. The width of cracks that occur in this area is controlled by the
reinforcement, whereas the crack width outside this area is uncontrolled. The relations
derived in the previous sections apply to the effective concrete area only.
c
a b
fc
concentrated crack
b d
effective area
many cracks with small width

Ac1 eff
c b
Fig. 9.12 Cracking behaviour of elements with concentrated reinforcement and a high
ratio of element width to transfer length (b / lt)
In elements loaded in bending, similar phenomena occur. In the case of deep beams, the
main tensile reinforcement limits the crack widths over an area close to the
April 2018 9-19

reinforcement. If no substantial additional reinforcement over the height of the cross-

section is provided (e.g. web reinforcement), the relatively small cracks at the bottom of
the beam join and develop into wide cracks (fig. 9.13).
Fig. 9.13 Crack pattern in deep beam with concentrated longitudinal reinforcement at
the bottom and hardly any longitudinal web reinforcement, loaded in pure
bending (left: side view; right: cross-section)
It is found from tests that the effective area around the reinforcement (the “hidden”
reinforced concrete tensile member) in beams, walls and slabs can be defined as shown in
fig. 9.14 [9.6] (EN 1992-1-1 cl. 7.3.2 (3) & fig. 7.1). The height of the effective area is:
hc,eff  2,5 (h  d )
bending: hc,eff  (h  x) / 3
tension : hc,eff  h / 2
The reinforcement ratio used in the crack width expression must be based on the effective
concrete area. In case of reinforced concrete (EN 1992-1-1 eq. (7.10)):
As
ρp,eff  (9.23)
Ac,eff
It is noted that the reinforcement ratio from expression (9.23) has a general setup by
using the subscript p. This makes it suited for combinations of reinforcing (s) and
prestressing steel (p).
April 2018 9-20

Fig. 9.14 Effective concrete area [9.6] (EN 1992-1-1 fig. 7.1)
9.7 Summary of the most important expressions for crack width

control in reinforced concrete
When calculating the maximum crack width, the serviceability limit state (SLS) applies.
The basis for crack width control is the reinforced concrete tensile member, subjected to
axial tensile loading. The general expression used to calculate the maximum crack width
is:
1 f ctm Ø 1
wmax =  σs   σsr   εcs Es  (9.24)
2 bm ρ Es
where
s is the steel stress in a crack under external tensile load;

sr is the maximum steel stress in a crack in the crack formation stage
cs is the shrinkage of the concrete (assumed > 0);
 is the reinforcement ratio As / Ac;
fctm is the mean tensile strength of the concrete;
,  coefficients according to table 9.3
April 2018 9-21

Table 9.3 Values for τbm,  and  from eq. (9.24) for various conditions.
crack formation stage stabilised cracking stage
short term  = 0,5  = 0,5

loading  =0  =0
long term or  = 0,5  = 0,3

dynamic loading  =0  =1
To distinguish between the crack formation stage and the stabilised cracking stage, the
following principles generally apply:
 Imposed deformation: Generally, the member is in the crack formation stage. It

should, however, be confirmed that:
0, 6 f ctm
 (9.25)
Es  p,eff
 External loads: Generally, the stabilised cracking stage applies, provided that the
tensile force is higher than the cracking force:
N > Ncr (9.26)
This chapter deals with the question how to calculate the largest crack width that occurs
in a structure in the serviceability limit state (SLS). However, requirements might also
focus on the mean crack width. In the crack formation stage, scatter in the crack width
only comes from the scatter in the concrete tensile strength. The relation between the
maximum and the mean crack width then is (approximated):
wmax = 1,3 wm (9.27)
In the stabilised cracking stage, the scatter is larger because now also the stochastic
nature of the crack spacing plays a role. The relation now is (approximated):
wmax = 1,7 wm (9.28)
April 2018 9-22

9.8 Some applications
In case of imposed deformations, which occur, for example, due to a decrease of the
temperature or restrained shrinkage, the structure will generally be in the crack formation
stage.
For a number of cases eq. (9.24) was used for the crack formation stage, see fig. 9.15.
The figure shows diagrams for concrete strength classes C20/25 and C35/45, both for
short term and long term or dynamic loading. An important question is which concrete
tensile strength has to be used in the calculations. In practice, there are various
influencing factors that reduce the tensile strength of the concrete, e.g. temperature
gradients from solar radiation, moisture gradients, or micro cracking from temperature
gradients in the hardening phase of the concrete. In practical situations, the tensile
strength will, therefore, be somewhat smaller than in laboratory conditions. To take all
these factors in account, a general reduction factor of 0,75 for the concrete tensile
strength is introduced. Furthermore, it turns out that the actual concrete strength in a
structure can be considerably higher than the characteristic cube compressive strength of
the mixture. The 28-day characteristic strength is the basis for most structural resistance
calculations (e.g. bending, shear, torsion, etc). With regard to crack width, however, an
increase of the strength results in an increase of the cracking force and, as a result, in an
increase of crack width. To be on the safe side, it is suggested to increase the
characteristic strength with 10 N/mm2. This implies that for a concrete C20/25, a
characteristic strength of 20 + 10 = 30 N/mm2 is used in the following calculations.
When implementing the two aforementioned corrections (namely a 0,75 tensile strength
reduction factor and a 10 N/mm2 compressive strength increase), the mean concrete
tensile strength in the actual structure is (using the EN 1992-1-1 table 3.1 expression):
f ctm  0, 75  0,30   20  10 
2/3
 2, 2 N/mm 2
For concrete C35/45 the result is:
f ctm  0, 75  0,30   35  10 
2/3
 2,9 N/mm 2
For the various cases, the data listed in table 9.4 apply. The long term bond stress is 80%
of the short term bond stress (table 9.2). With regard to concrete creep, it is assumed that
the creep coefficient is 2,0. Creep is taken into account by reducing the Young's modulus
of the concrete with a factor 1 + 2,0 = 3,0.
April 2018 9-23

Table 9.4 Basic values used to construct the design curves from fig. 9.15
concrete load Ec fctm bm αe = Es / Ec
duration [GPa] [N/mm2] [N/mm ] 2
[-]
short 30 2,2 4,4 7

C20/25
long 10 2,2 3,5 20
short 34 2,9 5,8 6

C35/45
long 11 2,9 4,6 18
In the crack formation stage, the maximum steel stress in a crack is (eq. (9.2)):
f ctm
 s   sr  1   e  

Substituting this equation in eq. (9.24) yields:

2
f ctm Ø (1   e ρ)
wmax = (9.29)
4 ρ 2 bm Es
In fig. 9.15a-d, the relations between wmax, Ø and ρ are presented for the four cases
considered. The allowable bar diameter Ø can be read for a given reinforcement ratio and
an allowed maximum crack width, provided that the tensile member is in the crack
formation stage.
April 2018 9-24

ρ [%]
[%] 3.0 ρ [%]
[%] 3.0
Wmax =0.1 mm
Wmax =0.1 mm
0.2
2.0 2.0
0.2 0.3
0.3
0.4
1.0 0.4 1.0
a b
0.0 0.0
0 10 20 30 40 0 10 20 30 40
 [mm]  [mm]
Ø [mm] Ø [mm]
ρ [%] ρ [%]
 [%] 3.0  [%] 3.0
Wmax =0.1 mm Wmax =0.1 mm
0.2
2.0 0.2 2.0 0.3

0.3
0.4
0.4
1.0 1.0
c d
0.0 0.0
0 10 20 30 40 0 10 20 30 40
 [mm] Ø
[mm]
[mm]
Ø [mm]
Fig. 9.15 Required reinforcement ratio for a given bar diameter and maximum crack
width for reinforced concrete tensile members in the crack formation stage:
a. C20/25, short term loading
b. C20/25, long term or dynamic loading
c. C35/45, short term loading
d. C35/45, long term or dynamic loading
Apart from the allowable crack width there is another important criterion: the yield stress
of the steel should not be exceeded at first cracking. On the basis of eq. (9.2), it can be
derived that the minimum reinforcement ratio to prevent yielding is:
1
 min  (9.30)
 s / f ctm   e
For s the design yield stress of the steel fyd = 435 N/mm2 (reinforcing steel B500) is
used. With regard to the tensile strength of the concrete, it should be taken into account
that it might be higher than the mean strength as derived from the characteristic 28-day
cylinder compressive strength (EN 1992-1-1 table 3.1). Previously, fck was increased with
10 N/mm2. To be at the safe side, the reduction with a factor 0,75 is not applied in this
case.
April 2018 9-25

So, for a strength class C20/25 it is found that:
f ctm  0,30   20  10 
2/3
 2,9 N/mm 2
When using αe = 7 (short term loading), eq. (9.30) results in ρmin = 0,70%.
An overview of the minimum reinforcement ratio for a range of concrete strength classes
is given in table 9.5. Note that EN 1992-1-1 table 3.1 uses a different expression for
strength classes > C50/60. When using the parameters discussed before, this expression
is:
2/3
 53  8  10 
f ctm  2,12  ln 1    4, 4 N/mm 2 for strength class C53/65.
 10 
Tabel 9.5 Minimum reinforcement ratio (pure tension)

minimum
reinforcement C20/25 C28/35 C35/45 C45/55 C53/65
ratio ρmin for
steel B500
[%] 0,70 0,82 0,93 1,07 1,10
Example 1a
A long slab without expansion joints has a thickness h = 150 mm and is on a stiff
bedding, see fig. 9.16. Due to the assumed high friction between the slab and the subsoil,
deformation of the slab from shrinkage of the concrete and a temperature drop, cannot
occur. In the calculation it is assumed that the shrinkage cs = 0,25 · 10-3 and that the
mean temperature will drop 25 ºC below the initial temperature during construction. The
concrete strength class is C20/25 and the reinforcing steel class is B500. The crack width
allowed is 0,20 mm. Calculate the required amount of reinforcement and the bar
diameter.
Fig. 9.16 Reinforced concrete slab subjected to imposed deformation
Solution
When the concrete shrinkage and the temperature drop are not restrained, the total
deformation of the slab is:
April 2018 9-26

 = 0,25 · 10-3 + 25 · 1,0 · 10-5 = 0,50 · 10-3
Because shortening cannot occur, tensile stresses will develop. Since it is a long term
deformation, creep should be taken into account. In case of a creep coefficient of 2, the
fictitious Young's modulus of the concrete is about 30 / (1 + 2) = 10 GPa. In case of an
uncracked slab, the fully restrained deformation would cause a concrete tensile stress:
c = Ec  = 10 · 103 · 0,50 · 10-3 = 5,0 N/mm2
The concrete will crack because its tensile strength is only about 2,2 N/mm2 (table 9.4).
It is now assumed that the cracked structure is in the crack formation stage (which often
holds in case of an imposed deformation). This assumption has to be validated
afterwards. From fig. 9.15b (long term loading) it follows that for wmax < 0,20 mm and Ø
= 12 mm, the reinforcement ratio should be at least ρ = 1,1%. If the reinforcement is
provided in two layers (at bottom and top of the slab), the maximum bar spacing is
135 mm. The required reinforcement ratio is of course higher than the minimum
reinforcement ratio (ρmin = 0,70%, table 9.5) since minimum reinforcement is applied to
prevent the steel from yielding whereas the reinforcement ratio from fig. 9.15 is required
to control crack widths. At a reinforcement ratio ρ = 1,1%, the transition point from the
crack formation stage to the stabilised cracking stage is at a strain:
0, 6 f ctm 0, 6  2, 2
   0, 60 103
Es  200 103  0, 011
This shows that the assumption that the structure is in the crack formation stage, was
correct.
Example 1b
Now, the same situation is regarded, but for a slab having a thickness of 400 mm (fig.
9.17). A reinforcement ratio of 1,1% would also in this case be sufficient to meet the
crack width requirement. However, it will be discussed whether a lower reinforcement
ratio might be applied if the reinforcement is concentrated in the outer zones (top and
bottom) of the slab since concentrating reinforcement in the outer zones has proven to be
more effective than uniformly distributing it over the full cross-section (fig. 9.12c).
Solution
The height of the effective tensile area at the bottom and the top of the slab (fig. 9.14):
heff  2,5  h  d   2,5  c  12 Ø   2,5   25  12 12   78 mm
It is now assumed that the slab is in the crack formation stage. This implies that the
tensile force in the slab is constant: N = Ncr. The steel stress directly after cracking and
steel stress at the SLS both follow from the cracking force off the full cross-sectional area
of the slab:
April 2018 9-27

f ctm
 s   sr  1   e  

effective tensile
member; 78 mm
effective tensile
member; 78 mm
Fig. 9.17 Thick reinforced concrete member subjected to an axial force from an
imposed deformation (dimensions mm)
After substituting the previous expressions into eq. (9.24) and assuming  = 0,3,  = 1
and bm = 2 fctm, the result is:
1 1 Ø 1  f ctm f 
wmax =   1 +  e ρ   0,3 ctm 1 +  e ρ   εcs Es 
2 2 ρp,eff Es  ρ ρ 
(9.31)
1 1 Ø 1  f ctm 
wmax =   0, 7 1 + e ρ   εcs Es 
2 2 ρp,eff Es  ρ 
In this expression, a clear distinction is made between ρ (based on the total height h of the
cross-section) and ρp,eff (related to the effective tensile area and used to calculate the
crack spacing in the effective tension zones or “hidden” tensile members). The ratio
between ρp,eff and ρ is:
ρp,eff 400
=  2, 6
ρ 2  78
To meet the requirement wmax = 0,20 mm, using ρp,eff = 2,6 ρ, fctm = 2,2 N/mm2, αe = 21,
Ø = 12 mm and cs = 0,2510-3, it is found from eq. (9.31) that
1 1 12 1  2, 2 
wmax =  3 
0, 7  1 + 21ρ   0, 25 103  200 103 
2 2 ρp,eff 200 10  ρ 
1,5 1  1,54 
0, 2= 5  1 + 21ρ   50 
2, 6 ρ 10  ρ 
April 2018 9-28

ρ > 0,8010-2, or ρ > 0,80% (As = 3200 mm2/m)
When applying reinforcing bars Ø 12 mm, the maximum bar spacing is 70 mm.
It is now checked whether the slab is actually in the crack formation stage as assumed.
The transition point is at a strain:
0, 6 f ctm 0, 6  2, 2
   0,83 103  0,50 103 -> crack formation stage
Es  200 10  0, 0080
3
Example 2
A slab, spanning in one direction, is subjected to a variable load qq of 4 kN/m2 (fig. 9.18).
The concrete strength class is C20/25. The reinforcement (at the tension side) consists of
bars Ø12 – 175 mm, which is As = 645 mm2/m. The concrete cover c = 15 mm. The
longitudinal reinforcement ratio:
As 645
sl    0, 25 102
bd  12 
1000   275  15  
 2
It can be assumed that the maximum load acts only for a short period of time at the
structure. The maximum crack width in that situation should be not larger than 0,4 mm.
4 kN/m22
qqq q==4kN/mm
275
6000
15
12-175
Fig. 9.18 Slab spanning in one direction with qq = 4 kN/m2 (dimensions mm)
Solution
According to table 9.4, the cracking stress of concrete C20/25 is 2,2 N/mm2. The flexural
tensile strength is (EN 1992-1-1 eq. (3.23)):
fctm,fl = (1,6 – h) fctm
When applying this expression to the slab (h = 0,275 m), the cracking stress is
2,9 N/mm2.
The maximum bending moment in SLS is:
April 2018 9-29

M max  18  (0, 275  25  4, 0)  6, 02  48,9 kNm/m

The cracking moment is:
M cr   cr W  2,9  16 1000  2752  36, 6 106 Nmm/m  36, 6 kNm/m
The structure cracks in bending and will be in the stabilised cracking stage (the bending
moment at the SLS is greater than the cracking moment). Figure 9.19 shows the stresses
in the cross-section of the slab before and after cracking.
compression
tension steel force
Fig. 9.19 Internal equilibrium in a cross-section just before (left) and after cracking
(right)
The height of the compression zone follows from:
x
  e sl  ( e sl ) 2  2 e sl
d
Using αe = 7 for short term loading and ρsl = 0,25 · 10-2 (based on d = 275 – 15 =
260 mm) the result is:
x
 0,17
d
This implies that x = 43 mm and the internal lever arm after cracking is:
z = d – x / 3 = 254 – 43 / 3 = 240 mm
The maximum steel stress s in a crack follows from:
M max 48,9 106

s    316 N/mm 2
As z 645  240
For a structural element loaded in bending, the height of the effective tension area around
the steel reinforcement is the smallest value of 2,5 (c + Ø/2) and (h - x) / 3 (see EN 1992-
1-1 cl. 7.3.2 (3) and section 9.6).
The governing value is
(h - x) / 3 = (275 - 43) / 3 = 77 mm (see fig. 9.20).
April 2018 9-30

hidden
77.3
77 hidden
tie tensile member
12
Fig. 9.20 Hidden reinforced tensile member in the slab (dimensions mm)
The effective reinforcement ratio of the hidden tensile member:
ρp,eff = 645 / (77 · 1000) = 0,0084
Note:
It is noted that the EN 1992-1-1 section on cracking (7.3) does not use the flexural
tensile strength; crack formation and the behaviour of the hidden tensile member
are related to the uni-axial tensile stress.
The steel stress sr in a crack directly after cracking follows from:
M cr 36, 6 106
s    236 N/mm 2
As z 645  240
Note:
The steel stress in the tensile member in a crack in the crack formation stage
according to EN 1992-1-1:
f ctm
 sr 
 p,eff
1      0,2,0084
e p,eff
2
 1  7  0, 0084   277 N / mm 2
Because the maximum moment in SLS is larger than the cracking moment, the structural
slab is in the stabilised cracking stage. The load considered has a short term character. It
then follows from table 9.3 that  = 0,5,  = 0 and bm = 2 fctm.
Substituting these values into eq. (9.24) results in:
1 f ctm Ø 1 1 1 12 1
wmax =  σs  0,5 σsr        316  0,5  236   0,35 mm
2 bm ρp,eff Es 2 2 0, 0084 200 103
It can be concluded that the requirement wmax < 0,4 mm is met.
9.9 Crack width control in prestressed concrete

9.9.1 Introduction
One of the main aims of the full prestressing of a structure is to keep it uncracked under
any possible SLS load combination. So, basically, a crack width control for fully
April 2018 9-31

prestressed structures (no tension in concrete allowed), or structures with reduced

prestressing (small tensile stresses in the concrete allowed, but no cracking), makes no
sense.
Experience has shown, however, that it is practically impossible to guarantee that a

prestressed concrete structure will not crack. Possible causes for unexpected cracking are,
among others, differential shrinkage and creep of adjacent structural parts having
different dimensions, temperature effects and differential settlements.
Fortunately, practice has learned that limited cracking in prestressed concrete does not
endanger its durability. On the basis of this important conclusion, a new way of
prestressing was developed, the so-called partial prestressing, which allows controlled
crack formation. In Chapter 12 the technology of partial prestressing will be dealt with in
more detail. In this section it will, in advance of the more general aspects of partially
prestressed concrete, be investigated how the crack width can be calculated and
controlled in the case of cross-sections provided with a combination of reinforcing and
prestressing steel.
9.9.2 Behaviour of a concrete tensile member with reinforcing and prestressing

steel
The bond properties of prestressing steel are not as good as those of reinforcing steel. In
practice, therefore, reinforcing steel is often applied additionally to prestressing steel.
Figure 9.21 shows the development of the stresses in reinforcing steel, prestressing steel
and concrete. Since the bond properties of prestressing steel are less good than those of
reinforcing steel bars having the same diameter, prestressing steel has a larger transfer
length than reinforcing steel. However, the same crack width must be found for both
types of steel. This implies that the increase of the stress in the crack is smaller for
prestressing steel than for reinforcing steel.
The cross-sectional areas of the reinforcing and prestressing steel are denoted as As and
Ap, their diameters as Øs and Øp, and their bond stresses as bms and bmp. Also a
distinction is made between their transfer lengths, using lst for reinforcing steel and lpt for
prestressing steel.
The increase of the tensile force in the steel compared when the stress in the concrete is
zero, is:
N = As s + Ap p (9.32)
Since the crack width is the same for reinforcing steel and prestressing steel, ws = wp.
Assuming Es = Ep and neglecting the concrete deformation over the transfer length, it is
found that:
ws,max = wp,max (= wmax)
As a result:
April 2018 9-32

s  p
 2lst   2lpt (9.33)
2 2
or:
s lst = p lpt (9.34)
For the transfer lengths (fig. 9.21) this implies:
 s Øs
lst  (9.35a)
4  bms
 p Øp
lpt  (9.35b)
4  bmp
where:
p is the increase of the steel stress in the prestressing steel in a crack;
s is the steel stress in the reinforcing steel in a crack.
Substituting eqs. (9.35a-b) in eq. (9.34) results in:
 s  s Øs  p  p Øp
  
Es 4  bms Ep 4  bmp
Which leads to:
 bmp Øs
 p   s    s 1 (9.36)
 bms Øp
where 1 is the so-called bond factor, also taking into account different diameters (EN
1992-1-1 cl. 6.8.2 (2) & eq. (6.64); cl. 7.3.2 (3) & eq. (7.5)):
 bmp Øs Ø
1     s (9.37)
 bms Øp Øp
April 2018 9-33

N = As.s + Ap.p
Ap
4.As
sr  p
pr
s p f
e ctm
ne.fctm
steel stress
 st
 pt
area of influence uncracked
concrete stress
Fig. 9.21 Development of steel and concrete stresses in the area adjacent to a crack
As can be seen from eq. (9.37), the bond factor contains both the ratio between the bond
strengths (ξ) and the ratio between the bar and tendon diameters. In EN 1992-1-1 table
6.2 the bond strength ratio is given (see table 9.6).
Table 9.6 Bond strength ratio  according to EN 1992-1-1

prestressing steel pre-tensioned bonded, post-tensioned
< C50/60 > C70/85
smooth bars and wires not applicable 0,3 0,15
strands 0,6 0,5 0,25
indented wires 0,7 0,6 0,30
ribbed bars 0,8 0,7 0,35
Because of the high sensitivity of prestressing steel to corrosion, the allowable crack
widths are generally 0,1 mm smaller than for reinforcing steel that is under the same
environmental conditions (exposure classes).
By virtue of the use of the bond factor, combinations of prestressing and reinforcing steel
can easily be transformed to the case of only reinforcing steel. The same principles as
previously discussed in detail, can be used. The only difference is that the reinforcement
ratio is defined in another way.
For reinforced concrete a reinforcement ratio ρ = As / Ac is used. In case of combinations

of reinforcing and prestressing steel, the reinforcement ratio of the hidden tensile member
is defined as:
April 2018 9-34

As  1 Ap
 p,eff   s  1  p (9.38)
Ac,eff
For the calculation of the crack width, the stress in the reinforcing steel is used as a basis,
so eq. (9.24) can still be used:
1 f ctm Ø 1
wmax =  σs   σsr   εcs Es  (9.39)
2 bm ρp,eff Es
The only difference is that the (combined) reinforcement ratio ρp,eff follows from eq.
(9.38). Also with regard to the calculation of s and sr, the influence of the prestressing
steel can easily be taken into account.
The use of eq. (9.39) will be illustrated with an example in section 9.10. For detailed
information about crack width control for combinations of reinforcing steel and
prestressing steel, reference is made to [9.7].
9.10 Example of crack width control for a combination of prestressing

and reinforcing steel
The beam shown in fig. 9.22 is designed for a maximum variable load qq = 15 kN/m. The
selfweight of the structure is 8,5 kN/m and there is an additional permanent load of
0,9 kN/m. The layout of the post-tensioned prestressing tendons along the beam is
parabolic. At the beam ends, the centre of gravity of the prestressing steel coincides with
the centre of gravity of the cross-section of the beam. The concrete strength class is
C35/45.
Properties of the cross-section:
Ac = 0,34 m2
Ic = 0,031 m4
d = 0,850 m
zb = 0,635 m (distance from centre of gravity to concrete bottom fiber)
zt = 0,365 m (distance from centre of gravity to concrete top fiber)
Wb = 0,049 m3
April 2018 9-35

qq = 15 kN/m
qvar=15 kN/m
800
1200 kN
483
200
20 m
c=25
1000
10 Ac = 0.34 m2
635
12 57 zu = 0.635 m
= 0.031 m4
300
. Wu = 0.049 m3
3x75
97
qdl = 9 kN/m
.
225 d = 0.850 m
za = 0.365 m
103
dimensionsin mm
225
Fig. 9.22 Crack width control in a partially prestressed beam
The bending reinforcement is concentrated at the bottom of the beam, see fig. 9.22. It
contains mixed reinforcement, and has a height of approximately 300 mm (fig. 9.22, left).
This part of the beam contains 9 reinforcing bars Ø 12 mm (As = 1018 mm2) and two
Dywidag prestressing tendons 4/15,7 Y1860C, both having a cross-sectional area of
4 · 150 = 600 mm2 (Ap = 1200 mm2) and an external duct diameter of 57 mm.
For the calculation of the crack width, a tensile strength fctm = 3,2 N/mm2 (EN 1992-1-1
table 3.1) is used. The flexural tensile strength fctm,fl follows from the axial tensile
strength fctm using the relation (EN 1992-1-1 eq. (3.23)):
fctm,fl = (1,6 – d) fctm , with fctm,fl ≥ fctm and d in m.
Since in this case d > 0,60 m, fctm,fl = fctm = 3,2 N/mm2.
The prestressing force is 1200 kN, which implies a working stress pm∞ = 1000 N/mm2.
The allowable short term crack width is 0,2 mm. Verify whether this criterion is met.
Solution
For a prestressing force of 1200 kN and a parabolic layout of the prestressing tendons, the
equivalent prestressing load-balancing force is:
Pm Pm 1200

qp =  2   11, 6 kN/m
R l 20, 02
8f 8  0, 483
So, the maximum uniformly distributed load in SLS is:
qmax = 8,5 + 0,9 + 15,0 – 11,6 = 12,8 kN/m
April 2018 9-36

which results in a maximum bending moment in SLS:
Mmax = 1/8 · 12,8 · 20,02 = 640 kNm
The mean concrete compressive stress cpm in the cross-section caused by the axial
prestressing force is:
 cpm  Pm / Ac  3,5 N/mm 2
So, the cracking moment is:
M cr  Wb (3, 2  3,5)  328 106 Nmm  328 kNm
This demonstrates that, under the maximum moment at the SLS, the beam is cracked in
bending. To calculate the crack width, the maximum steel stress s must be known.
Therefore, first the internal equilibrium is regarded. To obtain equilibrium the following
requirements apply:
H  0 and M  0
The corresponding forces and strains for H = 0 are shown in fig. 9.23.
Fig. 9.23 Horizontal equilibrium in the midspan cross-section
April 2018 9-37

[N/mm2]
2723,3
1.75 3.50
 co
' [‰]
ε [‰]
c
Fig. 9.24 Design stress-strain relation for C35/45
In this case (fig. 9.22), the centers of gravity of reinforcing steel and prestressing steel
almost coincide, so that approximately ds = dp = d = 850 mm. The bond factor  is 0,5
(table 9.6).
When assuming that the concrete is still in the elastic stage (c  1,75‰, figs. 9.23 &
9.24), horizontal equilibrium requires:
Pm  Ap  p  As  s  N c1  N c2  0
or
dx d x 1 c
Pm  Ap1    c Ep  As    c Es   f cd bf x 
 x   x  2 1, 75 103
(9.40a)
1  c0
  f cd  bf  bw  x  hf   0
2 1, 75 103
where:
 x  200 
 c0   c (9.40b)
 x 
It is noted that the concrete stress-strain relationship from fig. 9.24 refers to ULS design.
It is strictly speaking not applicable in a SLS check. Since concrete is usually in the linear
elastic stage in SLS, the only relevant concrete property in SLS design is the Young's
modulus. This modulus can be read from EN 1992-1-1 table 3.1. However, this table
presents a modulus related to short term loading only. Long term loading can be
incorporated by taking into account the creep coefficient. As demonstrated in Chapter 6,
creep is often difficult to quantify precisely. This is partly caused by the difficulties
encountered when predicting the loading history (short and long term load components
and their duration). Therefore, in practice often estimated values are used. This is also
done in this case study, where the Young's modulus of concrete is derived from the ULS
stress-strain diagram. As mentioned before, this is strictly speaking not correct and a
precise calculation should reveal the magnitude of the creep coefficient.
April 2018 9-38

The diagram from fig. 9.24 results in a Young's modulus:
35
f cd 1,5
Ec    13,3 103 N/mm 2
 c3 1, 75 103
EN 1992-1-1 table 3.1 shows Ec = 34 · 103 N/mm2 for C35/45. This implies that
implicitly a creep coefficient of 34 / 13,3 - 1 = 1,6 is included in the analysis.
The requirement for M = 0 is also graphically represented in fig. 9.23.
To have equilibrium of moments, the following condition should be met when using the
centroidal axis as reference:
M max  N c1 e1  N c2 e2  Ap  p eps  As  s eps
or
1 c 1  c0
M max   f b x e1  
3 cd f
f cd  bf  bw  x  hf  e2 
2 1, 75 10 2 1, 75 103
(9.41a)
dx d x
 Ap1    c Ep  d  zt   As    c Es  d  zt 
 x   x 
where
e1  zt  13 x (9.41b)
e2  zt  hf  13  x  hf  (9.41c)
The bond factor of the tendons, consisting of strands, is ξ = 0,5 (EN 1992-1-1 table 6.2;
also see table 9.6). The ratio of the diameters of the prestressing and reinforcing steel also
must be included to find the actual bond factor ξ1. EN 1992-1-1 cl. 6.8.2 prescribes:
Øp  1, 6 Ap in case of tendons consisting of strands.
The result is:
Øp  1, 6 600  39 mm
The bond factor including the influence of the diameter (eq. (9.37)):
 bmp Øs Ø 12
1     s  0,5   0,39
 bms Øp Øp 39
April 2018 9-39

In the eqs. (9.40a-b) and (9.41a-c), two unknown variables exist, namely c and x.
Solving the equations results in:
x = 350 mm and c = 1,0 ‰.
Nc1 = 18,67 · 105 N

Nc2 = 2,46 · 105 N
ΔPm,∞ = 1,30 · 105 N
Ns = 2,91 · 105 N
e1 = 248 mm
e2 = 115 mm
The steel strain is:
d x  850  350 

s   c  
3
 1, 0 10  1, 43 10
3
 x   350 
The stress in the steel is s Es = 286 N/mm2, which also follows from the previously
presented steel force (Ns / As).
The same calculation procedure is used to calculate the steel stress directly after cracking.
The applied bending moment is Mcr = 328 kNm.
Results:
x = 590 mm and c = 0,59 ‰.
Nc1 = 18,57 · 105 N

Nc2 = 5,83 · 105 N
ΔPm,∞ = 0,24 · 105 N
Ns = 0,53 · 105 N
e1 = 168 mm
e2 = 35 mm
The steel strain is:
d x  850  590 

s   c  
3
  0,59 10  0, 26 10
3
 x   590 
The stress in the steel is s Es = 52 N/mm2 (= Ns / As).
The reinforcement ratio of the hidden tensile member is now calculated.
The height of the hidden tensile member is 2,5 (h - d) but not greater than (h - x ) / 3.
April 2018 9-40

h = 1000 mm
d = 850 mm
x = 350 mm at the SLS
Result: hc,eff = (1000 - 350) / 3 = 217 mm is governing.
As  1 Ap 1018  0,39  (2  600)

 p,eff    0, 030
Ac,eff 217  225
The crack width can now be calculated using eq. (9.39):
1 f ctm Ø 1
wmax =  σs   σsr   εcs Es 
2 bm ρ p,eff Es
where  = 0,5, bm = 2 fctm, s = 286 N/mm2, sr = 52 N/mm2 and Ø = 12 mm. No
shrinkage will be included (short term loading).
The result is:
1 1 12 1
wmax =      286  0,5  52  0   0,13 mm
2 2 0, 030 200 103
It follows that the requirement wmax < 0,20 mm is met.
In this calculation, x and c can be estimated to quickly verify whether the crack width
criterion might be critical. To present a procedure, the stress distribution is presented for
the maximum bending moment in SLS assuming that the concrete would not crack. The
stresses at the top and the bottom of the beam are (fig. 9.25):
Pm M max zt 1200 103 640 106  365

 ct       3,5  7,5  11, 0 N/mm 2
Ac Ic 0,34 10 6
0, 031 1012
Pm M max zb 1200 103 640 106  635

 cb       3,5  13,1  9, 6 N/mm 2
Ac Ic 0,34 10 6
0, 031 1012
April 2018 9-41

2
11,0
-10.8N/mm
N/mm2
Nt
458
466 mm
9.1 N/mm2 2
9,6 N/mm
Fig. 9.25 Concrete stresses for an assumed uncracked cross-section
The total tensile force carried by the tension zone:
Nt = 1/2 · 225 · 466 · 9,6 = 503 · 103 N
After cracking, the tensile force has to be taken over by the steel. Due to cracking, the
internal lever arm increases. Therefore, the force that has to be resisted by the steel, is
smaller than Nt. A value of 0,9 Nt can be regarded as a conservative estimation.
The stress in the steel is:
0,9 N t 0,9  503 103

s    305 N/mm 2
As  1 Ap 1018  0,39  (2  600)
The more accurate calculation resulted in s = 286 N/mm2.
If s = 305 N/mm2 is substituted in eq. (9.39), it becomes clear that the crack width
criterion is met. On the basis of this result it can be decided immediately that a more
accurate calculation is not necessary.
Apart from the control of the crack width at the level of the main reinforcement at the
bottom of the beam, it should be avoided that large cracks in the web occur, see fig. 9.13.
Figure 9.26 gives a design aid that shows up to which height crack distributing
longitudinal web reinforcement is required [9.8].
In the case considered, h = 1000 mm and h – x = 1000 – 350 = 650 mm. From fig. 9.26 it
follows that for this value, in combination with the requirement wmax < 0,2 mm, hw ≥
200 mm. This condition was already met, see fig. 9.22, so that no further crack
distributing reinforcement has to be provided.
April 2018 9-42

1000
hw [mm]
1 2
800 0. 0. 0.4
w= w= w=
x-c
600
h-x-c
400
hw
200
0
0 400 800 1200 1600 2000
h-x [mm]
Fig. 9.26 Height of the web that should be provided with crack distributing
reinforcement, in order to avoid uncontrolled cracking in the web [9.8]
If a crack distributing reinforcement is required, the amount can be determined easily.

The maximum strain of the longitudinal (main) reinforcement at the SLS is known. The
strain distribution between this reinforcement and the compression zone is also known
(Bernoulli's law). Since this is a case of imposed deformation, the diagrams from fig.
9.15 can be used. The effective tension area of the hidden tensile member follows from
the geometry, see fig. 9.27.
2,5 (c + Ø/2) = 2,5·(25 + 10 + 12/2) = 103 mm
Fig. 9.27 Effective tension area of a longitudinal bar in the web of the beam
For more information on crack width control in high beams, reference is made to [9.9].
April 2018 9-43

EN 1992-1-1 cl. 7.3.3 (3) states that a crack width calculation for the web can be carried
out by assuming that it behaves as a member loaded in pure tension at a steel stress that is
half the steel stress in the main (longitudinal) reinforcement.
9.11 Literature
9.1 Kay. E.A., Fookes, P.G., Pollack, D.J.: “Deterioration related to chloride ingress”,
Concrete, Nov. 1981, pp. 22-28.
9.2 Falkner, H.: “Risse im Beton – Theorie und Praxis”, Vorträge Lindauer
Bauseminar 1985, Veröffentlichung Bauakademie Bieberach, Band 38.
9.3 Bruggeling, A.S.G.: "Structural concrete: science into praxis“, Heron, Vol. 32,
No. 2, 1987.
9.4. Bruggeling, A.S.G.: “Partially prestressed concrete – a design challenge”, Journal

of the Prestressed Concrete Institute, March/April 1985, pp. 140-171.
9.5. König, G., Fehling, E.: “Grundlagen zur Rissbreitenbeschränkung im

Spannbeton-bau”, Beton- und Stahlbetonbau, Heft 12, 1988.
9.6. CEB/FIP Model Code 1990, Bulletin 2/3-214.
9.7 König, G, Fehling, E.: "Zur Rissbreitenbeschränkung im Spannbetonbau“, Beton-

und Stahlbetonbau, Heft 6 & 7, 1988
9.8. CEB Manual on "Cracking and deformations", Bulletin d’Information No. 143,
1981, pp. 2.1-2.80.
9.9 Braam. C.R., “Control of crack width in deep reinforced concrete beams”, Heron,
Vol. 35, 1990, No. 4.
9.10 Noakowski, P., “Die Bewehrung von Stahlbetonbauteilen bei

Zwangsbeanspruchung infolge Temperatur“, Deutscher Ausschuss für Stahlbeton,
Heft 246, 1978.
9.11 Falkner, H., "Zur Frage der Rissbildung durch Eigen- und Zwangsspannungen
infolge Temperatur in Stahlbetonbauteilen", Deutscher Ausschuss für Stahlbeton,
Heft 208, 1969.
April 2018 9-44

Chapter 10 - Detailing
10. Detailing of prestressed concrete structures
10.1 Introduction
Professional detailing of prestressed concrete structures not only results in more easy
construction, but also has a positive influence on the quality of the structure. By skillful
detailing, not only the amount of damage will reduce, but also the repair costs during the life
time of the structure will be minimized.
On the basis of literature [10.1] – [10.4], it turns out that a large part of the damage can be
traced back to lack of understanding of the flow of forces in the structure and the ignorance
of a number of simple basic rules for design and construction.
It is possible to avoid part of the errors that lead to damage, namely by representing the flow
of forces using strut and tie models and by using the equivalent prestressing load method
(load balancing method; prestressing is preloading). Such representations can be the basis for
detailing. This principle is used in this chapter.
10.2 Introduction of prestressing forces
10.2.1 Introduction of prestressing forces through anchors
In the case of prestressing with post-tensioned steel, the tensile forces in the prestressing
tendons are introduced in the member by prestressing anchors. The concentrated forces
gradually spread in the structure. The length required to have a uniform load distribution, is
called the disturbance length of de St.-Venant. This disturbance length s is independent of the
magnitude of the force, but depends on the geometry of the member and the position(s) of the
prestressing anchor(s). According to de St.-Venant, the length s is equal to the largest width
across which the load has to be spread.
Figure 10.1 gives an example of a centrically introduced prestressing force, where the stress
trajectories are represented by a strut and tie model. From the truss model shown in fig. 10.1,
it turns out that the splitting tensile force Nspl is required to have force equilibrium in
transverse direction. The magnitude of this force Nspl follows from equilibrium:
 h 
N spl  14 Fp 1  1  (10.1)
 h
It is noted that the force is not denoted as P, but as F, since the theory is not exclusively
applicable to prestressing forces, but holds for the introduction of forces in general.
Generally, the anchor force must be distributed in both the height and the width direction (the
splitting tensile force for the width direction is found by replacing h and h1 in eq. (10.1) by b
April 2018 10-1

and b1). Splitting reinforcement is then required in two principal directions. In order to carry
the splitting tensile force Nspl, stirrups or mesh reinforcement can be used. In a circular cross-
section one can also choose a spiral reinforcement (This reinforcement should not be
confused with the spiral reinforcement that is used as confining reinforcement in an
anchorage system to strengthen the concrete to resist the high compressive stresses under the
anchor plate).
sh  = Fp/Ac
Fp/2
h/2 b1
Fp/2
Nspl
h1
Fp/2
Fp/2 h/2
h/2 h/2
b
a b
Fig. 10.1 a. Truss model to determine the splitting tensile force Nsp
b. Splitting tensile reinforcement
The cross-sectional area Aspl follows from the relation Aspl = Nspl / s, where s is the
maximum steel stress allowed. It makes sense to choose a stress level s below the design
yield stress fyd. If s is equal to the design yield stress fyd, the equilibrium requirement from
ULS is met, but the crack width in SLS might be too large. EN 1992-1-1 cl. 7.3.3 presents (in
Tables 7.2N and 7.3N) a good indication for the maximum stress in the steel allowed to avoid
crack width problems (s = 200 – 400 N/mm2, depending on maximum crack width allowed).
The splitting tensile reinforcement Aspl should be distributed. It is preferred to use much bars
with a small diameter instead of a smaller number of bars with a larger diameter, see fig.
10.1b. The splitting tensile reinforcement should confine the cross-section and be anchored
such that it is also effective at the outsides of the cross-section.
In practice, the prestressing force is often introduced by more than one tendon (fig. 10.2a).
By using more tendons, the splitting tensile force is reduced. Figure 10.2a shows a truss
model to determine the splitting tensile forces and their location. The splitting tensile
reinforcement As1 follows from the splitting tensile forces in the same way as shown before.
It makes sense to apply also some reinforcement at the free end of the beam (As2). This
reinforcement is required to resist unforeseen loads like impact loads or forces from
differential shrinkage. Furthermore, in the stage of prestressing, splitting forces along the free
end may occur. Their magnitude and location depends on the sequence of prestressing. The
stress distributions at the concrete surface following from linear elastic analyses indicate that
April 2018 10-2

the surface tensile force between two point loads (Fp each) is about 0,1 Fp (Dutch code NEN
6720)
s1/3h  = 3.Fp/Ac As2 As1
Fp/2
Fp/2
Nspl Fp
Fp/2 Fp/2
Fp
Fp
a b
Fig. 10.2 Truss model and reinforcement to introduce a combination of prestressing

forces
Also when the prestressing force is introduced eccentrically, the position of the splitting
tensile force and its magnitude can be determined with a strut and tie model, see fig. 10.3. In
this case, the disturbance length is about equal to the beam height h because the force has to
be spread over this distance. The stress distribution at the edge of the disturbed area can
easily be determined, because here principally plane sections remain plane (Bernoulli). By
orientating the compression bars (struts) in the model such that they follow the compression
trajectories, the most efficient equilibrium system is obtained. From the forces in the tensile
ties T and the allowed steel stress s, the required cross-sectional area of the splitting
reinforcement is obtained.
April 2018 10-3

Fp
h1
D -
T D x
h
T
T
0
+
sh
Fig. 10.3 Strut and tie model for eccentric load introduction
The tensile reinforcement required to introduce a number of eccentrically applied

prestressing forces can also be easily determined in an approximating graphical way, see fig.
10.4. It is emphasized, that the distribution of forces is mostly a 3D (three-dimensional)
problem, rather than a 2D problem. This should be taken into account when detailing the
structure. The end of a TT-beam is used as an example to illustrate this (fig. 10.5).
h1
D1
h1 -
T D2
x
D3
T
h1
Fig. 10.4 Flow of forces for a number of eccentrically introduced prestressing forces.
The TT-beam is provided with a transverse end beam. Due to prestressing, tensile forces
occur in this transverse beam too. Therefore, in transverse direction always sufficient
reinforcement should be provided. In large bridge beams it might even be necessary to apply
prestressing in transverse direction.
April 2018 10-4

P
P
Fig. 10.5 Cracking in a transverse beam at the end of a TT-girder, caused by 3D-
spreading of the prestressing forces P
In prestressed TT-beams without transverse end beam, the tensile force caused by the
introduction of the prestressing forces can only occur in the deck. The introduction of the
forces occurs in almost the same way as in case of a rectangular cross-section, see fig. 10.6.
FpP FpP
Fig. 10.6 Introduction of prestressing forces P in a TT-beam without a transverse end

beam
A similar type of problem can occur in an I-shaped cross-section in which only the flanges
are prestressed. When the spatial spreading of the forces is not accounted for, cracks with
large widths occur at the end of the beam, see fig. 10.7.
April 2018 10-5

cracking
Fig. 10.7 Cracking in an element with an I-shaped cross-section caused by splitting

tensile forces between the groups of prestressing tendons
Parts of the structure outside the influencing zone of the prestressing force partially restrain
the deformations caused by prestressing in other parts of the structure, see fig. 10.8. Those
parts have to stay connected to the part that is stressed by prestressing (compatibility
requirement). In this respect, attention has to be given to the support areas as well. Most
girders are supported as close as possible to their ends. A part of the supporting forces then
might act outside the influencing area of the reinforcement. It should then be assured that the
support reaction is introduced without causing damage to the structure, see fig. 10.9.
Fig. 10.8 Reinforcement partly outside the influencing area of the prestressing force
Fig. 10.9 Correct (left) and poor (right) detailing at a support
April 2018 10-6

10.2.2 Introduction of the prestressing force by bonding
Introducing the prestressing force Pm0 from the steel into the concrete by bond, requires a
transmission length lpt (EN 1992-1-1 cl. 8.10.2.2 & eq. (8.16)). Further, a dispersion length
ldisp can be distinguished (EN 1992-1-1 fig. 8.16), which is required to develop a linear
distribution (over the cross-section) of the concrete stress caused by the prestressing force
(fig. 10.10).
h
a
c c
lpto
e
ldisp
prestressing
prestressing force
force Fp Pm0
1
/3 lpt
~1/3
a
o
a bond stress a
Fig. 10.10 Transmission length lpt and dispersion length ldisp in a beam prestressed with
pre-tensioned tendons
The centre of gravity of the bond stresses is at about 1/3 lpt from the end of the beam.
For the dispersion length ldisp it can be written:
ldisp  13 lpt  s  13 lpt  h (10.2)
where lpt is the transmission length and s is the interference length of de Saint-Venant. As
boundary condition it should hold:
ldisp  lpt (10.3)
EN 1992-1-1 cl. 8.10.2.2 provides the following expression for the transmission length lpt:
 pm0
lpt  1  2 Ø (10.4)
f bpt
where:
α1 depends on the way of releasing (gradual or sudden);
α2 depends on the type of tendon;
April 2018 10-7

Ø is the nominal diameter of the tendon;

σpm0 is the tendon stress just after release;
fbpt is the bond stress.
f bpt  p1 1 f ctd (t ) (10.5)
where:
ηp1 depends on the type of tendon and the bond situation at release;
η1 depends on the bond conditions;
fctd(t) is the design value of the concrete tensile strength at time of release.
As an example, the transmission length of a 9,3 mm (3/8") strand (7 wires, see table 2.3) in a
C50/60 concrete is calculated. It is assumed to be released when the concrete has reached a
strength class C28/35. The strand is at the bottom side of the element. The Y1860S7 strand is
assumed to develop its maximum allowed initial stress σp0 (EN 1992-1-1 cl. 5.10.3):
σp0 = min.(0,75 fpk; 0,85 fp0,1k) = min.(0,75 · 1860 ; 0,85 · 1674) = min.(1395; 1423) =
1395 N/mm2.
The bond stress is:
1,94
f bpt  p1 1 f ctd (t )  3, 2 1, 0   4,14 N/mm 2
1,5
The transmission length in case of gradual release:
 pm0 1395
lpt  1  2 Ø  1, 0  0,19  9,3   595 mm
f bpt 4,14
The importance of the transmission length is twofold:

 Firstly, the strands must have sufficient anchorage capacity in case the design load acts
on the element. This is discussed in section 10.2.3.
 Secondly, splitting might play a role. This is discussed in this section.
A short transmission length is favourable for the anchorage capacity but unfavorable with
respect to splitting action. Short transmission lengths result in a more concentrated transfer of
forces and, as a result, in more concentrated splitting forces which in turn lead to higher
splitting stresses. To have sufficient safety with respect to splitting, it is now assumed that
the splitting forces are introduced over half the transmission length from eq. (10.4).
Now the method will be discussed that enables the check and design of splitting action in a
structure prestressed with pre-tensioned steel.
April 2018 10-8

Three different types of tensile stresses can be distinguished over the dispersion length,
namely (see fig. 10.11):
 Spalling stresses.
 Bursting stresses.
 Splitting stresses.
The three cases are discussed individually.
Fig. 10.11 Different types of tensile stresses in concrete prestressed with pre-tensioned
steel (CEB bulletin no. 181, 1987)
Spalling stresses
These stresses are generated by the eccentric introduction of the prestressing force, see fig.
10.3. Figure 10.12 shows typical cracks caused by the spalling stresses. The tensile spalling
stresses are close to the loaded end surface and are most affected by tendon eccentricity.
They occur away from a single tendon or between tendons in case of multiple tendons.
Fig. 10.12 Typical cracks due to spalling stresses
April 2018 10-9

According to EN 1992-1-1 cl. 8.10.3, the required transverse reinforcement, in case of

prestressing with pre-tensioned steel, is determined from the stress distribution calculated by
the linear theory of elasticity. The prestressing force has to be multiplied with an extra load
factor γP,unfav = 1,2 to account for local effects (EN 1992-1-1 cl. 2.4.2.2 (3)). Strut and tie
models or another appropriate representation must be chosen.
A graphical method has been developed by Den Uijl1. This method is equivalent to the theory
of elasticity and uses the diagram shown in fig. 10.13, where k is the core radius, wspl is the
width of the cross-section at the level considered and ep0 is tendon eccentricity. The spalling
stress σspl is determined as follows: Starting from (ep0 - k) / h and lpt / ep0 the value for
σspl · wspl · ep0 / Fpi can be read from the diagram. After substitution of the known values for
wspl, ep0 and Fpi, the spalling stress σspl is obtained. In a calculation according to EN 1992-1-1
the force Fpi is Pm0. The maximum spalling stresses are considerably reduced as the
transmission length increases.
yy
spl.wspl.epo/Fpi
0.12
lpto//eeop0==00 1 sl
2
σspl spalling
splitting spl
stress
stress
0.10 3
4 0
x
0.08 5
6 wspl
y
0.06
8
0.04 10
x
h
15
k
epo
0.02 20
epo-k
50 Fpi
0 lo
0 0.04 0.08 0.12 0.16 0.20 0.24
(epo.k)/h lpt
(ep0 – k ) / h
Fig. 10.13 Diagram according to den Uijl to determine the spalling stress spl
It is also an option to use the equivalent prism analogy. The height of this prism (see fig.
10.14) follows from the condition that no shear stresses are transferred along the line B-B.
Therefore, the resulting axial force on the prism has to be zero. The length of the prism
follows from:
2
 0, 6 lpt 
lspl  Cspl h where Cspl  1    (10.6)
 h 
1
Uijl J.A. den, ”Verbundverhalten von Spanndraht – Litzen im Zusammenhang mit Rissbildung im
Eintragungsbereich”, Betonwerk + Fertigteiltechnik, Volume 1, 1985, p. 28-36.
April 2018 10-10

The spalling force can be determined by using the assumption that the horizontal internal
lever arm zspl between the resulting tensile and compressive forces (see fig. 10.14) is 0,5 lspl,
from which it follows that:
2M
N spl 
lspl
M M
B B B
B
h yy Nspl
x
Nspl
B B
 spl
Zspl  0,5.  spl
Fig. 10.14 Analogue equivalent prism to determine the transverse reinforcement required
when prestressing with pre-tensioned steel
The bending moment M follows from the elastic stress distribution above the line B-B, at the
end of the equivalent prism (at a distance lspl from the end face of the beam). The spalling
stresses σspl can be calculated from the spalling force Nspl by assuming that Nspl is linearly
distributed over a quarter of the prism length lspl:
N spl 16 M
 spl  
1
2  14  lspl b 2
lspl b
where b is the width of the cross-section at the level where the spalling stresses occur. In case
of expected cracking, the total spalling tensile force Nspl has to be carried by transverse
reinforcement.
Bursting stresses
The tendon force is gradually transferred to the concrete by bond forces. The bursting
stresses are related to the longitudinal component of these forces. The bursting stresses are
perpendicular to the beam axis. They are around the tendon and at a distance from the beam
end. Their occurrence can be explained as follows: In case of a concrete specimen subjected
to a concentrated load, tensile stresses perpendicular to the specimen’s longitudinal axis
occur, see fig. 10.1. The transverse tensile force is in equilibrium with the transverse
compressive force that occurs close to the point where the load is introduced. This transverse
stress distribution is well known from, e.g., a tensile splitting test on a cylinder or cube. All
individual longitudinal bond force components present along the transfer length, cause
individual (transverse) bursting stress distributions. Their combined action is a bursting stress
distribution that results in compressive stresses at the beginning of the transfer length and
April 2018 10-11

tensile stresses at a distance from the beam end (see fig. 10.11), in the anchorage zone. A
bursting crack will, therefore, occur at a distance from the beam end.
Tensile splitting stresses

The tendons are first tensioned and then cut (fig. 2.23). Because of the stress release, the
tendons not only shorten but also become thicker due to the loss of the lateral contraction
introduced by the prestressing (wedge action). In addition, for strands a certain twist (screw
action) is generated during the retraction into the concrete.
Next to the bonding stresses between the concrete and steel, the wedge action exerts a
pressure on the concrete (σr in fig. 10.15). As a result this internal pressure generates
tangential tensile stresses σt. The stresses develop in transverse direction.
The splitting stresses reach their maximum value where the bond stresses are highest. This is
close to the beam end, see fig. 10.11.
The splitting stresses are sufficiently accounted for when the reinforcement required for
bursting and spalling confines the tendons. If no confining reinforcement is present, the
concrete cover should meet certain requirements.
When the concrete cover on the strands and the mutual distance between the strands are large
enough, splitting and bursting are prevented to occur. On the basis of a linear-elastic stress
analysis, general rules can be derived. EN 1992-1-1 cl. 4.4.1.2 states that with respect to
transmitting bond forces safely, the cover on the prestressing wires or strands has to be at
least:
 1,5Ø for strands and plain wires;
 2,5Ø for indented wires.
EN 1992-1-1 cl. 8.10.1.2 & fig. 8.14 provide information on the arrangement of pre-
tensioned tendons. Minimum horizontal spacing is max. (2 Ø; 20 mm; max. aggregate size +
5 mm); minimum vertical spacing is max. (2 Ø; 20 mm).
It is noted that splitting stresses occur in pre-tensioned elements since in post-tensioned

elements the prestressing force is introduced locally, namely at the steel anchor plate, not
using bond. However, also in case of post-tensioning, the prestressing force causes bursting
stresses.
April 2018 10-12

r t

a tm
r
r
t
a = bond stress
r = radial compression stress
t = tangential tension stress (in circumference direction)
Fig. 10.15 Stresses caused by bonding
10.2.3 Anchorage capacity of pre-tensioned steel
After cutting a wire or strand, it slips into the concrete. During this process of slipping, the
strands/wires expand in lateral direction due to the loss of lateral contraction (the so-called
effect of Hoyer; wedge action). The slipping stops as soon as a sufficiently large frictional
force has been built up along the tendon. This equilibrium state is disturbed when the steel
stress increases because of external loading, e.g. when cracking occurs. This is now
discussed.
The stress development in the prestressing steel, present over the transmission length lpt, is
then part of the limit criterion for anchorage failure. Tests show that it is sufficiently accurate
to assume a linear steel stress along the transmission length. The steel stress increases from
zero to σp0 over the distance lpt.
A B
o
lpt
pl
σp0
A
B
p
Fig. 10.16 Stress increase in the steel near a flexural crack but outside the transmission
zone
When next to the transmission zone lpt a flexural crack occurs, another mechanism is
introduced. At an increase of the stress level in the steel, bond stresses are generated at both
sides of the crack. The stress peak in the steel at the crack decreases over a certain distance as
shown in fig. 10.16. The stress peak Δσp in the flexural crack cause a stress increase in point
April 2018 10-13

A at the end of the transmission length. In that case, the limit equilibrium state over lpt is
disturbed. The wire or strand then slips and anchorage failure occurs. When the stress peak
has just damped out in point A, the top of the peak (point B) is on the envelope for anchorage
failure, which is indicated by a dashed line in fig. 10.16.
EN 1992-1-1 cl. 8.10.2.3 provides information to determine the envelope for anchorage
failure (see also fig. 10.17).
The length lpt of the transmission zone follows from eq. (10.4). However, an additional factor
1,2 must be introduced to be on the safe side. Therefore, the transmission length to be used in
this analysis is lpt2 = 1,2 lpt. Further it holds for the total transmission length in case of a steel
stress increase to σpd (EN 1992-1-1 eq. (8.21)):
 2Ø  pd   pm 
lbpd  lpt 2 
f bpd
where fbpd follows from the actual concrete strength class in ULS.
Fig. 10.17 Ultimate limit state with respect to anchorage failure (EN 1992-1-1 fig. 8.17)
A = tendon stress; B = distance from end
In the example from the previous section with the 9,3 mm strand it was found that:
lpt = 595 mm.
April 2018 10-14

It is now assumed that the full design value of the tensile strength of the prestressing steel is
activated (which is a conservative approach):
σpd = fpk / 1,1 = 1860 / 1,1 = 1691 N/mm2.
The bond stress (EN 1992-1-1 eq. (8.20)):
f bpd  p2 1 f ctd
where:
ηp2 depends on the type of tendon and the bond situation;

η1 depends on the bond conditions;
fctd is the design value of the concrete tensile strength.
For C50/60: fctk,0,05 = 2,9 N/mm2, fctd = 2,9 / 1,5 = 1,9 N/mm2.
f bpd  1, 2 1, 0 1,9  2,3 N/mm 2
The initial stress σp0 was assumed to be 1395 N/mm2. If it is assumed that time-dependent
losses are 150 N/mm2, σpm∞ = 1395 - 150 = 1245 N/mm2.
The total anchorage length is:
lbpd  1, 2  595  0,19  9 ,3 

1691  1245  714  343  1057 mm
2 ,3
The check with respect to anchorage failure can now be done with the aid of fig. 10.18. At
the design load, the distance x from the support to the first flexural crack is calculated. EN
1992-1-1 (cl. 8.10.2.3 (1)) uses the characteristic tensile strength of the concrete (fctk,0,05) to
calculate the cracking moment. (Note that the cracking moment also has a component from
the axial prestressing force.) The steel stress in this first crack follows from:
M x Vx d cot  N p,x
N p,x   ;  p,x 
z z Ap
where Np,x follows from the equilibrium of moments about S, as shown in fig. 10.19.
Because the equilibrium of moments is set up about a point S in a shifted cross-section (shift
is d cotθ with respect to position x) it is often denoted as the parallel axis theorem. The
parameter z is the distance between the working lines of Np and Nc and is equal to the sum of
the eccentricities ep and ec (see fig. 10.19).
The result of this analysis is a prestressing steel stress σpx (fig. 10.18), calculated using the
bending moment at the cross-section x + d cotθ from the support. The prestressing steel stress
April 2018 10-15

can develop provided there is sufficient anchorage capacity to transfer the prestressing force
to the concrete. A distance x is available to transfer the force.
x d.cotg 
lpt2 lbpd - l-pt2

o po o
px

σp0 pl
f
fpkpu/ 1,1
Fig. 10.18 Check of the anchorage capacity of pre-tensioned steel
If σp,x is inside the envelope from fig. 10.18, sufficient anchoring capacity is guaranteed. The
prestressing steel force caused by the bending moment then is smaller than the force that can
be generated by bond; the strand is not pulled out. The positions of σpx and the envelope in
fig. 10.18 demonstrate that this is the case: There is sufficient anchorage length to anchor a
force that is somewhat higher than the force associated with σpx (fig. 10.18).
x d cot  d cot 
N
N cc
Vx S
crack Mx ec
VVx x+dcot
d cot x
z
centroidal axis ep
N p,x
prestressing steel
section of beam free-body diagram
Fig. 10.19 Calculation of the steel stress in the crack
10.2.4 Intermediate anchors
Prestressing tendons are sometimes anchored at intermediate places in a beam. An example

are large box girder bridges, where part of the prestressing tendons is only provided in the
top and/or the bottom flange, mostly to keep the web width as small as possible, see fig.
10.20.
April 2018 10-16

support prestressing
continuous prestressing
Fig. 10.20 Box girder with prestressing tendons only in top and bottom flange
In such a case, the prestressing tendons must be anchored at different locations along the
girder (fig. 10.20). For the detailing of such intermediate anchors, four effects have to be
considered. Ignoring them might cause to damage (fig. 10.21):
1. The cover can spall-off from the prestressing tendons because of the transverse pressure
from the curved tendon (action 1 in fig. 10.21).
2. Splitting cracks can occur due to the spreading of compressive stress trajectories (action 2
in fig. 10.21).
3. The corbel can crack-off (action 3 in fig. 10.21).
4. The corbel and its adjacent concrete area deform as a result of the introduction of the
prestressing force. From compatibility requirements, the area behind the corbel is forced
to follow these deformations. As a result, tension occurs behind the corbel. Ignoring this
effect can result in large cracks behind the corbel (action 4 in fig. 10.21, and the cracking
from fig. 10.22).
1
3
2
Fig. 10.21 Damage due to poor detailing at an intermediate anchor
Fig. 10.22 Compatibility cracks behind an intermediate anchor, caused by missing

suspension reinforcement
April 2018 10-17

The reinforcement should therefore be designed such that all those types of damage are
prevented. Figure 10.23 shows the various types of reinforcement required to avoid
problems. The functions of the reinforcement are:
As1 Tensile tie as main reinforcement in the corbel.
As2 Splitting reinforcement in relation to the spreading of compressive stress
trajectories (compare fig. 10.2).
As3 Reinforcement for an unforeseen pressure by curvature of the prestressing
tendons (it might be that the construction drawings indicate a straight tendon
profile, whereas due to ignorance at the construction site, the tendons have a
curved profile).
As4 Reinforcement for the (planned) force because of tendon deviation.
As5, As6 Suspension reinforcement for limiting the crack width behind the corbel (control
of compatibility cracking).
As6
As5
As4
As3
As2
As1
Fig. 10.23 Reinforcement detailing in a corbel that anchors a prestressing tendon
10.3 Effects of prestressing tendon curvature

Prestressed concrete structures often contain curved prestressing tendons. When detailing the
structure, the effects of tendon curvature should be regarded because otherwise damage
might occur.
Fig. 10.24 Tendon curvature associated with the assembly of the reinforcement
April 2018 10-18

Figure 10.24 shows a beam with an I-shaped cross-section. In the middle of the beam, the
tendons are positioned as low as possible in the cross-section to have a maximum internal
lever arm. At the end of the beam, the position of the anchorages is determined by
geometrical conditions. A complication arises because the tendons must be introduced in the
lower flange, and therefore must pass the stirrups that provide part of the shear force
resistance. Since the allowable curvature of prestressing tendons is relatively small (a high
radius), the shape of the stirrups has to be adapted over a considerable length. It is
questionable whether optimizing the position of the tendons is worth this effort. When in the
middle of the beam the tendons are positioned above each other instead of besides each other,
the internal lever arm is somewhat reduced, but construction is much easier.
It often occurs that one is not aware of the effect that the curvature of prestressing tendons
may have. Curved prestressing tendons do not only introduce compressive stresses but may
also introduce unintended tensile stresses. Figure 10.25 shows such a case. The figure shows
the bottom part of a box girder bridge. At the supports, the prestressing tendons have a
position high in the cross-section, whereas in the span, they are positioned in the bottom
flange to have a maximum internal lever arm. However, this implies that the tendons should
run from the top of the girder, through the webs, to the bottom flange. Then, they not only
have an upward curvature (to introduce the load balancing forces into the structure), but also
have a curvature in transverse direction (fig. 10.25a). This introduces forces in transverse
direction which can initiate longitudinal cracking in the bottom flange.
A similar effect occurs if the webs of a box girder are inclined (fig. 10.25b). Also in this case
the tendons have a (projected) curvature in the horizontal plane, which results in transverse
tension in the bottom flange [10.4].
a cracks b
bended curved tension cables

tension cables in sloping web
in
bottom plate
NT
tension force in bottom plate
Fig. 10.25 Possible cracking due to the transverse loading effect from curved prestressing
tendons
Special attention has to be given to the effect of tendon curvature in curved bridges. Figure
10.26 shows the top view of such a structure [10.1]. In this case, the prestressing tendons are,
as is usual in statically indeterminate structures, positioned high at the supports and low in
the spans. In this case altogether 12 prestressing tendons are applied. They are continuous
over the full length of the bridge. In each of the three webs of the 2-cell box girder, 4 bundled
April 2018 10-19

prestressing tendons are applied. They are stressed from both ends of the bridge. During the
prestressing operation of the last tendons four tendons broke out sideways over almost a full
span of the bridge (fig. 10.27). The cause of the damage was a combination of a relatively
high tendon curvature (small radius) and a small spacing between the high capacity tendons.
The damage would not have occurred, if the concrete cover would have been larger, and the
tendons would have been more uniformly distributed.
Fig. 10.26 Sideways breaking out of tendons during prestressing
Fig. 10.27 Compressive forces on the concrete cover caused by a curved profile of the
When a relatively flexible tendon duct is supported at a too high spacing, or when the
tendons have been pushed down during construction (e.g. by walking over the duct), an
unintended curvature might result. During prestressing, the tendon is straightened and forces
act on the wall of the duct. This can result in spalling of the concrete cover (fig. 10.27).
April 2018 10-20

Figure 10.28 shows damage that occurred during the construction of a lightweight concrete
bridge because of unintended curvatures.
Fig. 10.28 Push-off of concrete cover caused by unintended tendon curvatures in the deck
of a lightweight concrete bridge
10.4 Construction joints and joints between elements

In structures having large lengths, construction joints are often required. Such joints enable a
rational construction process, carried out in stages. However, sometimes such joints result in
a weak spot, because of the locally low tensile strength of the concrete.
Another problem concerns the hydration heat from hardening concrete. As a result of the
chemical process of hardening, the temperature of the concrete can considerably rise. When
the concrete starts to cool down, it implicitly shortens. New concrete is often cast against
older concrete that has hardened already. The shortening process of the new concrete is then
restrained by the older concrete. This results in tensile stresses and might cause cracking (fig.
10.29a) if the tensile stresses develop faster than the concrete tensile strength (fig. 10.29b).
By insulating the hardening concrete (fig. 10.29c), the cooling process is slowed down, so
that the tensile stresses develop slower (fig. 10.29d). The formation of temperature induced
cracks can then be prevented.
April 2018 10-21

The tensile strength of the concrete in a construction joint is mostly lower than that of the
adjacent concrete. Therefore, it is necessary to apply a well distributed reinforcement (ρs =
0,7 – 1,0%, bar spacing ≤ 100 mm) through the joint. In the youngest concrete additional
reinforcement is required because of stresses induced by restrained shrinkage.
a b
stress
fcct
c
T = 20° T = 70° c
cracks
old concrete fresh concrete time
c insulation
isolation
d
stress
fcct
c
old concrete fresh concrete time
Fig. 10.29 Stresses caused by the hydration heat development
If a construction joint is used for coupling prestressing tendons, another unfavorable effect
can occur. Long bridges are usually built in stages and couplers are mostly applied in joints
close the points of zero moment (fig. 10.30a). The deformations occurring in the following
stages are schematically shown in fig. 10.30b-e.
Fig. 10.30 Behaviour of a coupling joint during prestressing
April 2018 10-22

At first, the part that has hardened already is prestressed, see fig. 10.30c.
Subsequently, the prestressing tendons are coupled and a new section is cast, see fig. 10.30d.
After hardening of the concrete, also this new part is prestressed. The prestressing force,
introduced at the end of the next section, concentrates on the coupler in the construction joint
(fig. 10.30e). This, however, will be accompanied by deformations, which may result in
cracking at the ends of the construction joints.
The probability of cracking is higher, the more the couplers are concentrated at a small
number of positions. To prevent cracking, the couplers should be spread well over the cross-
section, so that their mutual distance is small. In the area between the couplers, reinforcement
for crack width control should be applied, having a length equal to the distance to the nearest
coupler, see fig. 10.31 [10.4].
k k
couple anchor
k surface reinforcement
k
k
k k k
k
web
3ap
primary crack crack width limitation
bottom plate reinforcement
6ap
prestressing cable
ap
ap a0
Fig. 10.31 Crack width control in the area close to a coupling joint [10.4]
10.5 Indirect support of structures

Indirect supports always require additional reinforcement. Especially in the case of large box
girder bridges, indirect supports can give rise to considerable tensile forces which require
careful detailing. Figure 10.32 shows that in statically indeterminate structures large
compressive forces can occur at the bottom of the webs near to the supports.
Schlaich and Schäfer [10.7] suggest two solutions, see fig. 10.32b and c. In the drawings the
dashed lines represent compression struts whereas the solid lines represent tensile ties.
April 2018 10-23

One solution, according to the principle shown in fig. 10.32c, has been worked out in fig.
10.33a.
a b
c
T1 T1
Fig. 10.32 Indirect support at an intermediate pier of a statically indeterminate box girder
bridge [10.7]
Figure 10.33a shows how the inclined compressive force D in the web of the box girder is
carried by a vertical tensile tie T1 and transferred to the top of the girder. Tensile tie T1 can be
constructed from straight prestressing tendons. As a next step, the force, which is now at the
top of the transverse internal diaphragm, has to be transmitted to the bearing. The
representation as a strut and tie model shows that this requires a second tensile tie T2. To this
aim, curved prestressing tendons can be applied, see fig. 10.33b and c.
T2
b
a
D D
T2
c
T1
T1
Fig. 10.33 a. Suspension (“hanging”) reinforcement

b. Representation of the flow of forces by a truss model
c. Transverse prestressing in the diaphragm [10.7]
10.6 Coping with deformations caused by prestressing

When designing a structure, not only the requirements for ULS and SLS should be regarded,
but the construction stage should be considered as well. In section 4.4.1 some possible
consequences of deformations from prestressing were already mentioned. With regard to this
April 2018 10-24

aspect, also the sequence of demoulding is important. As an example, a statically

indeterminate structure is shown, see fig. 10.34. Because of the weight of the concrete, which
flows during casting, the mould and the supporting structure below it deform. When, after
hardening of the concrete, the structure is prestressed, it will have an upward curvature from
the upward tendon pressure. The supporting structure, which is relatively flexible compared
with the concrete structure on top of it, is – however - only partially unloaded. So, the
selfweight of the structure is not yet fully mobilized as a load. If now the formwork is
removed at one span, the total selfweight in that span is activated. This results in bending
moments at the other spans. However, in those other spans the formwork still exerts an
upward pressure. The sum of the moments can be that high, that cracking occurs at the upper
part of the structure. Therefore, the formwork should be slowly removed, as much as possible
simultaneously at all spans.
removing the formwork from span 1
Fig. 10.34 Influence of a relatively flexible supporting formwork on the behaviour of a

statically indeterminate prestressed structure
It should also be planned carefully in which sequence the prestressing tendons are stressed.
The beam shown in fig. 10.35 is used as an example. If all tendons present in one web are
stressed, whereas the tendons in the other web stay passive (unstressed), the deformation at
the stressed side can give rise to tensile stresses which might lead to cracking. Also when at
both sides the lowest tendons are stressed first, the high eccentricity of this (temporary)
prestressing force can be the cause of cracking at the upper side of the structure. In such a
case it is advised to first stress the most centrically positioned tendons.
The sequence of prestressing should be documented on the basis of a calculation. This

information should be available in a prestressing protocol before the prestressing operation
starts.
April 2018 10-25

Fig. 10.35 Cracking by incorrect prestressing of a concrete beam
10.7 Design of a dapped-end beam

As an example of the detailing of the end of a beam, in which a number of effects described
in this chapter play a role, the design of a prestressed beam for the metro viaduct Rotterdam
is regarded. Figure 10.36 shows the main dimensions of the viaduct spans. The piers are
spaced at 50 m. In between the piers prestressed girders are provided, having a span of 33 m
(fig. 10.37).
All beams are prestressed using 4 tendons. The maximum support reaction force at SLS is
1550 kN. The effect of dynamic loading should be taken into account. In order to avoid
damage caused by fatigue loading, it was required that the beams, even at maximum load,
should be uncracked. This requirement was governing for the design of the beam ends.
Fig. 10.36 Metro viaduct Rotterdam [10.8]
April 2018 10-26

Fig. 10.37 Main girder of the metro viaduct
Dapped-end beams were used. They are mostly designed using strut and tie models. In
general, two solutions are possible, see fig. 10.38.
a b
T2
T1
Fig. 10.38 Two solutions for the beam end using strut and tie models
Often, these two basic solutions are combined. In the case considered, however, a number of
restrictions applied. The four prestressing tendons should be anchored at the beam ends. It
would have been a poor solution to anchor all four tendons either at the bottom or at the top
of the cross-section: the high capacity tendons require heavy anchors, which require a large
area. Furthermore, an eccentric introduction of the prestressing force would result in a creep
gradient, possibly resulting in cracking. Therefore, it was the most logical solution to anchor
two tendons at the top and two at the bottom of the cross-section. The vertical ties (T1 or T2)
can only be constructed from active prestressing steel because it was required that the
structure is uncracked at SLS. Because of the stress concentrations in the corner, the
prestressing steel should be provided as close as possible to this corner. For the control of the
tensile stresses in the corner, inclined prestressing tendons would give the best results. This is
an argument to apply model b from fig. 10.38. This would simultaneously solve another
problem, namely the sound detailing of the lower node of the truss model a from fig. 10.38.
This problem is illustrated in fig. 10.39.
T1
wrong
Fig. 10.39 Unfavorable position of prestressing anchors
April 2018 10-27

Therefore, only the solution shown in fig. 10.38b was acceptable. The corner was smoothly
curved to reduce stress concentrations. The principle of the solution chosen is shown in fig.
10.40.
Fig. 10.40 Principles of the solution
The stirrups mark A have a double function. First, they act as splitting reinforcement for the
two lower prestressing tendons. Second, they resist the force in the compressive diagonal
(strut) D. The required amount of stirrups mark B follows from the shear force design. The
stirrups mark C and the hairpins mark A confine the corbel part of the beam. The prestressed
vertical tendons are designed such that the prestressing force is higher than the tensile force
in SLS from the strut and tie analysis. Thanks to this prestressing force, no cracking in SLS
was observed, even not in the corner. Figure 10.41 shows an overview of all the
reinforcement applied.
stirrups Ø12 mm, 6 legs stirrups, 2 legs
Fig. 10.41 Overview of the reinforcement in the end region of the beam
The safety with regard to failure was determined as follows. A crack was assumed to start at
the corner and to run at an angle  (fig. 10.42). It was assumed that all reinforcing bars and
prestressing tendons that intersect the crack reach their yield strength. The depth of the
compression zone was calculated assuming that the concrete in the compression zone reaches
the strain limit. When using the condition of equilibrium of bending moments (M = 0), the
April 2018 10-28

corresponding maximum reaction Rd can be calculated. This calculation was repeated a

number of times for different values of the angle . The angle for which the support reaction
is the lowest, is governing in the design.
f
Vc f'c c
x N 'c RdVu
Nc
Ap.fpu
Ap.fpu
RVd,min
u,min
 .
Ap fpu
Ap.fpu
RVdd

Fig. 10.42 Determination of the ultimate load Rd
Because of the large number of identical beams to be used in this project, a test was carried
out. One of the full-scale beams was loaded to failure, see fig. 10.43. The test results
demonstrated that the design calculation was correct.
F6 F3 F1
a=305
Dywidag 18.6 2mm 0.5mm

1.5mm
2mm
4mm
7mm
11mm
50
10
Fig. 10.43 Full-scale test on a beam of the Metro viaduct (crack pattern and crack widths
in ULS; dimensions cm, crack widths in mm)
April 2018 10-29

10.8 Literature
10.1 Podolny, W.: “The cause of cracking in post-tensioned concrete bridge girders and
retrofit procedures”, Journal of the PCI. March-April 1985, pp. 83-139.
10.2 König, G., Maurer, R., Zichner, T.: „Spannbeton: Bewährung im Brückenbau“,
Springer Verlag, 1986.
10.3 Brakel, J., Doorn, L. van, Pol, F.J.M. van de: „Schadegevallen bij brugconstructies in
Nederland“, Stuvo rapport 101, December 1985.
10.4 Leonhardt, F.: “Rißschäden an Betonbrücken – Ursachen und Abhilfe”, Beton- und
Stahlbetonbau, 2/1979, pp. 36-44.
10.5 Kupfer, H.: „Bemessung von Spannbetonbauteilen einschließlich teilweiser

Vorspannung“, Betonkalender.
10.6. Uijl, J.A. den: „Verbundverhalten von Spanndraht – Litzen im Zusammenhang mit
Rißbildung im Eintragungsbereich“ Betonwerk + Fertigteiltechnik, Heft 1, 1985, pp.
28-36.
10.7 Schlaich, J., Schäfer, K., Jennewein, M.: „Toward a consistent design of structural
concrete“, Journal of the Prestressed Concrete Institute, May-June 1987, pp. 74-147.
10.8 Blokland, P.: Loading tests on a full-size suspended beam and a model of this beam
for a Metro viaduct at Rotterdam”, CUR-Report 40, 1969.
10.9 Uijl, J.A.: “Background of a CEB-FIP MC proposal on anchorage and transverse

tensile actions in the anchorage zone of prestressed concrete members”.
April 2018 10-30

Chapter 11 – Partially prestressed concrete
11. Partial prestressing
11.1 Introduction
Serviceability limit state cracks in precast concrete are not harmful from the point of view
of durability provided that their width is small. This finding is of great importance for the
design in structural concrete. It implies that it is possible to combine passive and active
(prestressed) reinforcement. For each application the most suited combination can be
chosen.
At the beginning of the application of prestressed concrete, in most countries only fully
prestressed concrete was allowed. The requirement no tension in the concrete could
hardly be accomplished as it required prestressing in three directions. The result was that
full prestressing was replaced by prescribing that no tensile stresses in the main load
bearing direction were allowed. In this direction, however, principal tensile stresses can
be caused by shear and torsion as well. This alternative definition of full prestressing was
therefore not very consequent: cracks caused by shear and torsion are more dangerous
than well distributed, fine, bending cracks [11.6].
This resulted in the so-called “limited prestressing” concept, where small tensile stresses
were allowed. This method of prestressing has resulted in economically sound structures,
which meet high requirements with regard to durability.
Limited prestressing, however, does not exclude all inconsistencies. There is a general
misunderstanding that the definition applies only to the main load bearing direction, and
that it is therefore not necessary to apply transverse prestressing. This in fact implies that
the structure is prestressed in the main load bearing direction and is reinforced in the
transverse direction. Implicitly, cracking in transverse direction is accepted. The result is
that, as an example, the relatively slender box girder from fig. 11.1 is not prestressed in
vertical and transverse direction. As a result, however, longitudinal cracks can occur in
the webs of the box girder, and especially in the dynamically loaded bridge deck. Under
certain conditions, the cracks in the bridge deck above the webs can penetrate to the level
of the longitudinal prestressing steel. The cracks from loading in the transverse direction
are longitudinal cracks that run parallel to the longitudinal prestressing. These cracks can
coincide over a considerable length [11.6].
April 2018 11-1

Fig. 11.1 (a) Box girder without transverse prestressing, (b) bending moment lines for
webs and top flange
Such cracks are more dangerous from the point of view of durability than the well-
distributed fine bending cracks, which one tries to avoid by prestressing. Furthermore, for
such structures it is not economic to apply prestressing to avoid cracking caused by
transverse moments. Because of the governing traffic load, the transverse prestressing
tendons should be applied almost centrically, because otherwise in the unloaded situation,
tensile stresses might occur in the pre-tensioned compression zone. Furthermore, because
of the small depth of the flange, there is hardly any possibility to use a draped tendon
profile. Full and limited prestressing are therefore not suited, since they would require an
uneconomically large amount of prestressing. The best solution is to apply a well-
distributed reinforcement, combined with a small number of prestressing tendons which
assist in controlling crack widths and deformations. The efficiency of the prestressing
then mainly results from introducing an axial compressive force. This force strongly
reduces the steel stress increase at the onset of cracking and, as a result, limits the
transmission length (see Chapter 9). Moreover, the compressive force might limit the
height of the effective tension area. This smart solution is denoted as “partial
prestressing”, because in SLS fine cracks are accepted to occur. These cracks will only
open at unusually high traffic loads, and will otherwise be (almost) closed.
There are many situations where partial prestressing is the best option, for example slabs
in office buildings. Flat slabs having a span larger than about 6 m can hardly be
constructed in reinforced concrete in an economic way. Full or limited prestressing has
the disadvantage that the governing design load hardly or even never at all occurs. As a
result, under quasi-permanent loading conditions (2 qq, where 2 = 0,3 for offices and
housing; see EN 1990 table A1.1 & NL National Annex), at the bottom of the slabs a
higher compressive stress occurs than at the top (fig. 11.2). Since creep deformation
depends on the magnitude of the compressive stresses, the slab has an increasing upward
deflection (camber) which can result in damage to the partitioning walls on the slabs.
April 2018 11-2

c c
due to Md.l. + Fpi + MQ due to Md.l. + Fpi + M

MEqpcondition)
(operating + Pm MEd +loading)
(ultimate Pm

Fig. 11.2 Increasing upward deflection caused by creep of a limited prestressed

concrete slab
If one only would have the choice between reinforced concrete on one hand and full or
limited prestressing on the other, one would always have deal with the negative effects of
deflections developing over time: in case of reinforced concrete an increasing downward
deflection, for prestressed concrete an increasing upward deflection.
However, most damage can be avoided by choosing the right combination between
reinforced and prestressed concrete (the right degree of prestressing, see section 11.2).
With regard to corrosion, there are no arguments that hamper the application of partially
prestressed concrete in flat slabs. In Switzerland, this way of prestressing has developed
to the standard one. In the Netherlands, the application of partially prestressed concrete
was allowed after the introduction of NEN 6720 in 1990.
11.2 Definition of partial prestressing
In fig. 11.3, the stresses caused by the maximum load q = qg + qq in SLS are shown for
full, limited and partial prestressing.
In the last case, the stress s in the reinforcing steel, or the increase of the stress in the
prestressing steel p, should be limited to ensure that durability requirements are met.
Depending on the bar spacing, EN 1992-1-1 table 7.3N allows steel stresses starting at
160 N/mm2. Crack widths can be limited to 0,4, 0,3 or 0,2 mm, depending on the
exposure class(es).
It is noted that the Dutch National Annex to EN 1992-1-1 reduces the maximum crack
widths by 0,1 mm in case prestressing steel is applied.
April 2018 11-3

Mg+q no tensile stresses

are allowed
Pm
limitation of the increase in stress in

reinforcing and prestressing steel
Fig. 11.3 Stress distribution as a result of the total load qg + qq for full, limited and
partial prestressing
Partially prestressed concrete can be classified by the so-called “degree of prestressing”.

This degree can be defined in several ways. Bachmann ([11.1] and [11.2]) used the
following definition:
M dec
K (11.1)
Mg  Mq
where
Mdec is the so-called decompression moment. This is the moment for which the
stress in the outer fiber of the tension zone is 0.
Mg + Mq is the maximum moment in SLS from permanent and variable loads.
The degree of prestressing indicates the part of the total load the structure can resist
without having tensile stresses at the tension side. K = 0 refers to a structure without
prestressing, whereas K > 1 refers to fully prestressed concrete. The most important
advantage of prestressed structures compared to structures without prestressing, is their
more favorable behaviour in SLS with regard to cracking and deflection. In ULS, the
behaviour of a partially prestressed structure is basically the same as for a reinforced
concrete structure. Therefore, an effective definition of the degree of prestressing should
be linked to the effect of prestressing in SLS. The definition of the degree of prestressing
given in eq. (11.1) satisfies this requirement.
Another possible definition is:
Ap f pd
 (11.2)
Ap f pd  As f yd
April 2018 11-4

where:
Ap is the cross-sectional area of the prestressing steel;
As is the cross-sectional area of the reinforcing steel;
fpd is the design value of the tensile strength of the prestressing steel;
fyd is the design value of the tensile strength of the reinforcing steel.
The ratio  indicates which part of the resistance can be attributed to the prestressing
steel. This definition is relatively simple, but it does not provide any insight into the
effect of prestressing in SLS.
The influence of the degree of prestressing is demonstrated on the basis of an example

[11.2]. It concerns a statically determinate slab, subjected to a maximum SLS bending
moment Mg + Mq = 124 kNm. At first, full prestressing is regarded (K = 1). Subsequently,
the cross-sectional area Ap of the prestressing steel, and, consequently, the degree of
prestressing K, is reduced. The cross-sectional area As of the reinforcing steel is increased
if this is required to keep the resistance (ULS) at the required level. A minimum
reinforcement ratio ρs = 0,3% for the passive reinforcement was always applied, even if
this was not required to provide sufficient resistance.
The diagrams from fig. 11.4 show the calculated safety against failure, the total cross-
sectional area of the steel A = Ap + As , the stress in the reinforcing steel s and the
increase of the stress in the prestressing steel p. The diagrams can be used to draw
some important conclusions:
 For high degrees of prestressing, the safety factor (i.e. the ratio between the bending
moment from the characteristic loads and the actual bending moment resistance) is
higher than strictly required. (In Switzerland, where this comparison was made, the
margin was  = 1,75).
 The required cross-sectional area of the reinforcement reaches, in this specific case,
its minimum value for K = 0,6.
 At high degrees of prestressing (K > 0,7) the stress in the reinforcing steel s and the
increase of the stress in the prestressing steel p are relatively low. For intermediate
degrees of prestressing (0,4 < K < 0,7), the stresses are still considerably lower than in
traditional reinforced structures.
April 2018 11-5

1000 mm
220
270
300
Md.l.+v.l.
M g+M = 124
=q124 kNm kNm
Ap As
2
 failure
collapsesafety
safety
1
0
mm2 300
minimal section reinforcing steel
200 Ap + applied reinforcing
A As cross-sectional
section area of steel
100 Ap,As, Ap+As
As
Ap
0
N/mm2 300
s stress increase in prestressing
200 steel and stress in reinforcing steel
stresses in reinforcing steel s
sp
stress increase in prestressing
100 p
steel p
0
0 0.2 0.4 0.6 0.8 1.0
grade of prestressing k
degree of prestressing K
Fig. 11.4 Influence of the degree of prestressing on some important parameters

(Bachmann [11.2])
Often, it will be most economic to choose the values for As and Ap such that As
corresponds with the minimum reinforcement and determine Ap such that the safety ratio
is not unnecessarily high.
Next, results of an experiment carried out at ETH Lausanne [11.7] are discussed. The
slabs tested are statically indeterminate, see fig. 11.5. The figure presents the side view of
one span and the cross-sections at midspan and at the support. In the calculation a linear-
elastic bending moment distribution was assumed. The combination of reinforcing steel
and prestressing steel in the section subjected to the highest moment was chosen such that
the moment resistance is the same in all cases researched. In the research, the definition
of the degree of prestressing according to eq. (11.2) was used, with  varying between 0
(slab B4) and 0,76 (slab B2).
April 2018 11-6

cross-section A-A cross-section B-B
Fig. 11.5 Series of tests on prestressed slabs at ETH Lausanne (dimensions cm) [11.7]
In fig. 11.6, the maximum deflection is shown as a function of the load applied. The
influence of the degree of prestressing is clear: already at a relatively low degree of
prestressing (mean axial concrete compressive stress md = 0,86 N/mm2; slab B3;  =
0,29), the deflection in SLS is 40% smaller than for the reinforced reference element. If
the prestressing is twice as high, the reduction is about 65%. A similar reduction is found
for the maximum crack widths at the intermediate support (fig. 11.7) and in the span. The
tests confirm that even a relatively low prestressing level can substantially improve the
behaviour in SLS. The vertical dashed lines in fig. 11.6 and 11.7 refer to the SLS / ULS
load ratio of 1 / 1,75 used in Switzerland at that time.
max [mm]
25
20
15
B4
10
B3
B2
0 0.2 0.4 0.6 0.8 1.0 1.2

qd / qu
q / qd
Fig. 11.6 Maximum deflection as a function of q / qd
April 2018 11-7

wmax [mm]
1.0
0.8
0.6 wmax
B4
0.4
B3 B2
0.2
0 0.2 0.4 0.6 0.8 1.0 1.2

qd / qu
q / qd
Fig. 11.7 Maximum crack width at the intermediate support as a function of q / qd
11.3 Applications of partial prestressing
In this chapter a number of applications of partially prestressed concrete will be

discussed. The first two examples are from reference [11.3].
The first example is a statically indeterminate box girder bridge, having a main span of
41,6 m and two end spans of 32 m each. The bridge is prestressed by 12/12.9 VSL
tendons at an effective prestressing force (1200 - 1350 kN). To have a fully prestressed
structure, 18 tendons would be required; 9 in each web. Figure 11.8 shows the positions
of these tendons in the cross-section at a mid support and at the end span. It is difficult to
position all the tendons in the cross-section. When aiming at the largest possible internal
lever arm, a substantial number of tendons must be in the flanges. The result is that not
only vertical but also horizontal curvatures of the prestressing tendons are required.
Horizontal curvatures should be avoided as much as possible, see fig. 10.25. If, however,
a configuration with all tendons in the webs is chosen, 20 tendons are required because of
the reduced internal lever arm. Furthermore, the high concentration of tendons in the web
could result in problems during casting (sieve effect from the ducts). If the design of the
bridge is based on partially prestressed concrete, only 12 cables are required. The degree
of prestressing then is K = 0,68. To obtain the required resistance, additional reinforcing
steel is only required at midspan. The passive reinforcement is effective in reducing the
increase of the steel stress at cracking, and, as a consequence, the crack width.
The transition from full to partial prestressing offers a number of advantages, such as a
more practical layout of the prestressing tendons and a ULS resistance that is not
unnecessarily high. Moreover, cracks only occur at midspan at full SLS loading and close
again at unloading.
April 2018 11-8

prestressing cable not not used as k = 0.68

1 prestressing tendons
not used if K = 0,68
Fig. 11.8 Location of the prestressing tendons in the cross-section of a box girder at the
mid support (1) and at midspan (2), for K = 1 (black and gray tendons) and K
= 0,68 (black tendons only) [11.3]
The second example is a submerged tunnel. The minimum load caused by a 22 m water
column on the roof is 220 kN/m2.
Figure 11.9 shows a cross-section of the tunnel. Such tunnels are mostly constructed in
segments having a length of 100-200 m. The segments are built in a temporary dock that
is inundated after completion of the segments. The segments are provided with temporary
end walls to make them float. They are transported to their destination, where they are
submerged to their final position. This causes large differences in the load on the tunnel
roof in the various stages (construction, transport, submerging, final use). Full or limited
prestressing is not an option since there is no load in the construction and the transport
stage: the tendons would give a high upward load, whereas the vertical load is not yet
applied. Constructing in reinforced concrete is not an option since it would require very
high amounts of steel bars of large diameters that would have to be placed in several
layers. In such conditions, it is very difficult to meet the crack width criteria.
The best solution is to apply partial prestressing. In this specific case, the best solution is
obtained for K = 0,73 at the intermediate support (ρs = 0,27%, ρp = 0,27%).
The smallest concrete compressive stress at the mid support from prestressing and the
force from transverse water pressure is:
 for t = 0 (just after prestressing): c = - 3,3 N/mm2
 for t =  (including losses): c = - 4,1 N/mm2
The calculated mean crack width:

 directly after prestressing: wm = 0,11 mm (inside)
April 2018 11-9

 at 22 m water pressure: wm = 0,16 mm (outside)

 at 26 m water pressure: wm = 0,39 mm (outside) (1 : 104 yr)
Fig. 11.9 Cross-section of a submerged tunnel (dimensions mm) [11.3]

concrete on after prestressing Øk 25-200 Øk 16-200
B.B.R.V. - UR - 500
stirrup
Ø 16
Øk 25-200 Øk 16-200
Fig. 11.10 Prestressing and reinforcing steel in tunnel roof [11.3]
The following examples refer to a number of applications in Germany [11.5].
The first example is a frame of the Hans Martin Schleyer hall in Stuttgart, see fig. 11.11.
The main bearing structure of this hall consists of a number of prestressed concrete
frames. At its end, a frame supports a truss of the roof structure, see fig. 11.11. The
picture shows the position of the prestressing steel in the frame and gives the degree of
prestressing in some important cross-sections. It was obvious to prestress the
cantilevering frame with the high load at its end, especially because its top part is
permanently exposed to outside weather conditions. During design, it turned out that full
or limited prestressing was not very efficient. Applying partial prestressing was
April 2018 11-10

stimulated by the possibility to prestress all the tendons at once. By reducing the number
of tendons and by adding reinforcing steel, all prestressing anchorages could be placed in
the end cross-section of a frame. The required degree of prestressing followed from the
requirement that the frame should be fully prestressed under permanent loads. The ratio
between the permanent loads and the total load was about 2/3. This resulted in the degrees
of prestressing shown in fig. 11.11.
Fig. 11.11 Hans Martin Schleyer Hall: partially prestressed frames with position of
tendons and degrees of prestressing
The second example concerns a viaduct in the highway A7 Ulm – Würzburg [11.5]. The
structure was designed in partially prestressed concrete. The viaduct is shown in fig.
11.12: a skew beam grid bridge having a span of 22,5 m and to be designed for a traffic
class 60 (heaviest vehicle = 600 kN). The main aim of the study was to determine the
most economic degree of prestressing.
In the transverse direction, the optimum degree of prestressing was K = 0,53. This
implied that full prestressing is available up to a load equal to the permanent loads plus
39% of the traffic load. In the longitudinal direction, the most economic degree of
prestressing was K = 0,60.
April 2018 11-11

Fig. 11.12 Viaduct over a highway, longitudinal cross-section [11.5]
11.4 Economic degree of prestressing
Using the method of partial prestressing, structures can be designed with excellent
durability (for instance, by not allowing cracks to occur at an intermediate support or by
limiting crack widths) and minimum creep deformation. Furthermore, economic solutions
are possible, providing a resistance that does not (unnecessarily) exceed the specified
value.
When determining the building costs, all aspects, such as the costs of the materials
(concrete, reinforcing and prestressing steel), the required temporary structures
(formwork, scaffolding), transportation, labour et cetera, have to be taken into account.
Therefore, Kupfer and Schulz [11.4] determined the most economic degree of
prestressing of the structures shown in fig. 11.13. The costs are presented as a function of
the degree of prestressing (fig. 11.14). For all types of structures regarded, the optimum is
found for degrees of prestressing in the range K = 0,3 – 0,7. Compared with structures
from reinforced concrete, savings are about 10%.
April 2018 11-12

30600
10 10 10
16 37.5 20
165
115
20
17.5
3x8
16 16
40 31000 40
15
60
54
5 15
165
115
35
3x18 18 bmF = 2.35m
T - girder continous gider
continuous girder
30600
4
22
5
I-girder slab floor
Fig. 11.13 Structures for which the total costs are calculated as a function of the degree
of prestressing K [11.4]
relative construction costs

110
-profile
100
-profile
continuous
continous
90 girder
girder
slab floor
0 0.2 0.4 0.6 0.8 1.0 1.2

k = grade of prestressing
degree of prestressing K
Fig. 11.14 Relation between total costs and the degree of prestressing for the types of
structures from fig. 11.13
April 2018 11-13

11.5 Design method
An important task for a structural designer is to calculate the required reinforcing and
prestressing steel such that all SLS and ULS requirements are met. The first step,
however, is to estimate the dimensions of the structure. One should realise that
“designing a structure” is in general not a linear straightforward process, but requires a
number of iterations. In this section, only the major steps are presented.
Step 1: Choice of the prestressing load.

It is a sound approach to prestress such that the most frequently occurring “mean” load is
balanced by prestressing. At the combination of prestressing and load, the cross-section
then is subjected to an axial load (from prestressing) only. The negative effect of creep is
then limited, because the structure only axially shortens, but does not develop a
deflection or camber over time.
If, for instance, the permanent load and a part of the variable load, qg + 1qq, are
balanced, then - because of the effect of the axial prestressing force - cracking will occur
at a load qg + 2qq, where 2 > 1. In other words: under the most frequent load qg +
1qq, the deflection is zero and cracking occurs only at a relatively high load, which is
not very frequent. Several SLS aspects, such as deformations, cracking and durability
(cracks open only temporarily at non-frequent loads and close at unloading) profit from
this solution. The designer is free to choose the most appropriate degree of prestressing.
Step 2: Determination of the cross-sectional area of the prestressing steel.

In general, a curved layout of the prestressing tendons is chosen. The upward “balancing”
load is:
8 Pm, f
qp  .
l2
The working prestressing force is estimated at first, for instance 0,8 Pmo. On the basis of
the condition from step 1, the cross-sectional area Ap of the prestressing steel is
calculated.
Step 3: Determination of the required amount of reinforcing steel.

The basic condition with regard to the amount of reinforcing steel required is the
condition that the design bending moment can be resisted. The design bending moment
is:
M Ed   g M g   q M q   p M p
April 2018 11-14

where:
Mg is the bending moment from selfweight and static loads;

Mq is the bending moment from variable loads;
Mp is the bending moment from prestressing loads.
To resist MEd, the following actions are present:
Pm∞ is the axial prestressing force;

Ap (σpu - pm,∞) is the additional tensile capacity of the prestressing steel (pm,∞ = stress
in the prestressing steel to generate the prestressing forces);
As fyd is the design yield force of the reinforcing steel.
The last component is still open. In each cross-section, the amount of reinforcing steel
applied must ensure that the requirements with regard to the resistance (ULS) are met
(also see sections 7.2 and 7.3).
Step 4: Detailing of the reinforcement

Detailing of the reinforcement is commonly the last step in the design process. In Chapter
9 on crack width control it is shown that the smallest crack width is obtained if the
reinforcement is applied as small diameter bars at a small bar spacing. Since the
reinforcement is close to the surface of the concrete, this is favourable to the quality of
the structure.
Especially in the case of high degrees of prestressing, a certain amount of minimum

reinforcement has to be applied, even if it is not required for resistance (ULS). Bachmann
[11.2] advised to use at least:
- for pure bending: ρmin = 0,3 – 0,4%
- for pure tension: ρmin = 0,6 – 0,8%
It should be noted that the flanges of box girders are almost in pure tension, see fig.
11.16.
Step 5: Control of the crack width

When the degree of prestressing has been chosen, the crack widths must be checked. If,
for instance, the prestressing is such that no cracking occurs at the most frequent load, the
crack width might play a minor role. A crack width control is, however, always required.
On the basis of the Swiss Code SIA 162 (1983) a number of design aids have been
derived. They enable a fast control of the crack width (fig. 11.15).
April 2018 11-15

350
[N/mm2]
permisseble stresses
300
pure bending
250 ae/t = 1.0
200
s en p
150
pure tension
100 ae/t = 0 ae/t = 0.5
50
0
0 50 100 150 200 250 300
bar distance s [mm]
Fig. 11.15 Allowed maximum stress in the reinforcement as a function of the bar
spacing s, for which wm  0,15 mm [11.2]
On the horizontal axis of fig. 11.15, the bar spacing s is shown (bonded prestressing
tendons are included in this spacing) and on the vertical axis the maximum steel stress in
the reinforcing steel. The diagram has been derived for a mean crack width of 0,15 mm.
The upper curve is valid for a high stress gradient over the depth of the cross-section
(such as in massive slabs). The lower curve applies when the stress gradient is 0, which
implies pure tension. The stress gradient is defined by the ratio ae / t (fig. 11.16), where ae
is the height of the effective tensile member (compare to fig. 9.14) and t is the depth of
the concrete tensile zone, assuming that the member is uncracked. The ratio ae / t should
be less than 1 (top part of fig. 11.16).
f
heff ae=t
s s
t
ae
maximum ae = 0.2 m
Fig. 11.16 Effective depth of the cross-section according to [11.2]
11.6 Working example - three-span bridge

This working example is an extended version of an example presented by Bachmann
[11.2] and concerns a three-span highway bridge. The span length and the cross-sections
are shown in fig. 11.17. Figure 11.18 shows the governing bending moment lines for
April 2018 11-16

permanent loads only and for permanents loads plus variable loading. The moment lines
are calculated assuming uncracked cross-sections. The cross-sections A, B and C will be
designed.
The design by Bachmann is based on the Swiss code SIA 162 from 1975, using an
overall load factor of 1,75 for both permanent and variable loading. No material factor is
used. Nowadays, codes do not use one overall load factor anymore, but use partial factors
for loads ("effect") and strength ("resistance"). Moreover, different load factors are used
for permanent and variable loading. Nevertheless, Bachmann's approach will be followed
since the example is meant to illustrate the design process.
This working example deals with the main items of partial prestressing. Therefore, only
preliminary design calculations are presented. More detailed analyses must be carried out
to verify whether all requirements are met. Since these analyses are based on the same
principles as the preliminary design, they will not be presented.
midspan Ac = 3,61 m2 support Ac = 4,45 m2

Ic = 1,08 m2 Ic = 2,29 m2
Fig. 11.17 View and cross-sections of the bridge (dimensions m; c = centroidal axis; kt =
top of kern area; kb = bottom of kern area) [11.2]
Loads:
volumetric weight concrete: 25 kN/m3
static loads: 2,4 kN/m2
variable loading (traffic): 7,0 kN/m2
April 2018 11-17

Material properties:
concrete C35/45; Ec = 37 · 103 N/mm2
prestressing steel Y1770 (fpk = 1770 N/mm2; fp0,1k = 1570 N/mm2 )
reinforcing steel B500
[kNm]
permanent loads + variable load
permanent loads
Fig. 11.18 Bending moment lines (unit kNm) [11.2]
Step 1: Choice of the prestressing load.
The decompression bending moment is chosen such that the governing cross-sections are
uncracked at permanent loading only. The decompression moments MD follow from the
bending moment lines in fig. 11.18:
MD,A = + 9860 kNm

MD,B = -20588 kNm
MD,C = + 5194 kNm
Step 2: Determination of the cross-sectional area of the prestressing steel.
To estimate the required amount of prestressing steel, the prestressing is first applied as
equivalent loads to the structure. The profile of the prestressing is estimated, see fig.
11.19.
In sections A and C, the tendons are about 0,25 m from the bottom fibre; in section B
about 0,15 m from the top fibre.
It is assumed that the tendons exert an upward load only. (In a detailed analysis, the exact
tendon profile and both the upward and the downward curvatures are used). The drape of
the tendons fA = 1,37 m in both end spans; fC = 1,50 m in the midspan.
The working prestressing force is assumed to be constant along the bridge axis.
Upward prestressing loads:

8 Pm, f A
at both end spans: qp   0, 0085 Pm,
l12
April 2018 11-18

8 Pm, f C
at midspan: qp   0, 0068 Pm,
l22
Pm, Pm,
qP = 0,0085 Pm, qP = 0,0068 Pm, qP = 0,0085 Pm,
0,88 Pm, 0,88 Pm,

0,39 Pm,
1,11 Pm, 1,11 Pm,

Fig. 11.19 Schematised tendon profile and loads and bending moments from
prestressing
The condition that no tensile stresses are allowed due to permanent loads results in the
following requirements.
Section A:
Pm, 0,88 Pm, zb M D,A zb

   0
Ac Ic Ic
Pm, 0,88 Pm, 1, 43 9,86 1, 43
   0
3, 61 1, 08 1, 08
Pm,  9, 05 MN
April 2018 11-19

Section B:
Pm, 0,88 Pm, zt M D,B zt

   0
Ac Ic Ic
Pm, 1,11Pm,  0, 72 20,58  0, 72
   0
4, 45 2, 29 2, 29
Pm,  11, 28 MN
Section C:
Pm, 0,39 Pm, zb M D,C zb

   0
Ac Ic Ic
Pm, 0,39 Pm, 1, 43 5,194 1, 43
   0
3, 61 1, 08 1, 08
Pm,  8, 67 MN
The prestress losses caused by shrinkage and creep of the concrete, relaxation of the
prestressing steel and tendon friction are estimated to be 20%. The initial prestressing
force required then is:
Pm,
Pm,0 
0,8
The condition at section B is governing. The prestressing force required:
11, 28
Pm,0   14,1 MN
0,8
It is now assumed that the initial prestressing steel stress allowed follows from EN 1992-
1-1 cl. 5.10.3:
min. (0,75 fpk = 1328 N/mm2; 0,85 fp0,1k = 1335 N/mm2 ) = 1328 N/mm2
The minimum required amount of prestressing steel:
1410 4
Ap  10  10, 62 103 mm 2
1328
When using tendons consisting of 12 seven-wire strands diameter 16,0 mm, Ap =

12 · 150 mm2 = 1800 mm2 per tendon. When applying 4 tendons per web, the total cross-
sectional area of the prestressing steel is Ap = 8 · 1800 mm2 = 14400 mm2.
At σpm,∞ = 0,8 · 1328 = 1062 N/mm2, the working prestressing force is Pm,∞ = 15,29 MN.
The intial prestressing force is Pm,0 = 19,12 MN.
April 2018 11-20

Step 3: Determination of the required amount of reinforcing steel.
The required amount of prestressing steel can be calculated if the required bending
moment resistance is known. Bending moments from temperature and differential
settlements (20 mm) are also taken into account and have a load factor 1,0. The moments
from these imposed deformations are:
MA = 1,4 MNm
MB = -2,6 MNm
MC = 1,1 MNm
Also the bending moments from the prestressing loads have a load factor 1,0. The
following bending moments are obtained:
section Mg+q [MNm] Mp,∞ [MNm] Mtotal [MNm]

A 1,75 · 18,81 + 1,4 -0,88 · 15,29 +20,86
B 1,75 · (-34,93) - 2,6 +1,11 · 15,29 -46,76
C 1,75 · 13,02 + 1,1 -0,39 · 15,29 +17,92
The axial prestressing force and the additional capacity of the prestressing steel must be
taken into account in a cross-sectional analysis.
The friction losses in section A are now assumed to be about zero because of some
overstressing of the steel to compensate for these losses, as well as for the wedge set at
the anchor. The total loss, including time-dependent losses, was assumed to be 20%. If it
is assumed that the friction losses are on average 10% for the bridge, the time-dependent
losses are on average 10% too.
This assumption has impact on the additional capacity of the prestressing steel. It is now
assumed that the friction loss is 0%, 7% and 14% in cross-sections A, B and C,
respectively. The time-dependent losses are estimated to be 10% in all cross-sections.
The working prestressing stress then is:
section A: σpm,∞ = (1,0 - 0) · 0,9 · 1328 = 1195 N/mm2

section B: σpm,∞ = (1,0 - 0,07) · 0,9 · 1328 = 1112 N/mm2
section C: σpm,∞ = (1,0 - 0,14) · 0,9 · 1328 = 1028 N/mm2
The additional capacity of the prestressing steel is given in the following scheme. It is
assumed that the stress can increase from the working stress to the stress at the 0,1%
strain limit, i.e. 1570 N/mm2. (A detailed analysis might demonstrate that this is a
conservative assumption).
April 2018 11-21

section σpm,∞ [N/mm2] fp0,1k - σpm,∞ [N/mm2]

A 1195 375
B 1112 458
C 1028 542
Figure 11.20 presents the forces in cross-section A. It is assumed that the compressive
force in the upper flange is at 0,1 m from the top fibre. The internal lever arms of the
prestressing and reinforcing steel forces can now be calculated.
loads resistance
20,86
Nc
17,21
Ap
Ap·375
As·500
As
Fig. 11.20 Calculation of the reinforcement in section A
The required amount of reinforcing steel follows from bending moment equilibrium
around the point where the resulting concrete compressive force is:
M total  20,86  0,37 Pm, ,A  1, 60   500 As  375 Ap 
where:
Pm,∞,A = 14400 · 1195 = 17,21 · 106 N = 17,21 MN and Ap = 14400 mm2 = 14,4 · 10-3 m2.
The lever arm of the axial prestressing force is 0,37 m; it is about 1,6 m for the resulting
force from the reinforcing steel stress and the prestressing steel stress increase, see fig.
11.20.
Result:
As = 0,0073 m2 = 7,3 · 103 mm2.
Similar calculations are carried out for sections B and C.
Section B:
M total  46, 76  1,18  0,1 Pm, ,B  1, 65   500 As  458 Ap 
where:
April 2018 11-22

Pm,∞,B = 14400 · 1112 = 16,01 · 106 N = 16,01 MN and Ap = 14400 mm2 = 14,4 · 10-3 m2.
Result:
As = 0,0225 m2 = 22,5 · 103 mm2.
Section C:
M total  17,92  0,37 Pm, ,C  1, 60   500 As  542 Ap 
where:
Pm,∞,C = 14400 · 1028 = 14,80 · 106 N = 14,80 MN and Ap = 14400 mm2 = 14,4 · 10-3 m2.
Result:
As < 0 m2 ; so that minimum reinforcement applies.
Step 4: Detailing of the reinforcement
Figure 11.21 shows the reinforcement chosen. Small bar diameters are used. They are
distributed along the sides of the cross-section.
In section A, the 24 Ø 20 mm (As = 7540 mm2) are located near the outer fibre and they
surround the tendons (12 bars at the bottom part of each web). In the rest of the webs,
minimum reinforcement is applied to control crack widths.
In section C, minimum reinforcement can be used in the entire web, down to the outer
fibre.
The reinforcement in section B extends over the full flange width. Codes, such as EN
1992-1-1 for buildings (cl. 5.3.2.1) and EN 1992-2 for bridges, give design rules to
calculate the effective width of a flange. All reinforcement applied within the effective
width can be assumed to be effective in resisting the external load.
It is decided to concentrate almost all the required reinforcement close to the webs. Bars
Ø 20 mm are applied near the webs (66 Ø 20 mm: As = 20734 mm2, which is almost
equal to the total amount required, namely As = 22,5·103 mm2). The remaining parts of
the upper flange are provided with minimum reinforcement.
April 2018 11-23

required
12 Ø20 minimum
Ø12-150
4 * 12/16,0
4 * 12/16,0
cross-section A-A cross-section C-C
minimum required minimum
4 * 12/16,0
cross-section B-B
Fig. 11.21 Detailing of the reinforcement in the three cross-sections analysed
The minimum reinforcement chosen, namely Ø12 - 150 mm, provides a minimum
reinforcement ratio of 0,38% in the webs and 0,75% in the flanges.
The previous calculations refer to ULS; SLS is checked later. This might imply that
additional reinforcement is required to meet crack width requirements, see step 5.
Step 5: Control of the crack width
It is assumed that the crack width must be restricted to wmax = 0,15 mm. To control if this
requirement is met, the stress in the reinforcing steel is calculated.
April 2018 11-24

In section B the following forces apply, see fig. 11.22:

 A bending moment from permanent loads and variable loading (traffic) of
34,93 MNm, see fig. 11.18.
 A bending moment from an imposed temperature gradient and differential settlements
(2,6 MNm).
 A bending moment from the upward curvatures in the prestressing tendons (1,11 Pm,∞
= 1,11 · 15,29 = 16,97 MNm).
 An axial prestressing force Pm,∞,B = 14400 · 1112 = 16,01 · 106 N = 16,01 MN.
(As+ξ1Ap)σs
εs (As+ξ1Ap)σs
34,93
0,72
16,01 16,01
2,6
20,56
1,18
1,08
16,97
Nc εc Ec Aflange εc
Fig. 11.22 Calculation of the forces and moments in section B at SLS loading (forces in
MN, moments in MNm)
The bond factor of the prestressing steel is calculated using EN 1992-1-1 table 6.2 and
clauses 7.3.2 and 6.8.2:
Øs Øs 20
1      0,5   0,38
Øp 1,6 Ap 1,6  12 150
The concrete compressive force is assumed to be at the centre of the flange. Two
equilibrium conditions hold:
M  0:
  A   A  E 1, 65  P
s s 1 p s m, 1, 08  20,56
H  0:
 A  A E
s s 1 p s  Pm,   c Ac,flange Ec
where Ac,flange = 5000 · 200 mm2 = 1,0 m2.
It is assumed that Ep = Es = 200·103 N/mm2 = 200·103 MN/m2 and Ec = 37·103 MN/m2.
April 2018 11-25

From the bending moment equilibrium it follows:

 s  20734  0,38 14400  106  200 103 1, 65  16, 011, 08  20,56
εs = +0,62 · 10-3; from which σs = +124 N/mm2.
It appears that the reinforcing steel is in tension. This result indicates that the cross-
section is cracked in SLS. The concrete compressive strain follows from horizontal force
equilibrium:
εc = 0,52 · 10-3
Tension zone height:

s 0, 62
t (h  0,15  0,10)  (1,9  0,15  0,10)  0,90 m
s  c 0, 62  0,52
ae 0, 20
  0, 22 (fig. 11.16)
t 0,90
Figure 11.15: In case bar spacing s = 100 mm (fig. 11.21), σs,max = 220 N/mm2 >
124 N/mm2.
The calculations are carried out for cross-sections A and C too.
In section A, the forces are:

18,81 MNm, see fig. 11.18.
(1,4 MNm).
= 0,88 · 15,29 = 13,46 MNm).
The two equilibrium conditions:

M  0:
  A   A  E  (1,90  0,10  0, 25)  P
s s 1 p s m,  (0, 47  0, 25)  6, 75
H  0:
 A  A E

 s  7540  0,38 14400  106  200 103 1,55  17, 21  0, 22  6, 75
April 2018 11-26

Result: εs = +0,73 · 10-3; σs = +146 N/mm2.
The reinforcing steel is in tension; the cross-section is cracked in SLS. The mean
concrete compressive strain:
εc = 0,26 · 10-3
Tension zone height:
s 0, 73
t (h  0,10  0, 25)  (1,9  0,10  0, 25)  1,14 m
s  c 0, 73  0, 26
ae 0, 20
  0,18 (fig. 11.16)
t 1,14
Figure 11.15: In case bar spacing s = 100 mm (average value) (fig. 11.21), σs,max =
220 N/mm2 > 146 N/mm2.
In section C, the forces are:

13,02 MNm, see fig. 11.18.
(1,1 MNm).
= 0,39 · 15,29 = 5,96 MNm).
The two equilibrium conditions:

M  0:
  A   A  E  (1,90  0,10  0, 25)  P
s s 1 p s m,  (0, 47  0, 25)  8,16
H  0:
 A  A E

 s  7540  0,38 14400  106  200 103 1,55  17, 21  0, 22  6, 75
Result: εs = +0,73 · 10-3; σs = +146 N/mm2 (the same result as for section A).
The reinforcing steel is in tension; the cross-section is cracked in SLS.

εc = 0,26 · 10-3
Tension zone height (also see the results for section A):
s 0, 73
t (h  0,10  0, 25)  (1,9  0,10  0, 25)  1,14 m
s  c 0, 73  0, 26
April 2018 11-27

ae 0, 20
  0,18 (fig. 11.16)
t 1,14
Figure 11.15: In case bar spacing s = 150 mm (average value) (fig. 11.21), σs,max =
160 N/mm2 > 146 N/mm2.
Also in sections A and C, the cross-section is cracked at SLS.
The degree of prestressing can be calculated for the cross-sections considered. The
decompression bending moment follows from the bending moment from prestressing and
the axial prestressing force.
Section A:
P 17, 21 1, 08
M D  m, Wb  M pm,    13, 46  17,1 MNm
Ac 3, 61 1, 43
Section B:
P 16, 01 2, 29
M D  m, Wt  M pm,    16,97  28, 4 MNm
Ac 4, 45 0, 72
Section C:
P 14,80 1, 08
M D  m, Wb  M pm,    5,96  9,1 MNm
Ac 3, 61 1, 43
The degree of prestressing K is the ratio between the decompression moments and the
bending moment from permanent loads:
17,1
KA   0,85
18,81  1, 4
28, 4
KB   0, 76
34,93  2, 6
9,1
KC   0, 64
13, 02  1,1
Note that the bending moments from differential settlements are included in the bending
moments from permanent loads. This is in accordance with EN 1992-1-1 cl. 2.3.1.3
which states that differential settlements must be regarded as permanent loads, whereas
loads from temperature gradients are variable loads (EN 1992-1-1 cl. 2.3.1.2).
The degree of prestressing is relatively high (a minimum of 0,64). This indicates that the
prestressing force might be reduced. However, to meet the bending moment resistance
requirement, additional reinforcement As = 7200 mm2 has to be provided in the webs at
midspan (section A). When applying less prestressing, the amount of reinforcement
required strongly increases and it might be difficult to apply all the reinforcement at the
April 2018 11-28

bottom of the two webs (there is no bottom flange at midspan). On the other hand, the
bending moment resistance calculation is conservative, since it is assumed that the
prestressing steel reaches a stress fp0,1k. A more detailed analysis might demonstrate that
the prestressing steel stress is between fp0,1k and fpk at ULS, which reduces the amount of
additional reinforcing steel required.
It becomes clear that the design is iterative, balancing the prestressing force on one hand
and on the other ULS (additional As) and SLS requirements (crack width from σs).
11.7 Literature
11.1 Bachmann, H.: “From full to partial prestressing”, Edition “Prestressed concrete
in Switzerland”, 9th FIP congress, 1982, pp. 11-18 (in German).
11.2 Bachmann, H.: “Design of partially prestressed concrete structures based on

Swiss experiences”, Journal of the Prestressed Concrete Institute”, July- August
1984, pp. 84-105.
11.3 Bruggeling, A.S.G.: “Partially prestressed concrete structures – a design

challenge”, Journal of the Prestressed Concrete Institute, March-April 1985, pp.
140-171.
11.4 Kupfer, H., Scholz, U.: “Economy as a criterion for the choice of the degree of
prestressing”, Concrete Plant + Precast Technology, No. 5 / 1986, pp. 289-293.
11.5 Peter, J.: “Structures with partial prestressing”, Beton- und Stahlbetonbau 6/1986,
pp. 150-152 (in German).
11.6 Walther, R.: “Partial Prestressing”, Beton- und Stahlbetonbau 4/1975, pp. 79-82.
11.7 Walther, R.: “Partial prestressing”, Edition Prestressed Concrete in Switzerland,

9th FIP-Congress, 1982, pp. 19-27.
April 2018 11-29

April 2018 11-30

Chapter 12 – Unbonded tendons
12. Prestressing with unbonded tendons

12.1 Introduction
In most prestressing systems, the tendon ducts are injected with a grout after the tendons
are prestressed (see also section 2.2.1). However, there are also systems where injection
is not necessary, because the prestressing steel inside the duct is protected by grease and
is embedded in a plastic sheet. Both are applied already in the factory (section 2.2.2).
This technology is denoted as “prestressing with unbonded tendons”. For the prestressing
tendons almost exclusively 7-wire strands Y1860 having a cross-sectional area of 100 or
139 mm2 (12,9 mm and 15,2 mm) are used. Some of the corresponding prestressing
forces and design strengths are discussed in section 2.4.
The most important characteristics of this prestressing system are:

 The prestressing steel is already protected against corrosion when it arrives at the
building site; no injection at the site is necessary.
 It is not necessary to apply a large concrete cover on the tendons, because of the
corrosion protection. In combination with the small diameter of the tendon, a large
eccentricity of the prestressing steel can be realized. This is an advantage, especially
in case of small depth slabs.
 The tendons can be installed quickly and easily (fig. 12.1)
 After cracking, the increase of the deformation of the prestressing steel is large since
there is no bond between the prestressing steel and the duct. As a consequence, there
is also no bond between the prestressing steel and the concrete of the structure. Since
the strain of the tendon is constant over the full length between the anchorages, the
deflection of the structural element can be relatively large at increasing load. As a
result, the strain of the concrete in the compression zone can be high. Therefore, the
concrete fails long before steel fracture occurs. The resistance of a structure with
unbonded tendons is therefore lower than the resistance of a similar structure
provided with the same amount of bonded tendons. The resistance can be increased
by adding reinforcing steel.
Fig. 12.1 Positioning of unbonded prestressing tendons at the building site
April 2018 12-1

In a number of cases, the use of unbonded tendons results in increased economy, for
example slabs in buildings, containment structures (fig. 12.2), and shell structures.
Fig. 12.2 Monostrands applied in a segmental containment structure
12.2 The behaviour of structures with prestressed unbonded tendons

Figure 12.3 shows parts of two beams. The beam from fig. 12.3a is reinforced and/or
contains prestressing steel that is bonded by injection grout. The steel stress is largest in
the crack and decreases away from the crack thanks to bond action (Ns < Nsr). As a result,
at a certain distance from the crack, the stresses in the steel are equal again to those in the
uncracked area. This distance is denoted as ar in fig. 12.3a. The magnitude of the distance
ar depends on the magnitude of the bond stresses.
ar l
a A r B b A er B
cr c cr c
Nc Nc
zero
M curv e M Np
Ns
Nsr Ns
s fc s fcσct
σct
Fig. 12.3 Cracking of a beam [12.3]

a. (left): in a beam with bonded prestressing steel
b. (right): in a beam with unbonded prestressing steel
The pre-cracked situation is present again outside the influencing area of the crack.
There, new cracks can occur once the concrete tensile stress σct reaches the tensile
strength of the concrete. This will result in a well-distributed crack pattern, with cracks at
regular, relatively small, distances.
April 2018 12-2

In a beam with unbonded prestressing steel, a different mechanism occurs after the
formation of the first crack. However, in spite of the absence of bond, also here cracks
will appear at regular distances. Figure 12.3b shows the stress trajectories for the area
adjacent to the crack and gives the stress distribution over a section where the original
stress distribution is restored. The distance required to have the original stress distribution
depends on de St.-Venant’s disturbance length (here denoted as le) which is about equal
to the height of the concrete cross-section. The elongation of the prestressing steel is very
large, because the prestressing steel is only fixed at the anchorages at both ends of the
beam. This implies that, in case of one crack, the elongation of the steel (increase of steel
strain multiplied with the tendon length between the anchors) concentrates in one crack.
Therefore, the crack widens considerably and substantially reduces the depth of the
uncracked compressive area. Furthermore, the compressive zone is highly curved and
tends to separate itself from the low-stressed concrete parts in between the cracks, see fig.
12.4.
Fig. 12.4 The formation of transverse cracks at the tip of bending cracks
This results in a crack pattern characteristic for beams with unbonded tendons, namely
the so-called “fork-shaped” cracks, see fig. 12.5. This figure also shows the failure mode
typical for beams with unbonded tendons: the concrete fails in compression before the
steel has reached its tensile strength or its 0,1%-strain limit.
dh

Fig. 12.5 Crack pattern in beam prestressed with unbonded tendons[12.4]
April 2018 12-3

Figure 12.6 shows, as a qualitative comparison, the moment-deflection relation of two

almost identical prestressed beams. The only difference is that one has bonded tendons;
the other one has unbonded tendons. The cracking moment is the same for both beams.
After crack formation, the beam provided with bonded tendons has a much larger
stiffness than the beam provided with unbonded tendons. In the beam with bonded
tendons, the steel yields at a certain load. From then on the M- relationship demonstrates
an almost horizontal branch: the steel yields up to the instant of failure because the
concrete reaches its compressive strain limit. In the case of the beam with unbonded
tendons, the horizontal branch is not reached. The beam fails after a considerable
deflection because the concrete strain limit is reached in the compressive area. The
prestressing steel stress has hardly increased.
M with bonding
without bonding

Fig. 12.6 Qualitative comparison between two prestressed beams: one with bonded and
one with unbonded tendons
12.3 Serviceability limit states

When checking the structure in the serviceability limit states (SLS), a number of
differences between structures with unbonded and bonded tendons must be addressed.
The most important differences are:
 In the case of unbonded tendons, the prestressing steel does not contribute to crack
width control, see also section 12.2. There are two options to control cracking. The
first is to design such that the structural element does not crack in SLS. The second is
to allow cracking in SLS (partial prestressing), but to limit the crack widths by adding
(passive) reinforcing steel.
 Since the prestressing steel is well protected, the crack width requirements of
reinforcing steel, which are less severe, apply.
 The mean axial compressive stress in the concrete is most often relatively low (order
of magnitude 1 - 2 N/mm2). Therefore, the load-balancing action of the prestressing
steel is more important than its axial effect, see fig. 12.7 (This is mostly also the case
for slabs with bonded tendons).
 In case, apart from the prestressing steel, also reinforcing steel is available, crack
width control can be based on the stress in the reinforcing steel s, just as shown in
section 9.9. The bond factor  = 0 for unbonded tendons (eq. (9.37), table 9.6).
April 2018 12-4

qG + 0,4 qQ
d.l.+0.4 p.l.
Pm
Fp Fp
f Pm
Fig. 12.7 Balancing part of the load by prestressing (in this example: the permanent
load plus 40% of the variable load)
12.4 Bending moment resistance

To be able to calculate the bending moment resistance, it is necessary to know the
maximum force in the prestressing steel in the ultimate limit states (ULS). Due to the lack
of bond, the elongation of the tendon occurring after cracking of the concrete, is
uniformly distributed over the full length of the prestressing tendon. Therefore, it is not
possible to determine the increase of the strain in the prestressing steel by using
Bernoulli’s principle that plane cross-sections remain plane. As a consequence, in the
case of unbonded tendons, the deformation of the structure as a whole should be taken
into account.
For slabs, as a simplifying assumption for the calculation, it is assumed that the total
deformation concentrates at one crack, see fig. 12.8. The slab is modelled as two rigid
bodies that rotate about their intersection point.


w

Fig. 12.8 Simplified calculation of the increase of the stress in the prestressing steel in
a slab
The vertical displacement at midspan is denoted δ. From the geometry it follows that:

 1
2 l
The crack width:
w  2 z
April 2018 12-5

If the internal lever arm z (distance from the prestressing steel force to the resulting
concrete compressive force) is estimated to be 0,75 h, then
w  1,5  h
The total elongation l of the prestressing steel is equal to the crack width w, so that:
 
l  1,5  h  1,5 h3 h
1
2 l l
It was found experimentally that the deflection at failure is usually more than l / 50.
When using this conservative value
l
 (12.1)
50
it is found that the elongation l of the tendon is at least
3h h
l   (12.2)
50 17
This is a conservative estimation for slabs. For a more accurate calculation, reference is
made to [12.3] and [12.4].
Figure 12.9 shows a continuous slab over a number of supports, subjected to a special
load configuration. In this case it is assumed that the entire slab is subjected to a quasi-
permanent load qg +  qq and that for one span the variable load is gradually increased.
Failure is assumed to occur at the ultimate load value qu at the highest loaded span.
Cracks are assumed to occur at the two supports of this span, as well as at its mid
position.
If it is assumed that the deflection in the highest loaded span is  = l / 50, then the total
elongation of the prestressing tendon is 2 h / 17. This is two times the result from eq.
(12.2) since a crack develops at each of the two intermediate supports on each side of the
span. Both these two cracks have a width w / 2. This follows from compatibility of
rotating rigid bodies, assuming that the crack width at midspan is w. Moreover, it is
assumed that the internal lever arm z = 0,75 h at the supports too and that the adjacent
spans do not deform.
As a result of the elongation, the increase of the force in the prestressing tendon is:
2
h
P  17 Ep Ap
L
where L is the total length of the tendons between the anchorages.
April 2018 12-6

Since the prestressing force is constant over the full slab length (note: unbonded
tendons!), the upward equivalent (balancing) prestressing load for each span is:
8 ( P  P) f
qp  (12.3)
l2
It now looks as if there is no difference between spans with and without cracks. However,
the tendon profile is forced to follow the slabs’ discrete rotations that occur at the cracks.
This causes kinks in the tendon profile. Their impact is discussed in detail in the chapter
“External prestressing”.
qplu.qp+vl.qvl
g qg + q  qq
pl.qpl+vl..qvl
Pm + ΔPm
Fp + Fp
f
f
 FpP).f
88.(FPp+
f
m m
F2
8.(Fp+Fp).(f+f) l
F
Fig. 12.9 Prestressing loads on a continuous slab with unbonded tendons
According to EN 1992-1-1 cl.5.10.8 (2) the bending moment resistance may be

calculated assuming that the stress in the prestressing steel increases with 100 N/mm2. In
the Dutch National Annex this increase is reduced to 50 N/mm2. These are conservative
values. It is allowed to use a higher value, but then a detailed calculation must be made.
The minimum radius of curvature R of the prestressing tendons should, according to the
Dutch code NEN 6720, satisfy the following conditions:
R  20 ØT for strands in a smooth duct (unbonded)
R  40 ØT for strands in a ribbed duct (bonded)
where ØT is the diameter of the tendon (including the duct).
In practice, the radius of curvature at an intermediate support is about 1,5 m for unbonded
tendons and about 2,5 m for bonded tendons.
April 2018 12-7

12.5 Shear resistance

Up to a few decades ago, it was assumed that shear reinforcement is not effective in
structures with unbonded tendons. This idea was based on the cracking patterns observed
in tests. If the steel has no bond to the concrete, the bending cracks generated can only be
perpendicular to the member axis (figs. 12.4 and 12.5). Stirrups perpendicular to the
beam axis would therefore not be intersected by inclined cracks, and hence would not be
able to contribute to the shear resistance. The shear resistance would then predominantly
come from arch action, so that the strength of the concrete arch seems to be of much
more importance (fig. 12.10).
Fig. 12.10 Load carrying mechanism based on arch action.
This conclusion is basically correct for slabs not provided with additional reinforcing
steel. However, additional (bonded) reinforcing steel is often provided. This might come,
for instance, from crack width reduction (SLS) or from bending moment resistance
(ULS) requirements. Thanks to the bonding effect of the reinforcing steel, inclined
cracking occurs.
Research [12.5, 12.9] demonstrated that shear reinforcement can also be efficient for
another reason, even if no additional reinforcing steel is provided. This is illustrated in
fig. 12.11a, showing the failure mode of an I-beam prestressed with unbonded tendons.
The beam has no additional reinforcement in the tensile zone and has no shear
reinforcement. The first inclined crack that appears in the web immediately results in
failure of the beam. This inclined crack is no bending tensile crack, starting from the
bottom of the beam, but is caused by splitting of the compression strut between the point
where the load is applied and the support (fig. 12.11b). Stirrups can be effective in
controlling crack formation in the strut.
a b
Fig. 12.11 a. Failure mode of an I-shaped beam, prestressed by unbonded tendons,

without shear reinforcement [12.5].
b. Formation of a splitting crack in the compression strut
April 2018 12-8

Figure 12.12 shows the behaviour of a similar beam provided with shear reinforcement.
Figure 12.12a shows the reinforcement. In fig. 12.12b it is shown that a bending crack
originates at the bottom of the beam, just at the position of the load. Well-distributed
inclined cracks originate in the web. After the formation of the first inclined crack, the
load could still be considerably increased until the compression strut failed (web crushing
failure, fig. 12.12c).
a two legged stirrup
Fig. 12.12 Test on an I-shaped beam, prestressed with unbonded tendons and shear
reinforcement.
a. Reinforcement
b. Crack pattern
c. Failure mode
The behaviour displayed in fig. 12.12 shows a large similarity with the behaviour of a
beam with (bonded) reinforcing steel, as illustrated by fig. 12.13. Figure 12.13a shows a
truss model that simulates the behaviour of a beam reinforced with passive steel. Figure
12.13b shows a model to simulate the behaviour of a beam prestressed with unbonded
tendons. The only difference between both models is the force in the longitudinal tensile
tie at the bottom of the truss. In case (b), this force, that only slightly increases during
loading, acts as a concentrated force at the beam ends. In case (a), on the contrary, there
is a tensile force inside the beam. In both cases, the system is in equilibrium; in case (b)
with a compressive force at the bottom of the beam, and in case (a) with a tensile force
provided by the reinforcing steel. This implies that the shear resistance of structural
members with unbonded tendons can be calculated using the same procedure as for
members with bonded reinforcement.
April 2018 12-9

- - - -
a - + - + - - + - + -
+- + + + +-
- - - -
b - + - + - - + - + -
- - - -
Fig. 12.13 Truss models for a beam with passive reinforcement (a) and for a beam
prestressed by unbonded tendons (b)
12.6 Slabs prestressed by unbonded tendons
Prestressed slabs are predominantly applied for structures like office buildings, parking
houses, schools, hospitals and warehouses. These structures, with spans from 7 to 10 m
and live loads up to 5 kN/m2, mostly have flat slabs provided with concentrated tendons
in column strips in one direction. In the case of larger spans and/or higher loads, the
prestressed slabs are provided with drop panels, have columns strips with concentrated
tendons in two directions or are designed as waffle slabs.
Compared with other structural solutions, prestressed flat slabs (both with bonded and
unbonded tendons) have the following advantages:
 Larger spans and smaller slab thickness, see fig. 12.14.
 The selfweight of the slab structure is reduced, which has a favourable effect on the
dimensions of the structure and its foundation.
 Thanks to the smaller height of a storey, the total height of the building can be
reduced, or the number of storeys in a building with a fixed height can be increased.
This also holds for underground structures.
 Slabs exposed to a predominantly permanent load, show a favourable behaviour with

regard to deflection and cracking.
 By an intelligent configuration of the prestressing tendons, a high punching shear

resistance at the columns can be achieved.
April 2018 12-10

d c b
300
250 a
floor thickness [mm]
a
b
200
c
d
150
100
5 6 7 8 9 10 11 12 13 14 15
span [m]
Fig. 12.14 Slab thickness in relation to the span [12.11]:

a = mushroom floor in prestressed concrete;
b = mushroom floor in reinforced concrete;
c = flat slab floor in prestressed concrete;
d = flat slab floor in reinforced concrete.
Figure 12.15 shows four options for arranging the prestressing tendons.
In tendon configuration (a) the same number of tendons is provided in each direction.
The distance between the tendons in the span is relatively large and the tendons are
predominantly concentrated in the column strips since this is advantageous for the
transmission of the balanced load (exerted by the tendons) to the column. A simple
equilibrium model of the punching cone shows that the part of the load directly
transferred to the column, is 2 Pm,∞ sinα per tendon, see fig. 12.16. The smaller the radius
of curvature, the more the tendon contributes to the punching shear resistance.
In tendon configuration (b), see fig. 12.15, all prestressing tendons are concentrated in the
column strip. Between the columns strips, the slab is reinforced with (passive) reinforcing
steel.
In the configurations shown in fig. 12.15 (c) and (d), in one direction column strips are
provided with concentrated tendons. In the other direction, the slab is prestressed by
distributed tendons (c) or is only reinforced (d).
April 2018 12-11

a b
prestressing cables
reinforcing steel in
in the
in the field
span steel reinforcement
the span
in the field
prestressing
prestressingcables
tendons
in the collumn strip
in the column strip
c d
prestressing cables
in the span
in the field
Fig. 12.15 Various options to arrange unbonded prestressing tendons in a flat slab
[12.13]
Pm sin FpP.msin 
sin
Fp.sin
punch cone
punching shear cone
Fig. 12.16 Direct transfer of the balanced load from the prestressing tendons to the
column
A prestressed flat slab has to be checked for the ultimate limit states (ULS). One of them
is the bending moment resistance check. The bending moments in the slab caused by
loading can be determined on the basis of the theory of elasticity. EN 1992-1-1 does not
provide detailed information about the bending moments in flat slabs. EN 1992-1-1
annex I presents some simplified rules to distribute the negative (hogging) and positive
(sagging) moments over the column and middle strips (table I.1).
NEN 6720 cl. 7.5.3 presents more detailed information. Bending moments were
calculated using the theory of elasticity. The locally very high negative moments at the
columns were distributed over the column strip. The reinforcement moments (bending
moments plus torsion moments) were calculated and transformed into coefficients given
in tables (NEN 6720 tables 19 - 26).
April 2018 12-12

The tendon configurations from fig. 12.15 do not in all cases perfectly match with the
distributions of the moments that follow from the theory of linear elasticity. In particular
the arrangements (c) and (d) give deviating results. This might imply that there is some
redistribution of forces when, under increasing load, the slab cracks.
Especially the areas with high concentrated moments (near the columns), are sensitive to
this bending moment redistribution. Therefore, NEN 6720 cl. 9.9.2.2.c specifies a
minimum reinforcement required in those areas. This reinforcement should be provided
in both main load bearing directions and extended up to a distance of 0,25 lx and 0,25 ly
in x and y direction, respectively, from the centre of the column, see fig. 12.17. The
minimum reinforcement ratio for loading in bending has to be applied, see Chapter 7.
x
0.25 x 0.25 x 0.25 x
0.25 y
0.5 x
0.25 y
y
0.25 y
ar 0.25 x
ar 0.25 y 0.5 y
x
Fig. 12.17 Minimum reinforcement in the column area in a flat slab prestressed with
unbonded tendons
As an example of the determination of the required prestressing force and the

corresponding arrangement of the tendons, the roof of an underground parking house is
designed (fig. 12.18) [12.1].
The parking house has only one storey. Expansion joints subdivide it into three units. In
the design, a top layer of 0,4 m soil on top of the concrete roof has to be accounted for.
Furthermore, it is assumed that vehicles class 30 (total vehicle weight 300 kN) can pass
and load the roof.
By prestressing the slab instead of only reinforcing it, the slab thickness is reduced from
500 to 350 mm, whereas, simultaneously, the span increases from 5,0 to 8,75 m.
April 2018 12-13

In the project, unbonded monostrands Ap = 139 mm2, Y1860, are applied. The ULS
design value for the ULS (design load) of such a tendon is 139 · 1860 / 1,1 = 235 · 103 N.
The code used in the design allows a force directly after prestressing of 188 kN, that can
be temporarily increased to 204 kN. For the calculation of the (small) prestressing losses
by friction μ = 0,06 is used, in combination with k = 0,009 rad/m for the Wobble effect.
Thanks to these low values, it is possible to even prestress the longest tendons, which
pass five spans, from one side, without too high losses.
NEN 6720 cl. 7.5.3 was used to determine the moments in the column strip and the
middle strip from the selfweight and the soil layer. The tendons are arranged cross-like,
with a concentration of tendons in the column strip and an increased tendon spacing in
the middle strip (see fig. 12.15a). The prestressing system can effectively balance the
load.
For the roof of the parking house, the balanced (equivalent) load qp is equal to the
selfweight of the roof qg,1 and the soil layer qg,2. The traffic load is not yet considered
because the structure is designed in partially prestressed concrete. This implies that the
requirements from ULS (failure safety) and from crack width limitation (under maximum
load in SLS) will be met by adding reinforcing steel if required.
At a volumetric weight of 25 kN/m3 for the reinforced concrete and 19 kN/m3 for the soil,
the total load to balance is:
qp = -(qg,1 + qg,2) = - (0,35 · 25 + 0,40 · 19) = - 16,4 kN/m2
The prestressing force required to provide this balancing load is calculated using the
relation:
qp l 2
Pm, 
8f
At first, the slab strip between axes 14 and 16 (see fig. 12.18) is considered. The main
span in this direction is 8,75 m. The distance from the tendon to the outer concrete fibre
is assumed to be 0,04 m, both at midspan and support.
The drape of the tendon profile then is f = 0,35 – 0,04 – 0,04 = 0,27 m (neglecting the
downward curvatures at the supports).
Per m1 width, the effective prestressing force must be:

16, 4  8, 752
Pm,   581 kN/m
8  0, 27
When assuming a 20% prestressing loss from long term effects, the required initial
prestressing force is:
April 2018 12-14

581
Pm,0   726 kN/m
0,8
726
This corresponds to a theoretical number of tendons of  3,9/m1 .
188
For a width of 7,50 m (see fig. 12.18 on line F), the number of tendons is n = 7,50 · 3,9 =
29,30  30.
About half of these tendons are arranged in the column strips, whereas the other half is
uniformly distributed over the middle strip (between the column strips).
The mean compressive stress in the concrete σcm,p is:
30  0,8 188 103
 cm,p   1, 72 N/mm 2
7,5 1000  350
This stress is relatively low (much lower than, for instance, in beams). This demonstrates
that the most important component of the prestressing is the balancing part.
Friction, shrinkage and relaxation losses, and losses by wedge slip, can be determined in
the usual way. Depending on the degree of prestressing, they will amount up to,
approximately, 20%. The elastic shortening of the slab by the axial compressive force is
small.
The resistance check in ULS should be carried out following the method presented in
section 12.4. Furthermore, crack width control should be carried out according to the
method presented in Chapter 9. This is not further presented here.
Special attention should be paid to the behaviour of the slab in the vicinity of the
columns. As pointed out already, the punching shear resistance is considerably increased
by the component Vp = Pm,∞ sinα. It should be emphasized that this component is
automatically taken into account when using the equivalent prestressing load (balancing)
method, in which the prestressing loads are directly transferred to the column by the local
downward pressure of the curved tendons. So, additionally reducing the punching shear
force by Vp = Pm,∞ sinα would be erroneous, because then the effect of prestressing
would be taken into account twice! The effect of prestressing can be optimized by
positioning the inflexion point of the tendon (where the upward curvature changes into a
downward curvature and vice versa) in the punching cone. In that case the maximum
effect of load balancing is achieved. In the case considered, no shear reinforcement was
required because of this effect.
April 2018 12-15

13 14 16 17
7.50 7.50 4.875
F
4.705
E
a a
8.75
D
section a-a
4
10
= =
35
4
2e bundle 3e bundle 4e bundle
Fig. 12.18 Arrangement of the prestressing tendons in the design example (dimensions
m (top) and cm (bottom))
For easier handling during positioning, the tendons can be bundled at the construction
site. According to clause 9.4.2c of NEN 6720 a bundle can consist of a maximum of two
monostrands. In other countries larger bundles are allowed. For instance in Germany, it is
allowed to bundle up to four monostrands in the column strip (fig. 12.19) and up to two
in the middle strip. In the anchorage region, the distance between the strands has to be
increased to 120 -140 mm, to provide space for the anchors, see fig. 12.18.
Fig. 12.19 Bundles of monostrands in the column strip
April 2018 12-16

Fig. 12.20 Arrangement of tendons in a slab of a parking garage: prestressing tendons

combined with reinforcing steel [12.12]
12.7 Other applications

Typical examples of other application are heavily loaded industrial floors, taxi platforms
at airports and speed skating stadiums (fig. 12.21).
Prestressing with unbonded tendons also offers advantages for the design of containment
structures and shell structures. In those cases failure safety does not play a dominant role,
but crack width control is mostly governing. The elastic ducts fully function as corrosion
protection, if the concrete is cracked whereby the tendon is exposed to aggressive
substances. Particularly with regard to this aspect, a number of special applications are
possible, such as prestressing of bridge decks and containers with aggressive fluids.
For an extended example of the design and detailing of an office building slab,
prestressed with unbonded tendons, reference is made to [12.15].
April 2018 12-17

Fig. 12.21 Speed skating stadium (60 m x 30 m) [12.12]
12.8 Literature
12.1 Gerber, C., Özgen, W.: “Flachdecke mit teilweiser Vorspannung ohne Verbund”,
Beton- und Stahlbetonbau, Juni 1980, Heft 6., Seiten 129-132.
12.2 Grasser, A.: “Bemessung von Beton- und Stahlbetonbalken bei Biegung mit
Längskraft, Schub und Torsion”, Deutscher Ausschuss für Stahlbeton, Heft 240,
1979.
12.3 Ivanyi, G., Buschmeyer, W.P.: “Biegerißbildung bei Plattentragwerken mit
Vorspannung ohne Verbund”, Beton- und Stahlbetonbau 9/1981, S. 215-220.
12.4 Ivanyi, G., Buschmeyer, W.P.: “Additional strains in unbonded tendons during
loading”, Magazine of Concrete Research, Vol. 37, No. 37, March 1985.
12.5 Ivanyi, G., Samol, J.: “Versuche zur Schubtragfähigkeit an Balkentragwerken mit
Vorspannung ohne Verbund”, Forschungsbericht No. 31, Universität Essen, Juli
1985.
12.6 Ivanyi, G., Buschmeyer, W.P., Müller, R.A.: “Entwurf von vorgespannten
Flachdecken”, Beton- und Stahlbetonbau 4/1987, S. 95-105.
12.7 Ivanyi, G., Buschmeyer, W.P.: “ Kontrollierte Biegerißbildung zur Sicherung der
Verformungsfähigkeit bei teilweiser Vorspannung ohne Verbund”, Der
Bauingenieur 62, 1987, S. 339-343.
12.8 Kordina, K., Hegger, J., Teutsch, M.: “Anwendung der Vorspannung ohne
Verbund”, Deutscher Ausschuß für Stahlbeton, Heft 335, 1984.
12.9 Kordina, K., Hegger, J.: “ Zür Ermittlung der Biegebruch-Tragfähigkeit bei
Vorspannung ohne Verbund”, Beton- und Stahlbetonbau 4/1987, S. 85-90.
12.10 Marti, P., Ritz, P., Thürlimann: “Prestresssed Concrete Flat Slabs”, Institut für
Baustatik und Konstruktion ETH Zürich, Bericht No. 68, Feb. 1977.
April 2018 12-18

12.11 Matt, P.: “Vorspannung ohne Verbund: Beispiele und Möglichkeiten der
Anwendung”, Beton- und Stahlbetonbau 9/1981, S. 212-215.
12.12 Schütt, K.: “Vorspannung ohne Verbund: Möglichkeiten und Beispiele
ausgeführter Bauwerke”, Beton- und Stahlbetonbau 6/1968, pp. 153-154.
12.13 Wölfel, E.: “Flachdecken mit Vorspannung ohne Verbund”, Der Bauingenieur 55,
1980, S. 185-195.
12.14 Leeuwen, J., Tukker, T., Veenstra, P.: “Voorspanning zonder aanhechting”, CUR
Rapport 95, December 1979.
12.15 Bouquet, G.Chr., Groeneveld, J., Keusters, A.C.A.M., Pauw, J.H., Veen, C. van,
Zielinski, A.J.: “Toepassingen van VZA en VMA bij vloeren in de
utiliteitsbouw”, Stuvo-rapport 95, mei 1992.
April 2018 12-19

April 2018 12-20

Chapter 13 – External prestressing
13. External prestressing

13.1 Introduction
Since a number of years, bridges in which the prestressing tendons have a position outside the
concrete cross-section are built. This idea is not totally new: in the pioneering era of
prestressing a number of bridges were already built in this way, for instance the Saale bridge
(Alsleben, Germany) in 1928, and the Aue bridge (Germany) in 1936. The first applications,
however, were not very successful. The execution was relatively expensive and almost all
structures suffered damage by corrosion, since the technology of corrosion protection still had
a low standard. As a consequence, the prestressing technology developed into the direction of
cast-in prestressing tendons, whereas the idea of external prestressing was more or less
abandoned.
The interest in external prestressing has, however, grown again over the last decades. For this
increased interest a number of reasons can be mentioned [13.1], [13.2]:
 The necessary repair of bridges with corroded prestressing tendons.
 The need for methods to strengthen structures because of increased traffic loads.
 To simplify execution, by avoiding a complicated layout of the tendons inside the
concrete. This offers advantages with regard to casting (fig. 13.1), prestressing (no or
small friction losses) and the injection of the tendon ducts (no leakage). Furthermore,
unintended curvature pressures are avoided (fig. 13.2).
 The need to be able to repair (bridge-) structures, without temporarily closing them.
 To reduce the web width, by getting rid of the tendon ducts (fig. 13.3).
 To be able to control the durability of the structure and to eventually replace corroded
tendons.
 The possibility to apply additional prestressing in the case of unexpected deformations.
Fig. 13.1 Problematic casting of concrete Fig. 13.2 Unintended effects by

due to the presence of prestressing curvatures in cast-in
tendons prestressing tendons
February 2019 13-1

Fig. 13.3 Reduced strength in the construction phase from empty ducts in the web
Figure 13.4 shows an application of external prestressing. It concerns the Long Key Bridge in
Florida (USA). After placing the box-girder type of segments on a temporary steel truss, they
were prestressed together by external tendons. The tendons were guided through concrete
anchor blocks (saddles) at the bottom of the cross-section. At those saddles, the prestressing
forces were transmitted to the structure.
35.96
Fig. 13.4 Long Key Bridge: schematic representation of the construction sequence and the
tendon profile [13.3]
Another possibility is to apply transverse beams (fig. 13.5a) or stiffeners (fig. 13.5b) for
changing the direction of the tendons.
Fig. 13.5 a. Support of the tendons by transverse beams

b. Support of the tendons by vertical stiffeners [13.4]
February 2019 13-2

Especially in France (70 bridges) and the USA (30 bridges), experience is acquired with
externally prestressed bridges. Reference is made to publications of Virlogeux [13.4]-[13.8]
and Müller [13.9]. Nelissen [13.13] and Bruggeling [13.14] prepared State-of-the-Art papers
in the Dutch journal “Cement”. Vermeulen [13.22] demonstrated that external prestressing
can be of interest in Dutch conditions.
13.2 Technology of external prestressing

13.2.1 Prestressing steel
Principally, for external prestressing the same prestressing steel can be used as for “normal”
prestressing. An important point is, however, the way in which external prestressing steel is
protected against corrosion. There are various possibilities.
Galvanising:
In this case the corrosion resistance depends on the type of galvanising and the thickness of
the coating. Galvanised prestressing steel has been applied in France several times. However,
there is some doubt about the long-term corrosion resistance. Furthermore, the coating can be
damaged during mounting and replacing of the tendons.
Epoxycoating:
The technology of applying melted polymers on the steel has been developed in the USA,
especially for reinforcing steel. It is questionable whether it is a good solution for prestressing
steel. At the anchorage the coating is locally interrupted, so that problems might occur here.
Special measures are necessary as well, to avoid damage during transport and mounting.
Protective ducts:
Suitable materials are steel and plastic (High Density PolyEthylene, HDPE). In order to
achieve a complete protection system, good solutions are needed for coupling, the anchorage
region and the saddles (where the deviations of the tendons occur). Injection of the ducts with
a cement grout is a good and economic solution. If the possibility of second prestressing (for
compensating unexpected deformations) should be kept open, no cement grout can be used.
Other possibilities are grease, paraffin, tar-epoxy or other products with plastic properties.
These products are not easy to inject; sometimes heating up to 100 ºC is required.
Furthermore, special measures are necessary to avoid leakage.
In this category monostrands, produced in factories, offer many advantages. Those strands are
surrounded by a plastic duct, which is filled with grease (fig. 2.20). In this way, the
prestressing steel is effectively protected from corrosion during transport, storage at the
building site and assembling. Monostrands can be applied as single units or in bundles. In the
last case they are mostly placed in an additional coating of steel or plastic, see fig. 13.6 (left).
The space between the strands is filled with grease or with cement grout. In the case of filling
with grease, the tendons, consisting of many monostrands, can be readily delivered at the
building site. Often internal spacers are used, guaranteeing the distance between the
monostrands, also at the saddles. The individual strands can, if necessary, be stressed or
released one by one.
February 2019 13-3

“Prestressing bands” are composed of a limited number of monostrands (fig. 13.6, right).
They need only a small volume of grease, also when they are composed to large units, see fig.
13.7. The prestressing bands can be bundled by piling up.
HDPE-Pipe Steel duct Grout Band with 4 monostrands
Prestressing wires Monostrand HDPE-Coating
Fig. 13.6 Different types of tendons for external prestressing [13.14]
Fig. 13.7 Bundle of prestressing bands [13.14]
13.2.2 Saddles at points of tendon deviations
At deviation points, a solution is necessary which enables tendon deviation and the
introduction of local prestressing loads into the structure without damage. This pertains to
both vertical and horizontal deviations. Horizontal deviations can occur in bridges with a
curved axis. The transition of the curved part of the tendons into the straight part should be
inside the area of the saddle, since otherwise kinks occur in the tendons. This concerns again
both the vertical and the horizontal curvature, see fig. 13.8. For the detailing of a saddle there
are various possibilities, like the ones shown in fig. 13.9.
February 2019 13-4

section B-B
A A
section A-A
undesirable deviation
B B
Fig. 13.8 Unintended deviations of the tendons at the saddle [13.14]
PE
a ST
steel tube
PE tube
ST = PE
socket
b steel tube
PE tube
ST < PE
socket
c steel tube
PE tube
ST > PE
Fig. 13.9 Some solutions for the junction of the external tendons to the saddle [13.15]
a & b: Discontinuous PE tubes connected to a steel tube
c: Continuous PE tube through a (larger diameter) steel tube
In most saddles a pre-deformed steel tube is applied, cast into the concrete or, in the case of a
steel saddle, connected by stiffening plates. The connection between the free part of the
tendon and the saddles should be detailed carefully not to damage the prestressing steel by
kinking. If the exchangeability of the tendons is a design criterion, this should be regarded in
the design.
Relative displacements between the prestressing steel and the saddle should be allowed to
occur. If the prestressing tendon and the surrounding tube have been rigidly connected, for
instance by a mortar, another solution is necessary. In this case the relative displacement can
be facilitated by a neoprene intermediate layer that allows a displacement of the tendon and
tube relative to the saddle, see fig. 13.10.
February 2019 13-5

For the design of the saddles, the minimum radius of curvature has to be taken into account to
avoid damage to the prestressing steel or duct. In [13.16] allowable radii of curvature are
given as a function of the tendon dimensions. These radii of curvature are between 2,5 m and
5,0 m.
mortar
morter
elastomeric bearing
elastomer
Fig. 13.10 Solution for the relative displacements between the saddle and a rigidly injected
prestressing tendon
In analogy with conventional prestressing systems, friction losses are calculated with eq.
(4.23). The Wobble-effect can be ignored, since the tendons have a straight alignment
between the saddles. On the basis of experimental results and experiences at the building site,
the following values apply for the friction coefficient :
 Untreated, dry strands pulled through a steel saddle:  = 0,25 - 0,30

 Untreated, greased strands pulled through a steel saddle:  = 0,20 - 0,25
 Untreated strands in a plastic duct, which are
directly pulled through a saddle:  = 0,12 - 0,15
 Greased monostrands, surrounded by plastic, combined to
a bundle in a plastic duct, pulled through a saddle:  = 0,05 - 0,07
13.3 General design processes

Structures with external prestressing behave in almost the same way as structures with
unbonded tendons. The best representation of the effect of prestressing is obtained through the
equivalent prestressing loading method (load balancing method), where the prestressing
forces are applied as loads at the anchorages and the saddles.
Figure 13.11 shows a beam on two supports with two deviation points. The prestressing
forces are regarded as being loads on the beam. At the deviation points the loads are applied
on the structure in the direction of the bisector of the angle of deviation (dashed arrow). If the
friction in the saddle is also taken into account, the resulting force indicated by the arrow
rotates over a certain angle.
The prestressing forces can be split into horizontal and vertical components. The calculation
can then be carried out in the usual way.
By a right choice of the tendon profile, a prestressing load can be generated, which balances
the loads (dead weight and part of variable load) on the structure in the optimum way.
February 2019 13-6

Sometimes more deviation points are created to have the optimum balancing effect, see fig.
13.12.
V2v V1v
Zv
Dv
e H2v
H1v
Normal force
A B - friction-
Fig. 13.11 Introduction of the prestressing force from an external tendon
Fig. 13.12 Cross-section over the bridge Val Durance [13.2]
In finding the optimum design, there is a strong dependence on the construction method. If it
is possible to provide reinforcing steel complementary to the external tendons, a solution in
partially prestressed concrete is possible. Then, the dead weight and a part of the variable load
are balanced. At relatively high loads cracks occur, which are well distributed and small, as a
result of the action of the reinforcing steel. By choosing a favourable ratio between the
amounts of prestressing steel and reinforcing steel, an economic solution can be achieved.
If, however, the structure is built in segments, the situation is totally different. The joints are
mostly profiled (fig. 13.13) to transmit the shear forces, and are bonded together by epoxy
glue. In this case no reinforcing steel crosses the joint, so that the crack width control is a
critical aspect. Since there is not sufficient knowledge about the long term behaviour of the
epoxy, it is mostly required to design the structure in such a way, that in SLS the joints are in
permanent compression. In such a case due notice should be taken of the effects of imposed
deformation (temperature gradients, differential settlement).
With regard to the behaviour in ULS, the best approach is to use a kinematic failure model. In
fig. 13.14 this is shown for a simple case. The elongation of the prestressing tendon is a
function of the deflection. There is also a relation between the elongation of the tendon and
the load. In section 13.4 an analysis is given for a specific case.
February 2019 13-7

5000
300
3x500 450 400 4x500
100
500
1000 1000
35
25
A
50
25
50
25
150
50
section A-A
25
A A
Fig. 13.13 Profiled joint in a segmental bridge (dimensions mm)

P

f
w = 4f .h w  = f .2
L L
Fig. 13.14 Kinematic model for an externally prestressed beam in the failure stage (two
anchors and two deviation points)
For a numerical analysis of a segmental bridge with external prestressing, reference is made to
Huang [13.19]. Figure 13.15 shows the calculated deformations in the failure stage. The
asymmetric behaviour is caused by torsional moments, corresponding to the eccentric loading
case considered. The analysis shows a distinct change of behaviour when the joints open, see
fig. 13.16. The calculation shows as well, that the transmission of shear forces close to the
support can become critical.
Fig. 13.15 Deformations of an externally prestressed segmental bridge at overloading [13.19]
February 2019 13-8

Fig. 13.16 Principal stresses in the web of a box girder bridge at the failure load [13.19]
Because of the small bottom flange, locally a combination of very high compressive stresses
and shear stresses occurs. Furthermore, fig. 13.16 shows that the transmission of forces in the
segments between the (opening) joints needs appropriate attention. Moreover, fig. 13.17
shows truss models for a segment with (fig. 13.17b) and without a tendon deviation point.
compression bars
Survey
tensile bars
a section bb cc
p p p p p p p p p p
P2
P1 P2 P1
Zv Zv
flow of
Flow of forces
forcesinin flow of
Flow of forces
forces in
in
segment
transversewithdiafragm
open segment withdiafragm
transverse open
joints
in theand deviation
span with joints andspan
in the no deviation
without
forceforce
deviation force
deviation force
T1 T2 T1 T2
p p p p p p
D1 D2 D1 D2
Zv
Truss modelll b Truss modelll c

truss model at truss model at
section b section c
Fig. 13.17 Truss models for a segment with and without tendon deviation point [13.19].
13.4 A practical application: the Long Key Bridge in Florida

An interesting example of the application of external prestressing is the Long Key Bridge in
Florida. The bridge was completed in 1980. It was built in the record-time of 12 months. The
upper structure of the bridge, which has a length of 3,7 km, is composed of 101 spans with a
length of 36 m. The segments have a height of 2,10 m and a width of 5,40 m.
February 2019 13-9

Fig. 13.18 Long Key Bridge in the construction phase [13.3]
Expansion joints have been provided at any 8 spans (distance 288 m). For the assembling of
the segments a moveable steel truss was used, see fig. 13.19. At first, all segments of a span
were put in position. In order to have a levelled bridge deck, the steel truss had a camber.
Between the pier segment and the first segment in a span, a joint having a width of 150 mm
was provided, filled with concrete at the site. Subsequently, the span was prestressed in one
operation. Per span always six tendons were used. These were anchored above the piers,
where the tendons of the adjacent spans overlap, see fig. 13.20.
35.96
Fig. 13.19 Placing the segments on the moveable supporting truss
February 2019 13-10

Deviation block Closure joint
Fig. 13.20 Tendon layout
In the saddles, the tendons were led through steel ducts. In between the saddles, they are
placed in plastic ducts. At the junction between the two types of ducts, neoprene sleeves take
care of the tightness and the required elasticity, see fig. 13.21. After prestressing, the ducts
were injected with a grout. In this way the upper structure had its full bearing resistance
directly after prestressing. By using this assembling system, two to three spans per week (72
to 108 m) could be finalized. In some periods the production even increased to five spans per
week (180 m).
A particular aspect was that no structural topping was applied on the deck: the traffic moved
directly on the concrete of the segments. Another particularity was, that dry joints were
applied between the concrete segments (concrete to concrete, without intermediate layer).
This requires a high production accuracy. Figure 13.22 shows a cross-section at a joint.
Steel pipe Neoprene sleeve
Polyethylene
duct
Fig. 13.21 Detail of a tendon junction at a saddle
February 2019 13-11

Fig. 13.22 Interface of joint between two segments
13.5 Analysis of the behaviour of an externally prestressed bridge under

an increasing load
In section 13.3 it was shown already, that for an externally prestressed structure it is necessary
to analyze the behaviour at increasing load, from the unloaded state to the failure state. In this
way the best insight is obtained in the safety of the structure and in the sensitivity for imposed
deformations (temperature gradients and differential settlements). The following example is
inspired by the Long Key Bridge, although the dimensions are not exactly the same.
Figure 13.23 shows the dimensions used in the analysis.
2500 mm
P P
c.a Ac = 4,07 m2
Ic = 3,51 m4
Fig. 13.23 Data used in the analysis
February 2019 13-12

The selfweight of the upper structure is 4,07 · 25 = 102 kN/m = 0,102 MN/m.
The variable (traffic) load is 500 kg/m2, which is 8 · 5 = 40 kN/m = 0,040 MN/m.
In situation 1, the full load is applied over the full length of the bridge. For that situation, the
resulting bending moments are:
 Support: Ms = - (qg + qq) l2 =  · 0,142 · 332 = -12,9 MNm

 Midspan: Mm = -0,5 Ms = 6,4 MNm
When there is no variable load on the bridge, the bending moments are:
 Support: Ms = - · 0,102 · 332 = -9,3 MNm

 Midspan: Mm = -0,5 Ms = 4,6 MNm
As a simplification, the cases “fully loaded” and “fully unloaded” are regarded as being
governing for the design.
The loads by external prestressing are determined using fig. 13.24. Only the vertical force
components FV are presented. The horizontal force components at the deviation points
(P·(1-cos)) are not indicated.
P P
FV FV
FV =Psin
P
F =2Psin(/2)
-FV a2 / l
FV a (l-a) / l
Fig. 13.24 Prestressing load.
If the deviation points of the tendon profile are applied at a = l / 3, then the bending moments
generated by prestressing are:
 Support: Ms = 2/9 FV l
 Midspan: Mm = -1/9 FV l
February 2019 13-13

This implies that the ratio │Ms / Mm │= 2.
For the angle  it is found that:
tan  = (2,0 - 0,4) / 11 = 0,145 so that  = 8,3º.
The force FV then is:
FV = P sin  = 0,144 P
Note:  = 0,144 rad is a small angle. This implies that sin =  can be used.
Now, the value for FV is determined to balance the dead weight and 50% of the variable load.
At the support, a bending moment of 0,5 · (-12,9 + -9,3) = -11,1 MNm has to be balanced by
prestressing. The force required follows from:
2
/9 FV l = 11,1 MNm
or:
2
/9 FV l = 2/9 · 0,144 P · 33 = 11,1 MNm
so that:
P = 10,5 MN
With pm,∞ = 1000 N/mm2 this would require a cross-sectional area of the prestressing steel
Ap = 10,5 · 106 / 1000 = 10500 mm2, or four tendons with Ap  2600 mm2 each.
The bending moments are shown in fig. 13.25.
In a fully loaded situation, the remaining moment at the support M = -12,9 –

-11,1 = -1,8 MNm. This moment has to be resisted. If the prestressing tendons in the end
spans are anchored at the gravity point of the cross-section (fig. 13.23), there is no additional
bending moment from anchor eccentricity.
However, the mean compressive stress in the concrete by the axial prestressing force has not
been taken into account yet:
cm = - P / Ac = -10,5 / 4,07 = -2,6 MN/m2
The SLS stress distribution in the governing cross-sections can now easily be calculated:
 Support, fully loaded, top side:
M 1,8
 ct    cm   2, 6  2,1 MN/m 2
I 3,51
yt 0,91
 Support, fully loaded, bottom side:
February 2019 13-14

M 1,8
 cb    cm    2, 6  3, 4 MN/m 2
I 3,51
yb 1,59
In a fully loaded situation in SLS, the remaining moment at midspan M = 6,4 – 11,1 / 2 =
0,85 MNm.
 Midspan, fully loaded, top side:
M 0,85
 ct     cm    2, 6  2,8 MN/m 2
I 3,51
yt 0,91
 Midspan, fully loaded, bottom side:
M 0,85
 cb    cm   2, 6  2, 2 MN/m 2
I 3,51
yb 1,59
Now the situation in case no variable load is applied, is dealt with. Only the dead weight and
the prestressing are active.
The remaining moment at the support M = 9,3 – 11,1 = -1,8 MNm.
 Support, not loaded, top side:
M 1,8
 ct    cm   2, 6  3,1 MN/m 2
I 3,51
yt 0,91
 Support, not loaded, bottom side:
M 1,8
 cb     cm    2, 6  1,8 MN/m 2
I 3,51
yb 1,59
The remaining moment at midspan M = 4,6 – 11,1 / 2 = -0,95 MNm.
 Midspan, not loaded, top side:
M 0,95
 ct     cm    2, 6  2, 4 MN/m 2
I 3,51
yt 0,91
February 2019 13-15

 Midspan, not loaded, bottom side:
M 0,95
 cb    cm   2, 6  3, 0 MN/m 2
I 3,51
yb 1,59
All values for both the fully loaded and the unloaded situation are given in fig. 13.25. It is
seen that the joints are fully compressed under all loading combinations.
-12,9 -12,9
-9,3 -9,3 -2,1 -3,1 -2,4 -2,8
4,6
G
G+Q
fully loaded Q=0
6,4
G+Q G
-5,5 -1,8 -3,4 -2,2 -3,0
support span
from prestressing
11,1 11,1
Fig. 13.25 Bending moments [MNm] and stresses [N/mm2] at support and midspan
In the next step, the behaviour at ULS is studied.
To open the joint, the support moment has to introduce an additional 2,1 N/mm2 tensile stress
at the top. The increase of the bending moment required is:
I 3,51
M   ct  2,1  8,1 MNm
yt 0,91
relative to the fully loaded situation. The corresponding increase of the uniformly distributed
load q follows from:
M = -1/12 q l2 = -8,1 MNm, so that q = 0,089 MN/m.
For the span the stress increase is 2,2 N/mm2 to have joint opening. This requires:
I 3,51
M   cb  2, 2   4,9 MNm
yb 1,59
The corresponding value for q follows from:
M = 1/24 q l2 = 4,9 MNm, so that q = 0,108 MN/m, which is not governing.
It turns out that joint opening occurs at an additional load of 0,089 MN/m = 89 kN/m (the
support cross-section is governing).
February 2019 13-16

The total load at joint opening at the support then is (selfweight + variable load + additional
load):
qtot = 102 + 40 + 89 = 231 kN/m
If the load is further increased, the midspan cross-section joint opens too and finally, the
mechanism shown in fig. 13.26 occurs.
l/6
Fig. 13.26 Failure mechanism with opening joints at supports and at midspan
It is now assumed that the parts between the opening joints are rigid, and that the force in the
prestressing tendons starts to increase only at joint opening. So, when the kinematic model
starts to work, the stress in the prestressing steel is still pm,∞ = 1000 N/mm2, with a
corresponding prestressing force P = 10,5 MN.
At the ends of the structure, the prestressing force P is applied centrically. At a deviation
point, a vertical force FV = P sin  = 0,144 P = 1,51 MN is exerted. The situation directly
after opening of the joints is shown in fig. 13.27, where the rotation of the bridge part is
exaggerated since actually there is almost no vertical displacement in this stage. At increasing
deformation, the compressive force in the concrete gradually moves downwards at the support
and upwards at midspan.
It is assumed that in the first stage after joint opening the deflection of the bridge is that small,
that the angle of the kink does not change. Moreover, it is assumed that joint opening results
in a shift of the concrete compressive forces to the centre of the flanges. The vertical positions
of the two horizontal forces at the right hand side of fig. 13.27 (10,5 MN tendon force and
10,5 MN top flange concrete compressive force) are not influenced by the small deflection.
The small deflection also makes that tendon elongation is small. It is, therefore, assumed that
the prestressing force does not change.
February 2019 13-17

10,5
10,5 10,5
1,51
1,51
1
/2ql
10,5
Fig. 13.27 Equilibrium of a rigid part during joint opening (dimensions MN and m;
displacements not to scale)
The corresponding external load is found on the basis of the equilibrium of moments, for
instance about point S. As stated before, it is assumed that the compressive forces in the
flanges are centric (distance to outer fibre is half the flange height). The forces acting on the
concrete rigid body are:
 A 10,5 MN compressive force at the centre of both top and bottom flange;
 a 1,51 MN vertical force at both kinks (at the deviator at the support and the deviator at
11 m from the support);
 a uniformly distributed load q;
 a vertical reaction force at the support (1/2 q l).
Bending moment equilibrium about point S results in:

 0, 25 0,15  1 1
1,51 11  10,5   2,5     2  2  33   2 q  33
1
 2 2 
so:
q = 0,299 MN/m = 299 kN/m
Note: In the previous calculation only the dominant bending moment contributions are
accounted for.
The stress in the ultimate fibre of a joint (at the support) is zero for q = 231 kN/m. The load at
which the joints are open appears to be q = 299 kN/m. The difference between both can be
explained as follows:
 Also at midspan the situation of zero stress in the joint must first be reached.
 The compressive forces in the concrete gradually shift towards the outer fibres (increased
internal lever arm).
At further increase of the load, the deflection increases rapidly and the tendons stretch. The
design value that follows from the characteristic tensile strength of the prestressing steel is
1860 / 1,1 = 1691 N/mm2. This stress is reached at a relatively high strain. It is therefore
assumed that the stress in ULS is somewhat lower: the stress in the tendons is assumed to
increase to the design value, which is related to fp0,1k and is about 90% of the design value
derived from the characteristic strength. The material factor γs has to be taken into account,
February 2019 13-18

which implies that the stress is assumed to increase to fp0,1k / γs = 0,90 · 1860 / 1,1 =
1522 N/mm2. Up to failure occurs, the tendon force can therefore increase with a value:
P = (1522 - 1000) · 10500 · 10-6 = 5,48 MN
Note that using a stress of 1522 N/mm2 implies that the prestressing steel is assumed to be
still in the linear elastic stage (just at the kink in the stress-strain relationship from EN 1992-
1-1 fig. 3.10).
In the kinematic model shown in fig. 13.26, with rigid moving parts, the tendon elongation is:
l = 2 w1 + w2 = 2 z1  + z2 2 (13.1)
where
 =  / ( 1 /2 l ) (13.2)
and
z1 = 2,50 – 0,50 – 0,075 = 1,925 m (13.3a)
z2 = 2,50 – 0,125 – 0,40 -   l / 6 = 1,975 -   l / 6 (13.3b)
It is now assumed that the failure mechanism from fig. 13.26 is active in all spans. This
implies that the increase of the strain of a tendon follows from the elongation over a span and
the span length.
Note: In case only one span would show a failure mechanism, the elongation of the tendon
has to be distributed over a much larger distance. The result is a much smaller strain increase.
From the equations (13.1) – (13.3) it follows that:
 42
l  15, 6 
l 3 l
Note that in this expression z1 and z2 are in m.
The increase of strain in the prestressing steel (which is assumed to be in the linear elastic
stage):
l E
   Es  Es  2s (15, 6   1,33  2 ) (13.4)
l l
To reach p = 522 N/mm2 = 522 MN/m2 (Ep = 195 · 103 MN/m2 and l = 33 m), it is found
that  = 0,190 m is required.
Note that the 2/ l component has hardly any impact on the result of the l expression:
15,6·0,19/33 = 0,090 m, whereas 4/3 / l2 = 0,001 m.
This implies that z2 = 1,975 m in (13.3b) would give almost the same elongation l.
February 2019 13-19

The ULS equilibrium situation is shown in fig. 13.28. The figure presents the forces exerted
on the concrete. The increase of the axial prestressing force is 5,48 MN. Note that the initial
prestressing force is 10,5 MN, which makes that the total prestressing force now is 15,98 MN.
δ = 0,190
0,79
10,5+5,48=15,98
10,5 + 5,48 = 15,98

½ql 11 m
16,5 m
Fig. 13.28 Equilibrium situation at the onset of failure (forces in MN, displacements in m)
The deflection at midspan is 0,190 m, which is relatively high when compared with the height
of the cross-section. The rigid body rotation from the deflection is therefore taken into
account, which influences two parameters:
1 – The angle over which the tendon is kinked at the deviation point.
2 – The vertical position of the horizontal forces exerted at midspan.
The deflection at the deviator at 11 m from the support is 11 / (0,5 · 33) = 2/3 of the deflection
at midspan:
2,5  0, 4  0,5  0,190  23

tan    0,157
11
which implies that θ = 0,156 rad = 8,9º and sinθ = 0,155.
The vertical reaction force at the deviator is:
P sin   10,5  5, 48   0,155  2, 48 MN
Bending moment equilibrium about the centre of the bottom flange at the support results in:
1
/2 q l · 1/4 l = 1/8 q l 2 = 15,98 · (2,50 - 0,25/2 - 0,15/2 - 0,190) + 11 · 2,48 = 61,0 MNm
In this expression only the dominant bending moment contributions are taken into account.
The first component after the = sign is the contribution of the (axial) tendon force
(15,98 MN); the second component is from the vertical component of the tendon force at the
deviator. It is noted that the internal lever arm between both compressive forces in the flanges
(bottom flange at support and top flange at midspan) is reduced by the deflection δ (0,19 m)
that is required to have the assumed increase of the tendon force.
February 2019 13-20

The corresponding uniformly distributed failure load:

61, 0  8
qu   0, 448 MN/m = 448 kN/m
33, 02
It is noted that a more extensive calculation might indicate that a higher value can be reached,
since, in the previous analysis, it was assumed that the prestressing steel stress at failure is
fp0,1k / γs = 0,90 ·1860 / 1,1 = 1522 N/mm2. There is still a reserve before fpk / γs = 1860 / 1,1 =
1691 N/mm2 is reached. However, the stress-strain relationship has a kink once fp0,1k is
reached, which implies that the "standard" Ep then can no longer be used. This implies that the
deformations must increase considerably before the end of the second linear part of the
prestressing steel stress – strain relationship is reached.
To find the global load-deflection relation (fig. 13.29), the deformation in the “uncracked
stage” (i.e. no joint opening) is calculated. For q = 231 kN/m1 the deflection is:
1 q l4 1 0, 231  334
    6,8 103 m  6,8 mm
384 EI 384 30 103  3,51
For q = 299 kN/m it is found that  = 8,8 mm when using the “uncracked stage” stiffness.
This is too optimistic since joint opening at the support then has already started (the joint in
the span is still closed at this load; opening this joint too requires app. 108 – 89 = 19 kN/m
additional loading).
500
450
400
350
q load [kN/m]
300
250
200
150
100
50
0
0 50 100 150 200
displacement [mm]
Fig. 13.29 Load-deflection relationship
The safety with regard to failure turns out to be very large. The requirement is:
qd =1,2qg + 1,5qq = 1,2 · 102 + 1,5 · 40 = 182 kN/m,
whereas 448 kN/m is obtained.
This high safety is especially due to the short free length of the tendon, which is assumed to
be only 33 m (cracks at all spans and supports are assumed, which makes that an elongation
February 2019 13-21

results in a relatively high strain increase), and the relatively high prestressing force, which
creates high compressive forces in the joints, see fig. 13.25 (right).
Bending moment equilibrium about the centre of the bottom flange at the support resulted in
the following expression:
1
/2 q l · 1/4 l = 1/8 q l 2 = 15,98 · (2,50 - 0,25/2 - 0,15/2 - 0,190) + 11 · 2,48 = 61,0 MNm
The following variables are introduced:

 The prestressing force in ULS is 15,98 MN = Pm + ∆P;
 the concentrated upward force at a deviator is 2, 48 MN=  Pm  P  sin  ;
 the span is 33,0 m = l;
 the internal lever arm at joint opening at the support in SLS is:
2,50 m – 0,50 m - 0,15 m / 2 = dsupport;
 the internal lever arm at joint opening at midspan in SLS is:
2,50 m – 0,40 m - 0,25 m / 2 = dmidspan;
 the deflection at midspan in ULS is 0,19 m = .
The deflection at the deviator in ULS results in the following angle:
2,5  0, 4  0,5  0,190  23

tan    0,157
11
At a small angle , tan = sin = . The upward force at the deviator then is:
2, 48 MN =  Pm  P  
where
2,5  0, 4  0,5  0,190  23

11
The bending moment equilibrium expression is rewritten:
 0,15 0, 25 
1
8 ql 2 =  Pm  P   2,50 - 0,50 -  0,50  2,50  0, 40   2,50  0, 40   
 2 2 
 2,50  0, 40  0,50  23  
 13 l  Pm  P   
3l
1
 
After inserting the previously defined variables, the result is:
 0,15 0, 25 1 
1
8 ql 2 =  Pm  P   2,50 - 0,50 -  2,50  0, 40   3 
 2 2 
1
8 ql 2 =  Pm  P   dsup port  d midspan  13  
The expression contains the total bending moment from the external q-load ( 18 ql 2 ). The
external bending moment is resisted by both the support and midspan cross-section. It appears
that their internal lever arms as defined in the stage of joint opening (dsupport, dspan) must be
February 2019 13-22

reduced (  13  ) This reduction follows from the vertical displacement of some of the
deviators:
Rewriting the bending moment equilibrium expression, as presented before, is not that
obvious. Therefore, this expression will be derived by using a number of variables, see the
following figure. Thanks to symmetry, only half of a span has to be modelled.
Bending moment equilibrium around the position of the concrete compression zone force at
the support cross-section:
1 1 1
 Pm  P   zB  epB  epA  zA      Pm  P  sin(  )  l  ql  l
3 2 4
e e 
    pA pB 
1 1
l l
3 2
 Pm  P   zA  zB     ql 2
1 1
 3  8
where zA = dsupport and zB = dmidspan
Note that three bending moment components were not taken into account since they have only
minor impact (both a small force and a small lever arm).
The deviators present at the supports prevent vertical tendon displacement to occur. The
internal lever arm at the support therefore does not change compared with the lever arm at
initial joint opening. At midspan, however, there is no deviator; the deviators are at 5,5 m
from the midspan cross-section. A vertical displacement  at midspan implies a vertical
displacement of 23  at these deviators. The tendon follows the deviator and, as a result, there
is a 13  reduction of the midspan internal lever arm (the hinge in the midspan compression
zone has a vertical displacement , whereas the tendons has a displacement of only 23  ).
It is noted that this calculation is based on assumed plasticity, where plastic hinges develop at
the support and midspan cross-section and resist the total bending moment exerted.
Redistribution of bending moments from support to midspan or vice versa is assumed to
occur, which requires sufficient rotational capacity.
February 2019 13-23

Sensitivity to temperature gradients and differential settlements

Because of the unreinforced joints, the quality of the design of the structure also depends on
the behaviour in case of imposed deformations. It is assumed that, because of solar radiation,
an effective temperature gradient T = 10º C can occur over the height h of the cross-section
(fig. 13.30a). If a free deformation could occur, the curvature would be:
 T
 (13.5)
h
where:
 is the temperature expansion coefficient of concrete = 10-5 K-1

h is the total height of the cross-section (in the case study 2,5 m)
However, the curvature is restrained by a bending moment M. This moment follows directly
from the well-known relationship:
M
 (13.6)
EI
If the curvature is fully restrained, the corresponding bending moment M is:

EI  T
M 
h
Fig. 13.30 Imposed deformations by a temperature gradient (a) and differential settlement (b)
At a temperature cycle of about 12 hours (sun) the short term E-modulus is not valid (EN
1992-1-1 table 3.1). Therefore, a reduction of 15% on this E-modulus is applied.
The result is:
Ec,eff = 0,85 · 30000  25000 MN/m2
February 2019 13-24

So one finds:
25000  3,51 105 10

M   3,51 MNm
2,5
This corresponds with a stress variation in the cross-section equal to:
 top: ct = M / (I / yt) = 3,51 / (3,51 / 0,91) = -0,9 N/mm2 (compression)
 bottom: cb = M / (I / yb) = 3,51 / (3,51 / 1,59) = +1,6 N/mm2 (tension)
This implies that, due to this temperature gradient alone, no joint opening would occur
(compare the value 1,6 N/mm2 with the compressive stresses at the bottom of the beam from
fig. 13.25).
In the case of a differential settlement z (fig. 13.30b) in the uncracked stage, the bending
moment is:
6 EI  z
Mz 
l2
In general, settlements occur gradually. According to section 6.7.4 a gradual increase of the
imposed deformation results in a reduction of the magnitude of the bending moments.
The reduction factor is:

(13.7)
 (1   )
Therefore, a differential settlement of 10 mm would result in the following bending moment

at the support:
6  30000  3,51 0.01 

Mz  
33 2
 (1   )
With  =  = 3 and  = 0,8 one finds:
Mz = 1,7 MNm
which results in the following stress variations in the cross-section:
 top side: ct = 1,7 / (3,51 / 0,91) = 0,44 N/mm2

 bottom side: cb = 1,7 / (3,51 / 1,59) = 0,77 N/mm2
To have joint opening at the support in the unloaded situation (fig. 13.25; “support’, “G”,
bottom; -1,8 N/mm2), a differential settlement of (1,8 / 0,77) · 10 = 23 mm is required. In the
fully loaded situation (fig. 13.25; “support”, “G+Q”, bottom; -3,4 N/mm2), (3,4 / 0,77) · 10 =
44 mm leads to joint opening.
If the mean prestressing is lower, which is possible on the basis of the traffic load, see fig.
13.25, the differential settlement required to have joint opening, would be lower.
February 2019 13-25

At the bottom of the beam joint opening occurs already at a differential settlement of 23 mm
(unloaded situation). It should be noted that, if there is a significant joint opening, this causes
a reduction of the shear resistance of the structure. This shows that a good estimation of the
effect of imposed deformations is of major importance.
In a structure without segmental joints, where reinforcing steel can continue along the full
structure, the sensitivity to imposed deformation is much smaller.
13.6 Failure mechanisms of structures with unbonded tendons

The failure mechanism of structures with unbonded tendons was introduced in the previous
chapter. It appeared that the elongation of the tendons, which follows from the crack
formation in the structure, causes a strain increase over the full tendon length: the tendons are
unbonded and anchored at their ends only. Their strain increase is, therefore, often limited.
When compared with the theory from this chapter on external prestressing, it appears that in
ULS, the unbonded tendon behaves like an external prestressing tendon. However, there
exists a significant difference between both: the unbonded tendon is inside the concrete and
follows the deformation of the structure, whereas the external tendon is outside the concrete
and follows the structure's deformation only at its deviation points and anchors.
A structure that cracks is discontinuous at the location of the crack; there is locally a discrete
separation between the structural parts at both crack faces. As a result of this discontinuity, an
unbonded tendon that crosses the crack and first had a smooth (linear or curved) profile, now
has a discontinuous profile at the crack. This causes kinks in the tendon profile. This is
demonstrated in fig. 13.31 for simply supported beams with a linear and a curved tendon
profile. As discussed in chapter 12, the increase of the stress in the prestressing steel follows
from the crack width at the level of the tendon and the tendon length. The bending moment
resistance of this type of structure is now discussed.
Fig. 13.31 Failure mode of simply supported uniformly loaded beams containing linear (left)
and curved (right) unbonded tendons
The simply supported beam with linear tendons is discussed first, see fig. 13.32. The beam
has a rectangular cross-section. The tendon eccentricity relative to the centroidal axis is
denoted as ep. The beam is uniformly loaded to failure and cracks at midspan only.
February 2019 13-26

Fig. 13.32 Deformations and force equilibrium of a simply supported uniformly loaded beam
with linear unbonded tendons
At a deflection δ at midspan, the angle θ, the elongation of the tendon and the increase of
prestressing steel strain are:

 
0,5
 p  2  z
 p
 pm 

The increase of the prestressing force ΔP follows from the total steel strain εp,tot (initial strain
εpm∞ plus strain increase Δεp) and the stress-strain diagram of the steel. The total prestressing
steel force is denoted as Pm∞ + ΔP. At a relatively small angle θ (sin =  and cos = 1), the
horizontal and vertical component of the prestressing force can be expressed as indicated in
fig. 13.32b. The figure presents only the left half of the beam, including the forces acting on
it. Note that at the crack, there is only the vertical component of the prestressing force caused
by the kink in the tendon. The force at the anchor is split in a horizontal and a vertical
component.
The concrete compressive force at the assumed "hinge" in the concrete compression zone
follows from the horizontal component of the total prestressing force:
N cu  Pm  P
February 2019 13-27

The "hinge" is assumed to be at a distance ηh from the outermost concrete compression fibre.
Bending moment equilibrium about the support:
N cu  h  h      Pm  P    12 l  12 ql  14 l+  Pm  P   12 h  ep  (13.8)
1
8 ql 2 =  Pm  P   h  h      12 l  12 h+ep    Pm  P   12 h  h  ep 
At the left hand side of the = - sign is the total bending moment in the midspan cross-section
from the uniform load on the beam; at the right hand side is the internal bending moment in
this cross-section. It appears that the bending moment capacity is not reduced by the
deflection of the beam as should be the case when it is an external tendon. This becomes clear
when looking at fig. 13.32: The deflection of the beam not only influences the working line of
the concrete compressive force, but is also imposed on the tendons. These two aspects have
the same quantitative effects, but are opposite. There is only one effect from the deflection at
midspan, which is the increase of the prestressing force from tendon elongation.
The following example is similar to the previous one. The only difference is that the
unbonded tendons now have a parabolic profile, see fig. 13.33a. The tendons are anchored at
the level of the centroidal axis. The tendon drape is f.
Fig. 13.33 Deformations and force equilibrium of a simply supported uniformly loaded beam
with parabolic unbonded tendons. The beam is assumed to be cracked at midspan
February 2019 13-28

Again, the elongation of the tendons results in a prestressing steel strain increase:

 
0,5
 p  2  z
 p
 pm 

The total prestressing steel force at failure is again denoted as Pm∞ + ΔP. At their anchors, the
tendons are at an angle α relative to the centroidal axis of the beam. This angle follows from
the tendon profile. In case of a parabolic profile α = 4f / l if the angle α is relatively small. The
tendons are kinked over an angle θ at both crack faces. At small angles α and θ (sin(α + θ) =
α + θ and cos(α + θ) = 1) the horizontal and vertical component of the prestressing force can
be expressed as shown in fig. 13.33b. At the crack, there is again only the vertical component
of the prestressing force. At an anchor, the angle between the tendons and a horizontal line
has increased from α to α + θ.
The concrete compressive force at the assumed "hinge" in the concrete compression zone
follows from the horizontal component of the total prestressing force:
N cu  Pm  P
The "hinge" is again assumed to be at a distance ηh from the outermost concrete compression
fibre.
Figure 13.33b presents the left half of the beam. Forces exerted on the concrete are presented,
namely:
1. the uniformly distributed load on the beam;
2. the uniformly distributed upward curvature pressure from the tendons;
3. at the anchor: a horizontal and a vertical component of the prestressing force;
4. at the cracked cross-section: a horizontal concrete compressive force and a vertical force
caused by the kink in the tendon;
5. at the support: a vertical support reaction;
Some of these forces will be discussed in detail.
2:
The prestressing force has increased from Pm∞ to Pm∞ + ΔP. The curvature of the tendon
profile has not changed; the two parts in which it is split by the crack, have rotated only over
an angle θ. The curvature pressure is (< 0, which is upward):
Pm  P
qp  
R
The radius of curvature can be expressed in the drape f.
The curvature pressure is not vertical; it is at an angle α + θ relative to a vertical line at the
support, gradually reducing to an angle θ at midspan. The angles are small. As a result, the
February 2019 13-29

vertical component of the curvature pressure is equal to the curvature pressure. There is also a
horizontal component of the curvature pressure. This force follows from sin(α + θ) = α + θ at
the support, gradually reducing to sin(θ) = θ at midspan. The resulting horizontal force,
however, is relatively small and has a small lever arm since the lever arm is related to the
beam height and the vertical position of the tendon. It will therefore have a negligible
contribution to the bending moment equilibrium.
3:
Assume that the original tendon profile was at the supports at an angle α relative to the
centroidal axis. When the beam is not loaded, it is the angle relative to a horizontal line. The
deflection of the beam causes its two parts to rotate over an angle θ. The angle between a
horizontal line and the tendon profile now is α + θ at the anchors. The forces at an anchor:
horizontal:  Pm  P  cos        Pm  P 

vertical:  Pm  P  sin        Pm  P     
4:
The original tendon profile has no kinks. The crack causes the two beam halves to rotate
relative to each other. The tendon profile remains horizontal over the small distance where it
passes the crack. However, its original parabolic profile outside the crack does not change
(the tendons are in their ducts). As a result, there is a kink (angle θ) between the two parts of
the tendon profile just inside the beam and outside the beam, in the crack. The result is a
concentrated vertical force at each of the two crack faces:
 Pm  P  sin      Pm  P  

The kink also results in two concentrated horizontal force at the crack; one at each of the two
crack faces:
 Pm  P 1  cos  

This force is small (cos(θ) ≈ 0) and has a small lever arm in a bending moment equilibrium
analysis. The force is therefore not presented in figure 13.33.
5:
The vertical support reaction is in equilibrium with the load on the beam. Note that the
uniform upward load from prestressing does not result in a support reaction.
The resulting load from upward curvature pressure:
8 ( Pm  P) f 4 ( Pm  P) f

1
2  qP   12  2

 
Vertical force equilibrium will now be checked first.
Downward load at the anchor:

 Pm  P  sin        Pm  P     
February 2019 13-30

Upward load at the kink:

  Pm  P  sin       Pm  P  
Resulting load from upward curvature pressure:

8 ( Pm  P) f 4 ( Pm  P) f
1
2  qP   12  2

 
Since the original tendon profile is parabolic, the following well known relationship between
the angle at the anchor and the drape can be applied:
2f 4f
   f  14 
1
2  
Vertical force equilibrium then results in:
4 ( Pm  P) 14 
   Pm  P     Pm  P       0 ; OK

Note that it is a statically determinate beam. Prestressing then does not introduce vertical
support reaction forces. On the other hand, in case of a statically indeterminate structure,
vertical reactions forces are often introduced.
8 ( Pm  P) f 1
N cu  h  h      Pm  P    12 l+ 12    4 l  12 ql  14 l+  Pm  P   12 h (13.9)
2
1
8 ql 2 =  Pm  P   h  h      12 l  12 h+f    Pm  P   12 h  h  f 
Again, the deflection has no influence on the internal lever arm of forces, just as in case of the
beam with linear tendons. The tendons follow the vertical displacement of the beam and the
vertical distance between the concrete compressive force and the tendon remains unchanged.
The parabolic tendons are now applied in a statically indeterminate beam. The tendon profile
is modelled using upward curvatures only; the downward curvatures at the supports are
replaced by kinks, see fig. 13.34. The drape of the tendons is f, which is the distance between
the upper and lower point of the parabolic tendon profile. The angle between the tendon
profile and a horizontal line at the supports is again denoted as α. The distance from the
outermost concrete compression fibre to the resulting compressive force (at the "hinges") is
η1 h and η2 h at the midspan and support cross-section, respectively.
Figure 13.34 presents the forces that are relevant from the point of view of bending moment
equilibrium.
February 2019 13-31

Fig. 13.34 Statically indeterminate beam prestressed with unbonded tendons. The tendon
profile is assumed to have upward curvatures only. The beam is uniformly
distributed loaded and cracked at midspan and support cross-sections
8 ( Pm  P ) f 1
N cu  h  1h      Pm  P    12 l+ 12    4 l  12 ql  14 l+N cu 2 h (13.10)
2
1
8 ql 2 =  Pm  P   h  1h  2 h      12 l+f    Pm  P  h  1h  2 h  f 
As expected, the deflection does not influence the internal lever arms of forces. If the distance
from the top fibre of the beam to the tendons is denoted as y at the support, the distance
between the concrete compressive force and the tendons is:
 h  2 h  y  at a support cross-section (13.11)

and
 y+f  1h  at a midspan cross-section (13.12)
February 2019 13-32

The summation of both internal lever arms, namely  h  1h  2 h+f  , is used in expression
(13.10) which also contains the total bending moment from the external q-load that has to be
resisted by the support and midspan cross-section ( 18 ql 2 ). This indicates that this calculation
is based on assumed plasticity: plastic hinges develop at the support and midspan cross-
section and together, they must resist the total bending moment exerted. Redistribution of
bending moments from support to midspan or vice versa is assumed to occur. This implies
that there must be sufficient rotational capacity.
13.7 Application of external prestressing for the retrofitting of structures

An important application of external prestressing is strengthening and retrofitting of existing
structures. On the subject “maintenance”, Wicke [13.21] made an interesting statement: “In
the case of concrete structures one should, in general, not speak about repair, since this is not
done as well for most other matters in daily life. In the case of a machine or a car, one does
not speak about repair, but about service. Therefore, we should get used to considering repair
and maintenance activities not to be the result of a design error, or an implicit weakness of the
material concrete, but as a consequence of a natural wearing process”.
Within the scope of such “service” activities, Seltenhammer described the upgrading
operations for the Wangauer Bridge [13.20], see also Wicke [13.21]. As a second example of
the use of external prestressing, namely the renovation of the Ruhr Bridge Essen/Werder, will
be considered [13.15].
Wangauer Ach Bridge [13.20, 13.21]

The Wangauer Ach Bridge in Austria consists of two traffic lanes. Each lane is built with a
cross-section as shown in fig. 13.35. In the span, the cross-section is open at the bottom. Near
to the support, the cross-section is extended with a bottom flange. The prestressed upper
structure of the bridge is built as a continuous structure over a substantial number of piers,
with spans that vary between 25,0 and 41,25 m. The structure was built in the period 1962-
1964.
span support
axis
Fig. 13.35 Cross-section of the Wangauer Ach Bridge before refurbishment (dimensions m).
25 Years after its completion, it was decided to refurbish the bridge. Since meanwhile also the
traffic loads were increased and the design codes were changed, it was decided to provide
additional prestressing. Since it is almost impossible to provide additional bonded prestressing
in an existing structure, it was decided to use external prestressing.
The next decision was to apply the prestressing tendons axially, at the level of the gravity line
of the structure, and not to apply them according to the bending moment curve. This decision
was predominantly taken on the basis of technical arguments. Although a tendon profile, that
February 2019 13-33

is adapted to the distribution of moments along the structure, is most effective and economic
with regard to the use of prestressing steel, a problem is the introduction of the high
prestressing forces in the structure at the deviation points. To apply a saddle in a structure as
shown in fig. 13.35, is a difficult and expensive task. In a bridge having a length of 385 m, the
number of saddles, intermediate anchorages and couplers is very large. As shown in fig.
13.32, the tendons have been placed at the outside of each web. An additional prestressing
force of 5000 kN per web was applied on the structure (fig. 13.36).
refurbished
additional
prestressing
Fig. 13.36 External prestressing at the outside of the webs of the Wangauer Ach Bridge
Special attention was given to anchoring the tendons at the bridge ends. The tendons are
distributed over the height of the webs to have a more uniform load introduction and to create
sufficient space for the anchorages. To introduce the prestressing force in the structure and to
spread it over its width, a force (or a tensile tie) is required in transverse direction (compare
fig. 10.5 and 10.6). Therefore, the old transverse end girder was extended with an additional
new part. In this newly cast part, three additional tendons were placed to balance the splitting
force from load introduction (fig. 13.37).
existing longitudinal prestressing
additional longitudinal additional transverse

prestressing prestressing
additional concrete
Fig. 13.37 Strengthening of the transverse end beam in front view (top) and in horizontal
cross-section (bottom), including the position of the old and new prestressing
tendons.
Ruhr Bridge Essen-Werden [13.14]

This bridge continues over two spans (length 66,40 m and 47,00 m, respectively). The cross-
section consists of boxes. At the intermediate support the bridge has a width of 34,41 m and
four boxes. In the direction of the abutments, the width and the number of boxes increases,
see fig. 13.38.
February 2019 13-34

Crack formation had occurred in the bottom flange and in the beam webs of the largest span.
The cracking was that severe, that repair was inevitable because the post-tensioned steel in the
cracked area had to be protected against corrosion. Injecting the cracks (which locally showed
widths up to 0,4 mm) was not regarded as being a satisfactory solution, because calculations
demonstrated that these cracks were caused by temperature gradients. It was therefore
expected that, very soon after injection of the old cracks, new cracks would occur for the
same reason. Furthermore, it was shown that the stress variations in the post-tensioned steel
were far beyond values allowed. It was clear that measures were not only necessary for
corrosion protection, but from the point of view of structural safety as well. Also here,
strengthening with external straight tendons appeared to be the best solution.
longitudinal cross-section
cross-section
construction
joint
cracks in bottom flange
plan view
Fig. 13.38 Ruhr Bridge Essen-Werden
At the abutments, a total of 24 longitudinal tendons (VSL Type 5-16 with a breaking load of
2833 kN), each about 75 m long, were placed. At the abutments, they were anchored in
anchorage blocks that were cast at the longitudinal beam (web) ends, see fig. 13.39. Also the
February 2019 13-35

transverse beam at the intermediate support was provided with additional vertical prestressing
to sustain the forces from the external longitudinal prestressing tendons.
plan view
tendon
tendon
a detail A
detail B
anchor tendon
tendon anchor
tendon coupled with

anchor
existing tendon c
b
Fig. 13.39 Longitudinal external prestressing applied in the longest span (a), anchored in
newly cast in-situ parts of the longitudinal beams (b). Strengthening of the
transverse girder at the intermediate support by external vertical prestressing (c)
(dimensions mm)
13.8 Literature
13.1 Bruggeling, A.S.G.: “External Cables: State of the Art Report”, Conference on Partial
Prestressed Concrete Structures.
13.2 Combault, J.: “Evolution et développement des ponts modernes a précontrainte
totalement extérieure au béton”, FIP-Congres New Delhi, 1985.
13.3 Müller, J.: “Construction of the Long Key Bridge”, PCI-Journal, Nov-Dec. 1980.
13.4 Virlogeux, M.P.: “Die externe Vorspannung”, Beton- und Stahlbetonbau 83, Heft 5, p.
121-126.
13.5 Virlogeux, M.P.: “External prestressing: From Construction History to Modern
Technique and Technology”, Naaman, A., Breen, J. (eds.): External Prestressing in
Bridges, ACI-SP 120, Detroit 1990.
13.6 Virlogeux, M.P.: “La Précontrainte Extérieure. Le point de la question aujourd’hui. La
Conception et la Construction des Ponts a Précontrainte Extérieure au Béton”, Annales
de l’Institut Technique du Bâtiment et des Travaux Publics 1991, No. 498, p. 1-47,
No. 499, p. 1-75.
February 2019 13-36

13.7 Virlogeux, M.P.: “Comparison Between Cast-in-situ and Precast Segmental

Construction”, IABSE-Colloquium “Structural Concrete”, Stuttgart, 1991,
IABSE/IVBH-report Vol. 62, 1991, Zürich.
13.8 Virlogeux, M.P.: “External Prestressing – Historical and Modern application”, Conti,
E., Foure, B. (eds.): External Prestressing in Structures, AFPC-Workshop, Saint-
Remy-les-Chevreuse, 1993, p. 13-41.
13.9 Müller, J.: “Some Recent International Projects with External Prestressing”. Conti, E.,
Foure, B. (eds.): External Prestressing in Structures, AFPC-Workshop, Saint-Remy-
les-Chevreuse, 1993, p. 71-96.
13.10 Beaupre, R.J., Powell. L.C., Breen, J.E., Kreger, M.E.: “Deviator Behaviour and
Design for Externally Post-Tensioned Bridges”, Workshop in Paris, American
Concrete Institute, ACI-SP 120, Detroit 1990.
13.11 Naaman, A., Breen, J.: “External Prestressing in Bridges”, American Concrete
Institute ACI-SP 120, Detroit 1990.
13.12 Wollmann, G.P., Kreger, M.E., Roberts-Wollaman, C.L., Breen, J.E.: “External
Anchorage in Diaphragms and Intermediate Slab Blisters”, Conti, E., Foure, B. (eds.):
External Prestressing Structures, AFPC-Workshop, Saint-Remy-les-Chevreuse 1993,
p. 113-122.
13.13 Nelissen, M.G.P.: “Doorbraak bij de toepassing van uitwendige voorspankabels”,
Cement 1987, No. 12, p. 56-59.
13.14 Bruggeling, A.S.G.: “Uitwendige voorspanning van bruggen”, Deel I, Cement 1994,
nr. 9, p. 52-60 & deel II, Cement 1994, nr. 11, p. 75-81.
13.15 Eibl, J., Ivanyi, G., Buschmeyer, W. und Kobler, G.: “Vorspannung ohne Verbund:
Technik und Anwendung”, Betonkalender 1995, Teil II, Ernst & Sohn, Berlin.
13.16 Aeberhard, H.U., Buergi. P., Ganz, H.R., Marti, P., Matt, P., Sieber, T.: “External Post
Tensioning”, VSL Report Series, Bern, 1990.
13.17 ENV 1992-1-5, Eurocode 2. Ontwerp en berekening van betonconstructies. Deel 1-5:
Algemene regels. Constructies met voorspanning zonder aanhechting en uitwendige
voorspanwapening, eerste druk, dec. 1994.
13.18 Girmscheid, G.: “Spannbeton-Hochstraße in Bangkok – Planung und Ausführung”,
Beton- und Stahlbetonbau 88 (1993), Ernst und Sohn, Berlin, 1993, p. 161-166.
13.19 Huang, J.: “Extern vorgespannte Segmentbrücken unter kombinierter Beanspruchung
aus Biegung, Querkraft und Torsion”, Dissertation & Heft 22 der Schriftenreihe des
Instituts für Massivbau und Baustofftechnologie der Universität Karlsruhe, 1994.
13.20 Seltenhammer, U.: “Außenliegende Vorspannung an der Wangauer Achbrücke”,
Symposium “Verstärken von Betontragwerken durch Vorspannung”, Heft 10,
Schriftenreihe des Österreichischen Betonvereins, p. 28-33.
13.21 Wicke, M.: Symposium “Beton, Eeuwig Jong”, Betondispuut 25 jaar, Proceedings.
13.22 Vermeulen, G.P.C., Kaptijn, N., van der Veen, C.: “Uitwendige voorspanning in
schuifbruggen”, Cement 1993, nr. 12, p. 46-51.
February 2019 13-37

February 2019 13-38

CIE4160 Textbook February 2019

Uploaded by

Copyright:

Available Formats

CIE4160 Textbook February 2019

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

CIE4160 Textbook February 2019

Uploaded by

Copyright:

Available Formats

CIE3150/4160

Prof.dr.ir.Dr.-Ing.h.c. J.C. Walraven

prof.dr.ir. J.C. Walraven

2. Materials and methods of prestressing

3. Centrically prestressed structures

4. Prestressing of structures subjected to bending

5. Statically indeterminate structures

6. Shrinkage, creep and relaxation

9. Crack width control

11. Partial prestressing

12. Prestressing with unbonded tendons

13. External prestressing

Consider a statically determinate simply supported beam of a homogeneous, linear-elastic

εct σct (-)

εcb σcb (+)

Fig. 1.2 Behaviour of reinforced concrete after cracking

April 2018 1-1

Crack formation at SLS can be prevented by applying an eccentric compressive force Pm on

April 2018 1-2

1.2 Application of the prestressing force

Method 1: Prestressing between fixed points

jack element pre-tensioned steel

April 2018 1-3

Method 3a: Post-tensioning with bonded tendons

Method 3b: Post-tensioning with unbonded tendons

1.3 Curved tendon profiles

April 2018 1-4

unloaded structure fully loaded structure

Fig. 1.5 Stress in the cross-section at midspan and support, respectively

section A-A section B-B

Fig. 1.6 Parabolic tendon profile

April 2018 1-5

Fig. 1.7 Tendon layout when several tendons are applied

Fig. 1.8 Kinked tendon profile applied with pre-tensioned steel

1.4 Prestressed concrete versus reinforced concrete

April 2018 1-6

Fig. 1.10 Prestressed offshore gravity-based structure, built at Rotterdam (1992)

April 2018 1-7

April 2018 1-8

To summarise: the technique of prestressing has considerably increased the opportunities

April 2018 1-9

April 2018 1-10

2. Materials and methods of prestressing

A member having a length of 10 m is pre-tensioned with reinforcing steel B500

steel bar in untensioned state

(pre)tensioning of the steel

casting of the concrete

elastic shortening (Δlce) of the

final state condition after shrinkage

Fig. 2.1 Stress-losses if reinforcing steel is applied to axially prestress a concrete

Before releasing the bar, the steel strain is:

April 2018 2-1

The corresponding total elongation of the reinforcing steel:

ls  1, 75 103 10 103  17,5 mm

lcs  0,3 103 10 103  3, 0 mm

This high steel quality is achieved by:

Several types of prestressing steel are available.

April 2018 2-2

Bars having a diameter of Ø15 to Ø50 mm

Table 2.1 Properties of prestressing bars (EN 10138-4)