You are on page 1of 15

Engineering Structures 174 (2018) 111–125

Contents lists available at ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

Improved prediction of long-term prestress loss in unbonded prestressed T


concrete members

Pablo M. Páez , Beradi Sensale
Facultad de Ingeniería, Universidad de la República, Julio Herrera y Reissig 565, Montevideo, Uruguay

A R T I C LE I N FO A B S T R A C T

Keywords: In this paper, an approach based on the finite element method to model the behavior of concrete beams with
Undonded prestressing unbonded prestressing steel over time, and an improved simplified equation for calculating the prestress losses in
Long-term loads statically determinate prestressed concrete members with unbonded internal tendons are proposed. Both
Serviceability methods take into account the effects of the concrete creep, concrete shrinkage, prestressing steel relaxation and
Prestress loss
the presence of the bonded non-prestressed reinforcement. In addition, a generalization of the Step-by-Step and
the Age-Adjusted Effective Modulus methods for the time cross-sectional analysis of prestressed concrete
members with unbonded internal tendons is presented. The accuracy of the proposed equation is evaluated
based on the results of previous studies and is contrasted with the computational implementation of the Step-by-
Step Method. As the main conclusions, we mention that the results show that the proposed equation adequately
predicts the prestress loss and has higher accuracy compared to simplified existing models.

1. Introduction member as a whole. In other words, the stress in the tendon in an un-
bonded prestressed concrete member subjected to external loads is
The term “unbonded prestressed” is used for prestressed concrete member-dependent instead of section-dependent [1].
members in which there is no bonding between the concrete and the Without loss of generality, we can establish that from the 1960s,
prestressing steel or, if it exists, it is so small that a perfect bond be- different researchers have placed greater emphasis on the study of the
tween the two materials cannot be considered. Unbonded prestressing behavior of structures with unbonded prestressing, in particular the
can be divided into two different types: internal and external pre- behavior of beams and slabs [2]. Thereafter, most research has focused
stressing. In internal prestressing, the tendons are embedded in the on predicting the behavior of these elements at ultimate strength
concrete, such as in the case of beams and post-tensioned flat slabs, [1–17]; however, there are relatively few research works focused on
whereas in external prestressing, the tendons are not embedded in the predicting the behavior of members in bending under short-term ser-
concrete, such as in box girder bridges. vice loads [12,14–16,19] and even fewer that allow prediction of the
The behavior of the members with internal prestressing is char- behavior in bending under long-term service loads [11,15,18,20].
acterized by the fact that the position of the tendon at any section does To study the time-dependent behavior of unbonded prestressed
not change with the deformation of the member, whereas in external concrete members under service load, CEB-FIP [18] establishes an
prestressing, when the element is deformed, the position of the tendon equation for estimating the prestress loss based on the Rate of Creep
is conditioned by the displacement of both the anchor and the deviator Method [21,22]. This method is based on the assumption that the rate
points of the tendon. of change of creep with time is independent of the age at loading;
The analysis of prestressed concrete members with unbonded pre- however, new researches show that this hypothesis is not correct [23].
stressing tendons is inherently more complex than the analysis of Gauvreau [11] proposes a simplified equation to estimate the loss of
members with bonded prestressing tendons, since, unlike bonded pre- prestress due to time effects in members with unbonded internal ten-
stressing tendons, there is no strain compatibility between the tendons dons. However, this equation does not take into account the relaxation
and the surrounding concrete, which means that the prestressing ten- of prestressing steel or the presence of non-prestressed reinforcement.
dons and concrete can move with respect to each other. Then, the stress For long-term analysis, Lou et al. [15] use a formulation based on the
in unbonded tendons will depend on the deformation of the structural finite element method. Although the methodology used by the authors


Corresponding author.
E-mail addresses: ppaez@fing.edu.uy (P.M. Páez), sensale@fing.edu.uy (B. Sensale).

https://doi.org/10.1016/j.engstruct.2018.07.038
Received 21 November 2017; Received in revised form 7 June 2018; Accepted 14 July 2018
Available online 26 July 2018
0141-0296/ © 2018 Elsevier Ltd. All rights reserved.
P.M. Páez, B. Sensale Engineering Structures 174 (2018) 111–125

for the time-dependent analysis can be called “step by step”, meaning Therefore, the stress–strain relationship for concrete in compression can
that it is done by dividing the time interval into small intervals, the be expressed as Eq. (1):
creep compliance function used is not the one recommended by the σc = Ec εe (1)
different codes but its form has the advantage of facilitating program-
ming with respect to the simulation of the history of applied stress. On where σc is the concrete stress produced by a compression strain εe and
the other hand, the use of the proposed procedure by the Finite Element Ec is the modulus of elasticity of concrete. The modulus of elasticity of
Method requires a high degree of computational programming. Guo concrete can be calculated from the models provided in the different
et al. [20] propose a model to predict the time-dependent losses in standards, for example, that provided by the MC10 [25].
prestressed concrete members. This model takes into account the effect The total concrete strain at time t in an uncracked concrete member
of shrinkage, creep of concrete, and relaxation of steel. However, al- uniaxially loaded at constant temperature may be expressed as the sum
though the authors use the method for either bonded or unbonded of the instantaneous strain, εe (t ) , the creep strain between the times t0
prestressing, the method is based on the bond hypothesis and therefore and t , εcr (t , t0 ) , and the shrinkage strain between the time ts and t ,
its use for unbonded prestressed members is not entirely correct. εsh (t , ts ) , with t0 the age of concrete at loading and ts the age of the
In view of current trends of the standards based on the “Limit States concrete at the beginning of drying [26] (Eq. (2)):
Design”, it is just as important to predict the behavior of the members ε (t ) = εe (t ) + εcr (t , t0) + εsh (t , ts ) (2)
under service load conditions, in particular under long-term service
loads, as it is to predict the behavior of the unbonded prestressed Shrinkage strain in concrete can be defined as the time-dependent
members at ultimate flexural strength. It is important to estimate the strain of a concrete specimen in a given environment, where displace-
prestress losses as accurately as possible, since an incorrect estimation ment of the specimen is not restricted and the specimen is not subject to
can cause serviceability problems; for example, the tensile strength of external load [27]. Shrinkage strain can be calculated from the model
concrete can be exceeded under service loads and/or can lead to ex- provided by MC10 [25].
cessive camber [24]. If a concrete specimen is subjected to a compressive stress at time
In most practical applications with prestressing, the members are t = t0 , σc (t0) , which is kept constant in time, the creep strain at time
calculated so that, under the action of long-term loads, all sections are t > t0 is given by Eq. (3):
in uncracked state (State I); that is, the tensile stress in the extreme fiber σc (t0)
of concrete has not reached its modulus of rupture, and therefore the εcr (t , t0) = φ (t , t0)
Ec (t0) (3)
creep deformations of the concrete will tend to be much greater than in
the case of partially prestressed concrete members. The effects of creep, where φ (t , t0) is the creep coefficient and Ec (t0) is the modulus of
shrinkage, and relaxation of the steel will cause a redistribution of elasticity of concrete at time t = t0 . The value of the creep coefficient
stresses between the steel and concrete, which will generally result in a can be calculated from the model proposed by MC10 [25].
loss of the prestressing force and therefore an increase in the de- If the magnitude of stress varies with time, which is usual in con-
formations. crete structures, creep strain can be obtained from the principle of su-
This paper is motivated by the need to predict the behavior of perposition. The principle of superposition was applied to concrete for
prestressed concrete members with unbonded internal tendons under the first time by McHenry [28] and requires the fulfillment of certain
long-term loads. The main objective of this work is to propose a sim- hypotheses to obtain sufficiently accurate results. These hypotheses are
plified equation to calculate the prestress losses in statically determi- often referred to as linearity assumptions [29] and can be stated as
nate prestressed concrete members with unbonded internal tendons, follows: (a) the stress in the concrete is less than 40% of the char-
taking into account the effects of concrete creep, concrete shrinkage, acteristic compressive strength, fck ; (b) ε is not decreasing; that is, strain
prestressing steel relaxation, and the presence of the bonded non-pre- of decreasing magnitude does not take place; (c) there is no significant
stressed reinforcement. The second objective of this work is to develop change in moisture content; and (d) there are no sudden changes in
the formulation of the Step-by-Step Method (SSM) and the Age- stress. In most practical applications of civil engineering, these as-
Adjusted Effective Modulus Method (AAEM) for long-term analysis of sumptions are fulfilled, so the principle of superposition is usually ac-
prestressed beams with unbonded internal tendons; to date, there is no cepted in most analyses and calculations. The strain in concrete at time
literature or research work that has developed such a formulation. t caused by a given stress history is obtained from Eq. (4):
Finally, an approach based on the finite element method (FEM) to σc (t0) τ=t 1 + φ (t , τ )
model the behavior of concrete beams with unbonded prestressing steel ε (t ) =
Ec (t0)
(1 + φ (t , t0)) + ∫τ=t 0 Ec (τ )
dσc (τ ) + εsh (t , ts )
(4)
over time is proposed. The originality of this technique lies in its
treatment of long-term effects. Also in this paper, a computational al- If the change in stress between t0 and t , Δσc (t ) , is known, Eq. (4) can be
gorithm based on the Step-by-Step Method from the basic principles of written in simplified form [30–32] as Eq. (5):
the Strength of Materials will be implemented (without using Finite σc (t0) Δσc (t )
Element Method). The accuracy of the proposed equation was eval- ε (t ) = (1 + φ (t , t0)) + + εsh (t , ts )
Ec (t0) Ec (t , t0) (5)
uated based on the results of previous studies and was contrasted with
the computational implementation of the Step-by-Step Method. The where Ec (t , t0) is the age-adjusted effective modulus, whose expression is
results show that the proposed equation adequately predicts the pres- given by Eq. (6):
tress loss and has higher accuracy compared to existing simplified Ec (t0)
formulations. Ec (t , t0) =
1 + χ (t , t0) φ (t , t0) (6)

2. Constitutive equations of materials: instantaneous and time- χ (t , t0) is the aging coefficient, whose expression is given by Eq. (7) [31]:
dependent behavior
σc (t0) 1 Ec (t0) 1
χ (t , t0) = − = −
σc (t0)−σc (t ) φ (t , t0) Ec (t0)−ER (t , t0) φ (t , t0) (7)
2.1. Concrete
where ER (t , t0) is the relaxation function of the concrete, defined as the
For normal strength concretes, that is, for concretes whose char- stress at time t due to a unit strain applied at time t0 and kept constant
acteristic compressive cylinder strength at 28 days is less than 50 MPa, throughout the period t0 to t . Values for the relaxation function can be
a linear elastic behavior in compression can be considered for stress calculated by applying Eq. (4) to the case of a constant unit strain
levels below about 40% of the characteristic cylinder strength. history beginning at t0 or they can be calculated from the approximate

112
P.M. Páez, B. Sensale Engineering Structures 174 (2018) 111–125

formula proposed by Bazant et al. [33]. Np = Ap σp (εp) (14)


where Ap and σp (εp) are the cross-section and the stress–strain function
2.2. Non-prestressed reinforcement
of the prestressing steel, respectively. The strain in unbonded pre-
stressed steel, εp , can be written as Eq. (15):
Under service loads, the non-prestressed steel reinforcement usually
works at relatively low stress and therefore its behavior will be linear εp = εp,0 + Δlp/ lp,0 (15)
elastic for any time. The stress–strain relationship for non-prestressed
where εp,0 is the initial strain of the tendon, Δlp is the change in tendon
steel can be expressed by Eq. (8):
length, and lp,0 is its initial length. The initial length is measured from a
σs = Es εs (8) reference state of deformations.
Under the assumptions of Euler-Bernoulli beam theory, we can as-
where Es is the modulus of elasticity of the non-prestressed steel.
sume that for unbonded internal prestressing, the distance dp (x ) mea-
sured from the top surface of the beam to the centroid of the pre-
2.3. Prestressing steel
stressing steel does not vary with the external load, where x is the
distance between one end and the cross-section of the beam being
In general, prestressing steels have a linear elastic behavior up to
analyzed. On the other hand, for beams commonly used in bridges and
approximately 70% of their characteristic yield strength (stress at a
buildings or for other prestressed structural members, the angle of in-
strain of 0.1%) and from there they present a non-linear behavior.
clination of the tendon α (x ) is small and therefore the tendon elonga-
However, the different standards allow the use of idealized stress–strain
tion Δlp can be written as Eq. (16) [11]:
diagrams, assuming a linear elastic behavior for stresses up to the
characteristic yield stress [25] (Eq. (9)): x=l
Δlp = ∫x=0 Δεc, p (x ) dx (16)
σp = Ep εp (9)
where l is the length of the beam between anchor points and Δεc, p (x ) is
where Ep is the modulus of elasticity of the prestressing steel. the average change in strain in the concrete at the level of the unbonded
In contrast to non-prestressed steels, prestressing steels usually work prestressed steel at any section x along the span; its expression is given
at a high stress level in service, which will result in a relaxation process. by Eq. (17):
Therefore for any time t > t0 , the stress–strain relationship for the
prestressing steel is expressed as Eq. (10): Δεc, p (x ) = εc, p (x )−εc, p,0 (x ) (17)

σp (t0)−Δσpr (t −t0) where εc, p (x ) is the absolute strain in concrete at the level of unbonded
εp (t ) = prestressing steel and εc, p,0 (x ) is the strain in concrete for the reference
Ep (10)
state at the level of unbonded prestressing steel.
where Δσpr (t −t0) is the intrinsic relaxation, that is, the loss of tension of
the steel at constant length and constant temperature (although 4. Time analysis of prestressed concrete beams with unbonded
Δσpr (t −t0) represents a reduction in tension, it has been considered a internal tendons: proposed equation to calculate the loss of
positive quantity). However, for prestressed concrete members, the prestress
steel relaxation will be a process of variable length, and therefore the
prestress loss should be less than in the case of intrinsic relaxation. One Consider a statically determinate prestressed concrete beam with
of the models used for the calculation of the steel relaxation at variable unbonded internal tendons such that the cross-section of the beam is
length is the relaxation coefficient χr ; the loss of stress at variable symmetrical about the z -axis, the orthogonal y -axis passes through its
length is expressed by [34] Eq. (11): centroid, and the bending moment is applied about the y -axis (Fig. 1).
Δσpr (t −t0) = χr Δσpr (t −t0) Consider that the prestressing force and all long-term external loads are
(11)
applied at time t = t0 and that under these conditions the member is in
where Δσpr (t −t0) is the reduced relaxation. The relaxation coefficient is State I, that is, uncracked. Let us also suppose that the prestressing steel
given by Eq. (12): and the non-prestressed steel reinforcement are placed sufficiently close
2 to one another in the cross-section that they can be considered to be
ξ =1 λ (1−Ωξ )−0.40 ⎞
χr = ∫ξ =0 (1−Ωξ ) ⎛⎜

λ−0.4
⎟ dξ
(12)
concentrated in one concrete fiber.
⎝ ⎠ Concrete creep, concrete shrinkage, and steel relaxation generate a
Ω and λ are given by Eq. (13): change in the redistribution of stresses between concrete and steel,
prestressed and non-prestressed. This change can be considered as a
Δσps (t − t0) − Δσpr (t − t0)
Ω= σp,0
redistribution of the self-equilibrating state of stress induced by pre-
σp,0 stressing; in addition, under the established assumptions, the sum of the
λ= f ptk (13)
where σp,0 is the initial prestressing stress, fptk is the ultimate tensile
strength, and Δσps (t −t0) is the change in stress in the prestressed steel
due to the combined effects of creep, shrinkage, and relaxation.
Therefore, it is necessary to perform an iterative process to determine
the value of the relaxation coefficient.

3. Compatibility conditions for unbonded prestressing steel

The behavior of prestressed concrete members with unbonded in-


ternal prestressing is characterized by the absence of strain compat-
ibility between the prestressing steel and the surrounding concrete. If
we assume perfect unbonding between the tendon and concrete, the
prestressing force Np will be constant along its length between the an- Fig. 1. Cross-section: statically determinate prestressed concrete beam with
chor points (Eq. (14)): unbonded internal prestressing and non-prestressed reinforcement.

113
P.M. Páez, B. Sensale Engineering Structures 174 (2018) 111–125

changes in the forces of the concrete and the steel must be null [34,35] σc, p,0 (x ) 1 ecp (x )2 ΔP (t )
Es As φ (t , t0) + Es As εsh (t , t0)−Es As ⎛ A + ⎞ p
Ec (t0) Ic
(Eq. (18)):
ΔPs (t , x ) = ⎝ c ⎠ Ec (t , t0)
Es As ecp (x )2
ΔFc (t ) = −ΔPrs (t ) = −ΔPp (t )−ΔPs (t ) (18) 1+ ⎛1 + ⎞
Ec (t , t0) Ac Ic
⎝ ⎠
where ΔFc (t ) is the change in the resultant compressive force in the (24)
concrete, ΔPrs (t ) is the change in steel force (prestressed and non-pre- Taking into account that ecp (x ) = ecp= constant, lp = l ; substituting Eqs.
stressed steel), ΔPp (t ) is the change in the prestressing force, and ΔPs (t ) (18) and (24) into Eq. (21), and equating Eqs. (21) and (20), we obtain
is the change in the reinforcing steel force. Eq. (25):
On the other hand, it can be shown that the change in stress in
1 φ (t , t0) x=l Δσpr
concrete at the level of the prestressing steel Δσc, p (t , x ) is [35] (Eq. l Ec (t0)
η (t , t0) ∫x = 0 σc, p,0 (x ) dx + η (t , t0) εsh (t , t0)− Ep
(19)): ΔPp (t ) =
1 ω ω2Es As
Ep Ap
+ Ec (t , t0)
− ωEs As ⎞
ΔFc (t , x ) ΔFc (t , x ) Ec (t , t0)2 ⎛1 +
Δσc, p (t , x ) = σc, p (t , x )−σc, p (t0, x ) = + ecp (x )2 ⎝ Ec (t , t 0) ⎠ (25)
Ac Ic (19)
where ω and η (t , t0) are given by Eqs. (26) and (27), respectively.
where Ac and Ic are the area and moment of inertia of the concrete 2
1 ecp
cross-section and ecp (x ) is the distance between the centroid of the ω= +
concrete cross-section and the centroid of the prestressing steel. Ac Ic (26)

ωEs As
4.1. Statically determinate beams with straight tendons and constant η (t , t0) = 1−
reinforcing steel
Ec (t , t0) 1 + ( Es As ω
Ec (t , t0) ) (27)
The value of ΔPp (t ) obtained by Eq. (25) is a negative quantity.
Consider that the profile of the unbonded prestressing tendons is
straight, that the reinforcing steel is constant along the beam, and that
4.2. Generic profile of the tendons: statically determinate structures
the prestressing steel and the non-prestressed steel reinforcement are
placed sufficiently close together in the section that they can be con-
When the profile of the tendons is generic, the distance between the
sidered to be concentrated in one concrete fiber. Then, the change in
centroid of the concrete cross-section and the centroid of the prestres-
strain of the prestressing steel over time can be expressed by Eq. (20):
sing steel is no longer constant; moreover, it cannot be considered that
ΔPp (t ) Δσpr Δlp (t ) the steel reinforcement and prestressing steel are located sufficiently
Δεp (t ) = + =
Ep Ap Ep lp (20) close together in the section and therefore some of the arguments set
out above are no longer valid.
Δσpr has been considered to be a positive quantity. Neglecting the non-prestressing steel reinforcement and taking into
The change in length of the unbonded prestressing steel can be account that ΔFc (t ) = −ΔPp (t , x ) (Eq. (18)), and then equating Eqs. (20)
obtained from Eq. (16). Then, applying the Age-Adjusted Effective and (21), we obtain Eq. (28):
Modulus Method to a concrete fiber coincident with the location of the Δσpr
1 φ (t , t0) x=l
steel, the change in length of the prestressing steel can be written as Eq. l Ec (t0)
∫x = 0 σc, p,0 (x ) dx + εsh (t , t0)− Ep
(21): ΔPp (t ) =
1 1 x=l ecp (x )2
+ ∫ ⎛1 + ⎞ dx
Ep Ap l·Ec (t , t0) x=0 Ac Ic (28)
x=l σc, p,0 (x ) x=l ⎝ ⎠
Δlp (t ) = ∫x=0 Ec (t0)
φ (t , t0) dx + ∫x=0 εsh (t , t0) dx
2
x=l
⎛⎜ 1 + ecp (x ) ⎞⎟ ΔFc (t , x ) dx 5. Generalization of the step-by-step and age-adjusted effective
+ ∫x=0 modulus methods
⎝ Ac Ic ⎠ Ec (t , t0) (21)

where σc, p,0 (x ) is the initial stress in the concrete at the same fiber as the In order to validate the proposed simplified equations, the time
centroid of the steel, accounting for the long-term external actions and cross-sectional analysis methodology is generalized for application to
prestressing. the analysis of prestressed concrete beams with unbonded internal
Taking into account the strain compatibility between non-pre- prestressing.
stressed reinforcement and concrete at the same fiber as the centroid of
the steel, the change of the strain in the concrete will be equal to the 5.1. Step-by-step method
change of the strain in the non-prestressed reinforcing steel (Eq. (22)):

Δεs (t , x ) = Δεc, p (t , x ) Consider again a prestressed statically determinate concrete beam


with unbonded internal prestressing. The cross-section of the beam is
Δσs (t , x ) σc, p,0 (x ) ecp (x )2
=
1
φ (t , t0) + εsh (t , t0) + ⎛ A + ⎞ ΔFc (t , x ) symmetrical about the z -axis and the origin O is an arbitrarily chosen
Es Ec (t0) Ic
⎝ c ⎠ Ec (t , t0) (22) reference point on the axis of symmetry. Suppose that the beam has ms
If the area of the non-prestressed reinforcing steel is As , the change in layers of non-prestressed reinforcing steel and mp layers of unbonded
force in the non-prestressed steel reinforcement is prestressing steel (Fig. 2). Consider that under the action of the pre-
ΔPs (t , x ) = As Δσs (t , x ) . Then, taking into account Eqs. (18) and (22), stressing force and all long-term external loads the member is un-
ΔPs (t , x ) can be written as Eq. (23): cracked.
The stress–strain relationships of concrete and steel for times t = t0
σc, p,0 (x ) and t = t j are:
ΔPs (t , x ) = Es As φ (t , t0) + Es As εsh (t , t0)
Ec (t0) For t = t0 (Eq. (29)),
1 ecp (x )2 ⎞ (ΔPs (t , x ) + ΔPp (t ))
−Es As ⎜⎛ + ⎟
σc,0 (x ) = Ec,0 ε0
⎝ Ac Ic ⎠ Ec (t , t0) (23) σs (i),0 (x ) = Es (i) εs,0
Δlp (i),0
Rearranging, the change in the force in the non-prestressing steel re- σp (i),0 (x ) = Ep (i) ⎛ + εp (i),0⎞
lp,0(i) (29)
inforcement is given by Eq. (24): ⎝ ⎠

114
P.M. Páez, B. Sensale Engineering Structures 174 (2018) 111–125

ε (t j, x ) = εr (t j, x ) + z·κ (t j, x ) (35)

where εr (t j, x ) and κ (t j, x ) are the strain at the reference point O and the
curvature, respectively.
From Eq. (32) and taking into account the relation given in Eq. (35),
the axial force resisted by the concrete in a section located at a distance
x and at time t j is given by Eq. (36):


Nc (t j, x ) = ∫A c
σc (t j, x ) dA = ∫A c ⎢Ec (t j )(εr (t j, x ) + z·κ (t j, x )−εsh (t j ))

i=j−1

+ ∑ Fe (t j, ti ) σc (ti, x ) ⎥ dA = Ac (x ) Ec (t j ) εr (t j, x )
i=0 ⎦
+ Bc (x ) Ec (t j ) κ (t j, x )−Ac (x ) Ec (t j ) εsh (t j )
i=j−1
+ ∑ Fe (t j, ti ) Nc (ti, x )
Fig. 2. Generic cross-section and arrangement of reinforcement in a prestressed i=0 (36)
concrete beam.
where Ac (x ) and Bc (x ) are the area of the concrete and its first order
Using the “rectangular rule” for numerical integration, Eq. (4) can be moment about the reference axis Oy , respectively.
rewritten as follows [36] (Eq. (30)): The axial force resisted by the reinforcing steel in a section located
at a distance x and at time t j is given by Eq. (37):
i=j
ε (t j ) = J (t j, t0) σc (t0) + ∑ J (t j, ti )Δσc (ti ) + εsh (t , ts ) i = ms (x ) i = ms (x )
i=1 (30) Ns (t j, x ) = ∑ As (i) (x ) Es (i) εr (t j, x ) + ∑ z s (i) As (i) (x ) Es (i) κ (t j, x )
i=1 i=1
where Δσc (ti ) = σc (ti )−σc (ti − 1) and J (t j, ti ) represents the creep function
calculated at time t j relative to a unit strain applied at time ti ; its ex- (37)
pression is given by Eq. (31): and the axial force resisted by the unbonded prestressing steel is given
1 + φ (t j, ti ) by Eq. (38):
J (t j, ti ) =
Ec (ti ) (31) i = mp (x )
Δlp (i) (t j )
where Ec (ti ) is the instantaneous elastic modulus of the concrete at time Np (t j, x ) = ∑ Ap (i) (x ) Ep (i) ⎜⎛ + εp (i),0−εp, rel (i) (t j )⎞⎟
ti . i=1 ⎝ lp,0(i) ⎠ (38)
Thus, for time t = t j , the constitutive relations of the materials can
The bending moment resisted by the concrete in a section located at a
be written as Eq. (32):
distance x and at time t j is given by Eq. (39):
i=j−1
σc (t j, x ) = Ec (t j )(ε (t j, x )−εsh (t j )) + ∑ Fe (t j, ti ) σc (ti, x )

i=0 Mc (t j, x ) = ∫A c
z·σc (t j, x ) dA = ∫A c
z ⎢Ec (t j )(εr (t j, x ) + z·κ (t j, x )−εsh (t j ))
σs (i) (t j, x ) = Es (i) εs (t j, x ) ⎣
i=j−1
Δlp (i) (t j ) ⎤
σp (i) (t j, x ) = Ep (i) ⎛ lp,0(i)
+ εp (i),0−εp, rel (i) (t j )⎞
(32)
+ ∑ Fe (t j, ti ) σc (ti, x ) ⎥ dA = Bc (x ) Ec (t j ) εr (t j, x )
⎝ ⎠ i=0 ⎦
where Fe (t j, ti ) = (J (t j, ti + 1)−J (t j, ti ))/ J (t j, t j ) . + Ic (x ) Ec (t j ) κ (t j, x )−Bc (x ) Ec (t j ) εsh (t j )
If Next (t j, x ) and Mext (t j, x ) are the external axial force and bending i=j−1
moment applied in a section located at a distance x from the end of the + ∑ Fe (t j, ti ) Mc (ti, x )
beam at time t j , and Nint (t j, x ) and Mint (t j, x ) , respectively, are the in- i=0 (39)
ternal axial force and bending moment in the same section and at the
same time with respect to reference coordinate axes, the equilibrium where Ic (x ) is the second moment of the concrete section about the
equations are expressed as Eq. (33): reference axis Oy .
The bending moment resisted by the reinforcing steel in a section
Nint (t j, x ) = Next (t j, x ) located at a distance x and at time t j is given by Eq. (40):
Mint (t j, x ) = Mext (t j, x ) (33)
i = ms (x )

The internal axial force and bending moment can be written as the sum Ms (t j, x ) = ∑ z s (i) As (i) (x ) Es (i) εr (t j, x )
i=1
of the axial forces and bending moments resisted by the concrete, the
i = ms (x )
reinforcement, and the unbonded tendons (Eq. (34)):
+ ∑ z s2(i) As (i) (x ) Es (i) κ (t j, x )
Nint (t j, x ) = Nc (t j, x ) + Ns (t j, x ) + Np (t j, x ) i=1 (40)
Mint (t j, x ) = Mc (t j, x ) + Ms (t j, x ) + Mp (t j, x ) (34) and the bending moment resisted by the unbonded prestressed steel is
where Nc (t j, x ) , Ns (t j, x ) , and Np (t j, x ) are the axial forces in section x given by Eq. (41):
and at time t j resisted by the concrete, the reinforcing steel, and the
i = mp (x )
unbonded prestressing steel, respectively, with respect to reference Δlp (i) (t j )
Mp (t j, x ) = ∑ z p (i) (x ) Ap (i) (x ) Ep (i) ⎛⎜ + εp (i),0−εp, rel (i) (t j )⎞⎟
coordinate axes, and Mc (t j, x ) , Ms (t j, x ) , and Mp (t j, x ) are the bending i=1 ⎝ lp,0(i) ⎠
moments resisted by said materials. (41)
Taking into account the Euler-Bernoulli hypothesis, the strain in a
concrete fiber at a distance z from the reference point O at section x and Substituting Eqs. (36)–(41) in Eq. (34), we can rewrite this equation as
at time t j is (Eq. (35)): follows (Eq. (42)):

115
P.M. Páez, B. Sensale Engineering Structures 174 (2018) 111–125

i = mp (x )
⎡ Next (t j, x ) ⎤− ∑ ⎡ Np (i),0 (x ) ⎤ = ⎡ RA (t j, x ) RB (t j, x ) ⎤ ⎡ εr (t j, x ) ⎤ ⎡ εr (t , x ) ⎤ = 1 ⎡ RI (t , x ) − RB (t , x ) ⎤ ⎛ ⎡ Next (t , x ) ⎤
⎢ Mext (t j, x ) ⎥ ⎢ Mp (i),0 (x ) ⎥ ⎢ RB (t j, x ) RI (t j, x ) ⎥ ⎢ κ (t j, x ) ⎥ ⎢ ⎜ Mext (t , x ) ⎥
⎣ ⎦ i=1 ⎣ ⎦ ⎣ ⎦⎣ ⎦ ⎣ κ (t , x ) ⎥⎦ ⎣− RB (t , x ) RA (t , x ) ⎥
(RA (t ) RI (t )−RB (t )2) ⎢ ⎦⎝⎢ ⎣ ⎦
i=j−1 i = mp (x )
Nc (ti, x ) ⎤
+ ∑ Fe (t j, ti ) ⎡ − ∑ ⎡ Np (i),0 (x ) ⎤−F (t , t ) ⎡ Nc (t0, x ) ⎤

⎣ c (ti, x ) ⎥
M ⎦ ⎢ Mp (i),0 (x ) ⎥ e, TB 0
⎢ ⎥
i=0
i=1 ⎣ ⎦ ⎣ Mc (t0, x ) ⎦
Ac (x ) ⎤ i = mp (x )
−Ec (t j ) εsh (t j ) ⎡ Ac (x ) ⎤ 1

⎣ Bc (x ) ⎥
⎦ + Ec (t ) εsh (t ) ⎡ + ∑ Ap (i) (x ) Ep (i) εp, rel (i) (t ) ⎡ z ⎤
⎢ Bc (x ) ⎦
⎣ ⎥ i=1 ⎣ p (i) ⎦
i = mp (x )
1 i = mp (x )
− ∑ Ap (i) (x ) Ep (i) εp, rel (i) (t j ) ⎡ y ⎤ Δlp (i) (t ) ⎞
⎣ (i) ⎦
p − ∑ Ap (i) (x ) Ep (i) ⎡ 1 ⎤
lp,0(i) ⎣ z p (i) ⎦ ⎟
i=1
i = mp (x ) i=1 ⎠ (48)
Δlp (i) (t j )
+ ∑ Ap (i) (x ) Ep (i) ⎡z1 ⎤
lp,0(i) ⎣ p (i) ⎦ The axial rigidity, the stiffness related to the first moment of area, and
i=1
the flexural rigidity are given now by Eq. (49):
(42)
i = ms (x )
where RA (t j, x ) , RB (t j, x ) , and RI (t j, x ) represent the axial rigidity, the RA (t , x ) = Ac (x ) Ec (t ) + ∑ As (i) (x ) Es (i)
stiffness related to the first moment of area, and the flexural rigidity, i=1
respectively, about the reference axes. Their expressions are given by i = ms (x )
Eq. (43): RB (t , x ) = Bc (x ) Ec (t ) + ∑ z s (i) As (i) (x ) Es (i)
i=1
i = ms (x ) i = ms (x )
RA (t j, x ) = Ac (x ) Ec (t j ) + ∑ As (i) (x ) Es (i) RI (t , x ) = Ic (x ) Ec (t ) + ∑ z s2(i) As (i) (x ) Es (i)
i=1 i=1 (49)
i = ms (x )
RB (t j, x ) = Bc (x ) Ec (t j ) + ∑ z s (i) As (i) (x ) Es (i)
i=1 6. Implementation of a computational algorithm to obtain the loss
i = ms (x ) of prestress in statically determinate unbonded prestressed
RI (t j, x ) = Ic (x ) Ec (t j ) + ∑ z s2(i) As (i) (x ) Es (i) concrete beams by means of the step-by-step method
i=1 (43)

The elongation of the unbonded tendon located in the i -th layer at time The steps of the computational procedure are as follows:
t j is obtained by Eq. (44):
1. The following input data are collected: the cross-section geometry
x=l x=l
Δlp (i) (t j ) = ∫x=0 (εr (t j, x ) + z p (i) ·κ (t j, x )) dx − ∫x=0 (εr ,0, ref (x ) and span of the beam, material properties, external load, initial and
final times, environmental conditions, and so on.
+ z p (i) ·κ 0, ref (x )) dx (44) 2. The time interval (tm−t0 ) is divided into m time intervals by the
Then, the strain at the reference point O and the curvature in a section series given by Eq. (50) [36]:
located at a distance x from the end of the beam are obtained from Eq. t1 = t0 +
(tm − t0)
m·tm
(45):
t j = t0 + (m ·tm)1/(m − 1) ·(t j − 1−t0) (50)
⎡ εr (t j, x ) ⎤
⎢ κ (t j, x ) ⎥
⎣ ⎦ 3. For each time j (t = t j ), the values of the shrinkage of concrete and

=
1 ⎡ RI (t j, x ) − RB (t j, x ) ⎤ ⎛ ⎡ Next (t j, x ) ⎤ the relaxation of the steel are obtained. For this purpose, specific
(RA (t j ) RI (t j )−RB (t j )2) ⎢ − ( , ) RA (t j, x ) ⎥ ⎜ ⎢ ⎥ procedures are used, which we have called “shrinkage” and “re-
⎣ RB t j x ⎦ ⎝ ⎣ Mext (t j, x ) ⎦
i = mp (x ) i=j−1
laxation”.
Np (i),0 (x ) ⎤ ⎡ Nc (ti, x ) ⎤ 4. The matrices J (t j, ti ) and Fe (t j, ti ) are obtained. These matrices have
− ∑ ⎡ ⎢ Mp (i),0 (x ) ⎥− ∑ Fe (t j, ti ) ⎢ Mc (ti, x ) ⎥
i=1 ⎣ ⎦ i=0 ⎣ ⎦ dimensions of m + 1 × m + 1.
i = mp (x ) 5. The beam is divided into n + 1 sections.
Ac (x ) ⎤ 1
+ Ec (t j ) εsh (t j ) ⎡ + ∑ Ap (i) (x ) Ep (i) εp, rel (i) (t j ) ⎡ z ⎤ 6. At t = t0 :
⎣ Bc (x ) ⎥
⎢ ⎦ ⎣ p (i) ⎦
i=1 6.1. At the time of prestressing: the initial length of the tendon lp,0 ,
i = mp (x )
Δlp (i) (t j ) ⎞ is obtained.
− ∑ Ap (i) (x ) Ep (i) ⎡z1 ⎤ 6.1.1. For each of the sections, the bending moment due to the
i=1 lp,0(i) ⎣ p (i) ⎦ ⎟ (45)
⎠ self-weight is calculated.
6.1.2. The geometric properties Ac (x ) , Bc (x ) , and Ic (x ) and the
cross-section rigidities RA (t j, x ) , RB (t j, x ) and RI (t j, x ) are
5.2. Age-adjusted effective modulus method
calculated.
6.1.3. At this time, the prestressing force is known, Np = Np,0 .
The time-dependent constitutive relationship for the concrete in
6.1.4. For each of the sections, the strain in the concrete at the
time given by Eq. (32) can be rewritten using the Age-Adjusted Effective
level of the prestressing steel is calculated. Then they are
Modulus Method for any time t greater than t0 [36] (Eq. (46)):
integrated to obtain lp,0 with lp,0 = l−Δlp,0 , where Δlp,0 is
σc (t , x ) = Ec (t )(ε (t , x )−εsh (t )) + Fe, TB (t , t0)·σc (t0, x ) (46) the change in the length of the beam at the level of the
centroid of prestressing steel due to the self-weight and
where Fe, TB (t , t0) is given by Eq. (47), the effective prestress.
χ (t , t0)−1 6.2. When applying the long-term external load:
Fe, TB (t , t0) = φ (t , t0) ⎜⎛ ⎞⎟
6.2.1. For each of the sections, the bending moments due to
⎝ 1 + χ (t , t0) φ (t , t0) ⎠ (47)
long-term external loads, including the self-weight, are
Following an analogous procedure to that described in the previous calculated.
section, Eq. (45) can be rewritten as Eq. (48): 6.2.2. An initial value of Np , with Np = Np,inic , is taken. In

116
P.M. Páez, B. Sensale Engineering Structures 174 (2018) 111–125

particular, it starts with Np,inic = Np,0 . 7.1. Short-term analysis: Stiffness matrix of the Navier-Bernoulli beam
6.2.3. From Eq. (45), the strains in the concrete at the level of element and the reinforcement steel
the prestressing steel are calculated and are integrated
(by numerical methods) to obtain the elongation of the By using Hermitian shape functions to model the vertical displace-
tendon, Δlp . For t j = t0 , Fe (t0, t0) = [0] and εp, rel (i) (t0) = 0 ; ment of the concrete beam element and linear shape functions for the
in most practical applications, it is usual to consider that axial displacement, with reference to local axes, the stiffness matrix of
εsh (t0) = 0 . the beam element [Kb(e) ] can be written as Eq. (51) [37,38]:
6.2.4. The strain in the prestressing steel is calculated, E A
E A
(Δlp / l − Δlp,0 / l) ⎡ c(e) c 0 0 − c(e) c 0 0 ⎤
εp = εp,0 + (1 − Δlp,0 / l)
, and the prestressing force Np is ⎢ l l

12 E I 6 E I 12 E c I c 6 E c Ic ⎥
obtained. ⎢ 0 c c c c
0 − (e) 3
(e ) )3 (e ) )2 ( l ) (e ) )2
6.2.5. If the new value of Np is different from the value of Np at ⎢ (l (l (l ⎥
⎢ 0 6Ec Ic 4Ec Ic
0 −
6Ec Ic 2Ec Ic ⎥
the beginning of the process, Np,inic (6.2.2), by less than a ⎢ (
(l )e ) 2 l (e ) (l )(e ) 2
l (e ) ⎥
certain tolerance, Np is found and the process is con- [Kb(e)] = ⎢ E A ⎥
c c Ec Ac
⎢− l(e) 0 0 0 0 ⎥
sidered finished. The values of Nc (t0, x ) and Mc (t0, x ) are ⎢ l (e )

12Ec Ic 6Ec Ic 12 E I 6 E c Ic
stored. ⎢ 0 − (e) 3 − (e) 2 0 (e
c c
) 3
− (e) 2 ⎥
⎢ ( l ) (l ) (l ) (l ) ⎥
6.2.6. Otherwise, Np,inic = Np , where Np is the value obtained in
⎢ 0 6Ec Ic 2Ec Ic
0 −
6Ec Ic 4Ec Ic ⎥
Step 6.2.5, and the process is repeated until con- ⎢ (l (e ) )2 l (e ) (l )(e ) 2
l (e ) ⎥ (51)
⎣ ⎦
vergence.
7. At t > t0 , the external load remains constant between t0 and tm . The reinforcing steel is treated by assembling truss elements. The
7.1. For each of the m remaining time intervals, the relative rigid- stiffness matrix of a truss element with two degrees of freedom is given
ities RA (t j, x ) , RB (t j, x ) , and RI (t j, x ) are calculated and, taking by Eq. (52) [37]:
into account the time-dependent variables, the procedure of Es As 1 − 1
Step 6.2 is applied. For t = t j , Np is obtained. [Ks(e)] = ⎡ ⎤
l (e) ⎣− 1 1 ⎦ (52)
7.2. With the value of Np obtained in the previous step, the re-
laxation value is adjusted (Eqs. (11)–(13)) and the process is According to the plane-section assumptions, the contribution of the
repeated (Step 7.1). truss element to the stiffness of the beam element can be taken into
7.3. For t = tm , the final prestressing force is obtained. account using Eq. (53):

[Ks(e) ]transf = [Ts]T [Ks(e) ][Ts] (53)

7. Finite element approach where [Ts] is the transformation matrix, obtained from the geometric
relationships between the displacements. An expression for this is given
In this section, an approach is presented that uses the finite element in Eq. (54):
method to model the behavior of concrete beams with unbonded pre- 1 0 es 0 0 0 ⎤
stressing steel over time, based on the assumptions made in Section 4 of [Ts] = ⎡
⎣ 0 0 0 1 0 − es ⎦ (54)
this work. An exception is made whereby the prestressing steel and the
non-prestressed steel reinforcement can be placed anywhere within the where es is the distance from the centroid of the beam to the centroid of
concrete section. the reinforcement steel, as shown in Fig. 3. Then, [Ks(e) ]transf is given by
Unlike methods proposed by other authors, in which the concrete Eq. (55):
and reinforcing steel are modeled as a single beam element or plane
⎡ 1 0 − es − 1 0 es ⎤
frame element [12,15], in this approach, the concrete beam with un-
⎢ 0 0 0 0 0 0 ⎥
bonded internal tendons and reinforcing steel is modeled by assembling
Es As ⎢− es 0 es2 es 0 − es2 ⎥
straight beam elements, which simulate the concrete section, and truss [Ks(e) ]transf = (e) ⎢ ⎥
elements, which simulate the reinforcement steel, while the unbonded
l ⎢− 1 0 es 1 0 − es ⎥
⎢ 0 0 0 0 0 0 ⎥
internal tendons are modeled by means of equivalent nodal forces. ⎢ e
⎣ s 0 − es2 − es 0 es2 ⎥ ⎦ (55)
The originality of this technique lies in its treatment of long-term
effects. A finite element approach is proposed based on the time-de- Eq. (55) assumes that the non-prestressed steel reinforcement is hor-
pendent constitutive relationship for the concrete given in Eq. (46). The izontal and that it is placed in a single layer, although different layers
concrete creep, concrete shrinkage and relaxation of the prestressing can be modeled, creating as many matrices [Ks(e) ]transf as there are layers.
steel within the time period t0 to t are taken into account by applying a Then, the stiffness matrix of the reinforced concrete beam element
system of external forces that are conveniently chosen, and by mod- can be written as Eq. (56):
ifying the stiffness matrix of the beam element.
[K (e) ] = [Kb(e)] + [Ks(e) ]transf (56)

Fig. 3. Structural model for the finite element approach.

117
P.M. Páez, B. Sensale Engineering Structures 174 (2018) 111–125

Fig. 4. Nodal variables of the Navier-Bernoulli beam, truss and tendon elements, and the relationship between them.

7.2. Unbonded prestressing steel where [D (e) ] is the vector of nodal displacements of the beam element
and [T2] is a transformation matrix. Expressions for these are given in
Let us consider a beam element with unbonded prestressing and Eqs. (59) and (60), respectively:
reinforcing steel, as shown in Fig. 3. x i(,gu) , yi(,gu) , x j(,gu) and yj(,gu) are the
global coordinates of nodes i and j of the i -th beam element in the [D (e) ] = [ui(e) wi(e) θi(e) uj(e) wj(e) θj(e) ]T (59)
undeformed position, respectively, and x i(g,̂ u) , y i(g,̂ u) , x j(g,̂ )u and y j(g,̂ )u are the
global coordinates of the nodes i ̂ and j ̂ of the tendon segment, re- ⎡1 0 − ecp, i 0 0 0 ⎤
spectively (Fig. 4). ⎢0 1 0 0 0 0 ⎥
[T2] = ⎢
To take into account the slip between the concrete and the tendon, 0 0 0 1 0 − ecp, j ⎥
⎢ ⎥
the unbonded prestressing steel is modeled in terms of equivalent nodal ⎣0 0 0 0 1 0 ⎦ (60)
loads [12]. The action of the prestressing tendon on the beam element
can be taken into account using Eq. (57): The strain of the prestressing steel is given by Eq. (61) (Section 6, Point
6.2.4):
[P (e) ]
(Δlp / l − Δlp,0 / l)
= −Np [−cos(α (e) ) −sin(α (e) ) ecp, i cos(α (e) ) cos(α (e) ) sin(α (e) ) εp = εp,0 + (1 − Δlp,0 / l) (61)
−ecp, j cos(α (e) )]T
and taking into account Eqs. (16) and (17), the change in tendon length
(57) is given by Eq. (62):
where [P (e) ] is the vector of the equivalent nodal forces due to pre- i=m
stressing, Np = Np (εp) is the prestressing force and α (e) is the angle of Δlp = ∑ lp(e, i)−lp(e,0,) i
inclination of the tendon, as shown in Fig. 4. i=1 (62)
Under the action of external loads and prestressing, the element is
deformed. If ui(e) , wi(e), θi(e) , uj(e) , wj(e) and θj(e) are the nodal displacements where m is the number of beam elements, lp(e,0,) i is the initial length of the
of the i -th beam element, the deformed positions of nodes i ̂ and j ,̂ tendon segment (the distance between nodes i ̂ and j )̂ and lp(e, i) is its
deformed length.
x i(g,̂ d) , y i(g,̂ d) , x j(g,̂ )d and y j(g,̂ )d , can be obtained from geometric relationships
Finally, for each step of the load an iterative procedure is used to
(the plane-section assumption is used), as shown in Fig. 4 and Eq. (58):
find the prestressing force, Np (εp)
(g )
⎡ x i ,̂ d ⎤ ⎡ x i(,gu) ⎤
⎢ (g ) ⎥ ⎢ ⎥
( g )
⎢ y i ,̂ d ⎥ ⎢ yi, u + ecp, i ⎥ 7.3. Long-term analysis using the age-adjusted effective modulus method
⎢ ⎥=⎢ (e )
⎥ + [T2][D ]
⎢ x (ĝ ) ⎥ ⎢ x (g )
j , u ⎥ Eq. (46) can be rewritten as Eq. (63), where the first term in par-
⎢ j ,d ⎥ ⎢ ⎥
⎢ y (g ) ⎥ ⎢ yj(,gu) + ecp, j ⎥ entheses represents the stress generated by the axial strain and the
j ̂d
, ⎣ ⎦ (58)
⎣ ⎦ second term in parentheses represents the stress generated by bending.

118
P.M. Páez, B. Sensale Engineering Structures 174 (2018) 111–125

σc (t , x ) = [Ec (t )(εr (t , x )−εsh (t , t0)) + Fe, TB (t , t0 )·Ec (t0 )·εr (t0, x )] The first and last terms in parentheses in Eq. (70) represent the axial
stiffness matrix for the element, considering the age-adjusted effective
+ [Ec (t )·z·κ (t , x ) + Fe, TB (t , t0)·Ec (t0)·z·κ (t0, x )] (63)
modulus of elasticity Ec (t ) , and the axial stiffness matrix for the element
Let us consider in the first instance the stresses and the strains gener- at time t0 .
ated by the bending of the beam. The principle of virtual work applied The virtual work of the internal stresses must be equal to the virtual
to the internal stresses of the beam element can be expressed as in Eq. work of the external forces, as in Eq. (71),
(64),
∫V δε (t, x )·σc (t, x ) dV = [δaa(e) (t )]T [qa(e) (t )]−[δaa(e) (t )]T [f a(e) (t0)] (71)
∫V δε (t, x )·σc (t, x ) dV = ∫V z·δκ (t, x )[Ec (t )·z·κ (t, x ) + Fe,TB (t, t0)·Ec (t0)· where [f a(e)
(t0)] and [qa(e)
(t )] are the vector of equivalent nodal forces
z·κ (t0, x )] dV = ∫ Ec (t ) Ic δκ (t , x ) κ (t , x ) dx
(e ) due to the axial loads on the element applied at time t0 and the vector of
l
equilibrating nodal forces at time t , respectively.
+ ∫ Fe, TB (t , t0) Ec (t0) Ic δκ (t , x ) κ (t0, x ) dx
(e ) Then, Eqs. (67) and (71) can be written as Eq. (72),
l

(64) Ec (t )
[K (e) (t0)][D (e) (t )] = −Fe, TB (t , t0)[Kb(e) (t0)][D (e) (t0 )]
The curvature can be written as in Eq. (65) [38], Ec (t0) b

6ξ (−1 + 3ξ ) 6ξ (1 + 3ξ ) ⎤ + Ec (t ) Ac εsh (t , t0)[−1 1]T + [q (t )(e) ]


κ (t , ξ ) = ⎡ (e) 2 − (e) 2 ·

⎣ (l ) l (e ) (l ) l (e) ⎥⎦ −[f (t0)(e) ] (72)
[wi(e) (t ) θi(e) (t ) wj(e) (t ) θj(e) (t )]T = [Bf ][af(e) (t )] (65) where [f (e)
(t0)] and [q(e)
(t )] are the vector of the equivalent nodal forces
due to the axial and transverse loads on the element applied at time t0
where ξ is a normalized local coordinate for the element (ξ ∈ [−1, 1]), and the vector of equilibrating nodal forces at time t , respectively.
[Bf ] is the curvature matrix for the element and [af(e) (t )] is the nodal The stiffness matrix for the reinforced concrete beam element at
displacement vector in bending at time t . Then, the virtual work of the time t can be written as Eq. (73):
internal stresses can be expressed as in Eq. (66),
Ec (t )
[K (e) (t )] = [K (e) (t0)] + [Ks(e) ]transf
+1 [Bf ]T [Bf ] Ec (t ) Ic l (e) Ec (t0) b (73)
∫V δε (t, x )·σc (t, x ) dV = [δaf(e) (t )]T ⎛ ⎜ ∫−1 dξ ⎞ [a (e) (t )]
⎟ f
⎝ 2 ⎠ The relaxation of the prestressing steel can be taken into account by
+ Fe, TB (t , t0)[δaf(e) (t )]T applying an equivalent nodal forces vector, as given by Eq. (74):

⎛ +1 [Bf ]T [Bf ] Ec (t0) Ic l (e) (e )


[ΔPpr ] = Δσpr (t −t0)·Ap ·
⎜ ∫−1 dξ ⎞⎟ [af(e) (t0)]
⎝ 2 ⎠ (66) [−cos(α (e) ) −sin(α (e) ) ecp, i cos(α (e) ) cos(α (e) ) sin(α (e) )
The expressions in parentheses correspond to the stiffness matrix for the −ecp, j cos(α (e) )]T
beam element considering the age-adjusted effective modulus of elas- (74)
ticity Ec (t ) (that is, replacing Ec (t0) with Ec (t ) in Eq. (51)) and the
stiffness matrix for the element in bending at time t0 , respectively.
8. Application
The virtual work of the internal stresses must be equal to the virtual
work of the external forces, as shown in Eq. (67):
Example 1: Concrete beam with a rectangular cross-section and straight
∫V δε (t , x )·σc (t , x ) dV = [δaf(e) (t )]T [qf(e) (t )]−[δaf(e) (t )]T [f f(e) (t0 )] (67)
tendon

where [f f(e)
(t0)] and [qf(e)
(t )] are the vector of the equivalent nodal forces Lan et al. [39] tested a series of prestressed concrete beams with
due to the transverse loads on the element applied at time t0 and the bonded and unbonded prestressing tendons to verify the concept of
vector of equilibrating nodal forces at time t , respectively. control of prestress loss through the use of smart steel strands. One of
Let us now consider the stresses generated by axial deformations of the beams with unbonded prestressing is used to show the validity of
the beam element. By applying the principle of virtual work to the in- the proposed equations (Fig. 5). The beam has a rectangular cross-
ternal stresses, Eq. (68) can be obtained: section of 100 × 200 mm and its span is 2000 mm. The prestressing
tendon is straight and has a nominal diameter of 15.24 mm
∫V δε (t, x )·σc (t, x ) dV = ∫V δεr (t, x )[Ec (t )·εr (t, x )−Ec (t )·εsh (t, t0) ( Ap = 141.88 mm2), an ultimate tensile strength of 1660 MPa, and a
+ Fe, TB (t , t0 )·Ec (t0 )·εr (t0, x )] dV (68) modulus of elasticity of 195 GPa, and the initial prestress is
σp,0 = 0.5·fptk . The ordinary steel reinforcement consists of four bars
The strain can be written as in Eq. (69) [37], with a diameter of 12 mm and a yield stress of 335 MPa. The nominal
1 1 cubic compressive strength of the concrete is 30 MPa (characteristic
εr (t , ξ ) = ⎡− (e) ⎤·[ui(e) (t ) uj(e) (t )]T = [Ba][aa(e) (t )]
⎣ l l (e) ⎦ (69)
where [Ba] is the strain matrix for the element (t )] is the vector and [aa(e)
of axial nodal displacements at time t . Then, the virtual work of the
internal stresses can be expressed as in Eq. (70),
+1 [Ba]T [Ba] Ec (t ) Ac l (e) ⎞ (e)
∫V δε (t, x )·σc (t, x ) dV = [δaa(e) (t )]T ⎛∫−1 ⎜
2
dξ [aa (t )]⎟

⎝ ⎠
+1 [Ba]T Ec (t ) Ac εsh (t , t0) l (e) ⎞
−[δaa(e) (t )]T ⎛ ∫−1
⎜ dξ ⎟

⎝ 2 ⎠
+1 [Ba]T [Ba] Ec (t0) Ac l (e) ⎞
+ Fe, TB [δaa(e) (t )]T ⎛ ⎜ ∫−1 dξ ⎟

⎝ 2 ⎠
[aa(e) (t0)] (70) Fig. 5. Cross-section of the beam tested by Lan et al. [39].

119
P.M. Páez, B. Sensale Engineering Structures 174 (2018) 111–125

Table 1 sections are small. On the other hand, although for this particular case
Parameters of the concrete used to calculate the prestress loss in the concrete the value obtained from the loss of prestressing with PE presents a
beam of Lan et al. [39]. significant difference from the measured value (18.5%), it is a better
Standard φ (t , t0) χ (t , t0) εsh (t , t0) [10−6] Ec (t0) [GPa] estimator compared to the equation proposed by Gauvreau [11]
(38.9%).
CEB-FIB [25] 0.853 0.921 −130.46 31.19
ACI 209.2R [40] 0.708 0.992 −160.96 27.49
Example 2: Concrete beam with a rectangular cross-section and curved
tendon profile
cylinder strength of 24 MPa). The beam was cured for 28 days at room
In order to monitor the time-dependent prestress loss, Guo et al.
temperature and then prestressed.
[20] tested a series of concrete beams with bonded prestressing and one
Under the action of the self-weight of the beam, the loss of prestress
with unbonded prestressing, called PC2. The beam was made with
in the tendon was measured up to 300 h, resulting in a prestress loss at
concrete of a nominal cubic compressive strength of 55 MPa (char-
t = 300 hs of Δσp = 46 MPa.
acteristic cylinder strength of 44 MPa) and has a rectangular cross-
Table 1 shows the values of the creep coefficient, shrinkage, aging
section of 200 × 300 mm while the span is 4200 mm (Fig. 6). The
coefficient and modulus of elasticity of the concrete, calculated ac-
prestressing tendons have a nominal tensile strength of 1860 MPa and
cording to the CEB-FIB [25] and the ACI 209.2R [40] standards; the
low relaxation and the nominal area of the cross-section is 140 mm2.
aging coefficient is calculated according to Eq. (7). The loss of prestress
Ordinary reinforcing steel with a diameter of 12 mm and yield strength
is calculated using these parameters. Table 2 shows the calculated va-
of 335 MPa was used.
lues of the prestress loss according to several approaches used in this
The profile of the unbonded tendon is defined by Eq. (75):
work, distinguishing between those that are valid for unbonded pre-
stressing and those that are valid for bonded prestressing. dp (x ) = −(4.5918·10−5 (x −1400)2) + 240 for 0 ⩽ x < 1400 mm
A cement with strength class 42.5 N, a mean temperature of 20 °C
dp (x ) = 240 for 1400 ⩽ x < 2800 mm
and a relative humidity of the ambient environment (RH) of 75% was
considered. The remaining parameters used to calculate the creep dp (x ) = −(4.5918·10−5·x 2−25.7143·10−2·x + 360) + 240for 2800 ⩽ x ⩽
coefficient and the shrinkage according to the ACI 209.2R [40] were a 4200 mm
cement content of 300 kg/m3, a fine aggregate percentage of 42% and a (75)
slump of 60 mm. These parameters were estimated according to ACI
211.1 [41] to give the characteristic cylinder strength of 24 MPa where dp (x ) is measured from the top surface of the beam and x is the
mentioned above. distance on the axis of the beam measured from the support.
Using the parameters of CEB-FIB [25], on applying the SSM, a The prestressing force is 103.9 kN and was applied at the age of
prestress loss of Δσp = 47.2 MPa is obtained, whereas on applying the 30 days. The beam was subjected to concentrated loads of 3.125 kN
AAEM and the FEM, the prestress losses are Δσp = 46.7 MPa and applied at the one-third and two-third points of the span immediately
Δσp = 46.4 MPa, respectively. If the proposed equation (PE) (Eq. (25)) is after prestressing.
applied, the prestress loss is Δσp = 54.5 MPa, which represents a dif- The prestress loss measured by the authors at t = 395 days and the
ference of the order of 18.5% from the measured value. If the Gauvreau values calculated by various methods based on the CEB-FIB [25] and
equation [11] is applied, the prestress loss is Δσp = 63.9 MPa, which the ACI 209.2R [40] standards for the parameters of the concrete are
represents a difference of the order of 38.9% (Table 2). A similar ana- shown in Table 3. The parameters of the concrete are shown in Table 4.
lysis can be made in relation to the parameters calculated according to A strength class of cement of 42.5 N, a mean temperature of 20 °C and
ACI 209.2R [40]. an RH of 73% were used. The remaining parameters used to calculate
The PE considers that the bottom reinforcement steel is sufficiently the creep coefficient and the shrinkage were estimated according to the
close to the unbonded prestressing steel that the assumption ds (2) = dp ACI 211.4R standard [43] to achieve the characteristic cylinder
can be used. In the previous example, if the SSM is applied with strength (high-range water-reducing admixtures were used, with a ce-
ds (2) = dp = 140 mm, we obtain Δσp = 51.8 MPa, which represents a ment content of 410 kg/m3, a fine aggregate percentage of 38%, a
difference of the order of 5.21% compared to the PE. If we also assume water-to-cement ratio of 0.42 and a slump of 50 mm).
As (1) = 0 , we obtain Δσp = 52.1 MPa, which represents a difference of According to the parameters used by the CEB-FIB [25] standard, the
the order of 4.61%. difference between the measured prestress loss and that calculated by
If we apply the formula proposed in Eurocode 2 (EC 2) [42] and the SSM is of the order of 0.93%; for AAEM, this is of the order of 0.37%;
equation proposed by Guo et al. [18], which are valid only for bonded and for FEM, it is of the order of 0.93%. On the other hand, the error
prestressing steel, the prestress losses according to the parameters of made using the PE (Eq. (28)) is of the order of 12.9%. In this case, the
CEB-FIB [25] are Δσp = 65.4 MPa and Δσp = 51.9 MPa, respectively. same result is obtained by the Gauvreau equation and by the PE, since
We can conclude from this example that considering the bottom the relaxation of the prestressing steel is negligible.
reinforcement steel to be close to the unbonded prestressing steel in Eq. The values of the prestress losses obtained by applying ACI 209.2R
(25) has a significant influence on the results, particularly when the [40] for the parameters of the concrete do not differ significantly from
the previous values, and only small differences are observed.

Table 2
Comparison of the values of the prestress loss for the prestressed concrete beam of Lan et al. [39], calculated according to the parameters of CEB-FIB [25] and ACI
209.2R [40].
Δσp : prestress loss at t = 300 h (MPa)

Valid for unbonded prestressing Valid for bonded prestressing

Standard Measured SSM AAEM PE (Eq. (25)) Gauvreau [11] CEB-FIB [18] FEM EC 2 [42] Guo et al. [20]

CEB-FIB [25] 46 47.2 (2.61%) 46.7 (1.52%) 54.5 (18.5%) 63.9 (38.9%) 64.4 (40.0%) 46.4 (0.87%) 65.4 (42.1%) 51.9 (12.8%)
ACI 209.2R [40] 46 47.9 (4.13%) 47.6 (3.48%) 56.0 (21.7%) 66.4 (44.3%) 67.9 (47.6%) 47.3 (2.83%) 67.0 (45.7%) 53.3 (15.9%)

120
P.M. Páez, B. Sensale Engineering Structures 174 (2018) 111–125

Fig. 6. Geometry of the beam PC2 tested by Guo et al. [20].

Example 3: Concrete I-beam with curved tendon profile Table 4


Parameters of the concrete used to calculate the prestress loss in the concrete
In 1953, Breckenridge and Bugg [44] conducted testing of long-term beam of Guo et al. [20].
prestressed concrete I-beams at the U.S. Civil Engineering Naval La- Standard φ (t , t0) χ (t , t0) εsh (t , t0) [10−6] Ec (t0) [GPa]
boratory. The beams were 609.6 mm deep, the bottom flanges were
355.6 mm wide, the top flanges were 304.8 wide and the web thickness CEB-FIB [25] 1.139 0.851 −264.53 36.08
ACI 209.2R [40] 1.150 0.911 −273.24 35.34
was 101.6 mm (Fig. 7). The gross concrete area of the beams was
Ac = 11.645·10 4 mm2 and the moment of inertia was
Ic = 51.476·108 mm4. Their span length was 12,192 mm. Two of these
beams were prestressed using unbonded internal prestressing steel bars
and under the action of their own weight for a period of approximately
six and a half years. For the prestressing steel, two high-strength steel
bars of the Stressteel Bar System of 28.5 mm diameter
( Ap = 641.29 mm2) and a nominal tensile strength of 1007 MPa were
used. The effective prestress of each bar after anchoring was
683.23 MPa, and their profiles are given in Eqs. (76) and (77). Fig. 8
shows these profiles schematically. The ordinary steel reinforcement
consisted of eight bars of 9.50 mm diameter ( As = 70.79 mm2) and a
yield stress of 335 MPa. The modulus of elasticity of the prestressing
steel and the reinforcing steel were 168.9 GPa and 200 GPa, respec-
tively. Normal strength cement was used, and the characteristic com-
pressive cylinder strength of concrete was fck = 36.27 MPa
(Ec (t0) = 26.44 GPa). Fig. 7. Cross-section of the beams tested by Breckenridge and Bugg [44].

Table 3
Comparison of the values of the prestress loss in the PC2 beam tested by Guo et al. [20], calculated according to the parameters of CEB-FIB [25] and ACI 209.2R [40].
Δσp : prestress loss at t = 395 days (MPa)

Valid for unbonded prestressing Valid for bonded prestressing

Standard Measured SSM AAEM PE (Eq. (28)) Gauvreau [11] CEB-FIB [18] FEM EC 2 [42] Guo et al. [20]

CEB-FIB [25] 53.9 54.4 (0.93%) 53.7 (0.37%) 60.9 (12.9%) 60.9 (12.9%) 61.7 (14.5%) 53.4 (0.93%) 59.9 (11.1%) 52.8 (2.04%)
ACI 209.2R [40] 53.9 55.2 (2.41%) 54.4 (0.93%) 63.1 (17.1%) 63.1 (17.1%) 64.2 (19.1%) 54.1 (0.37%) 61.9 (14.8%) 54.0 (0.19%)

121
P.M. Páez, B. Sensale Engineering Structures 174 (2018) 111–125

steel, the prestress loss obtained is 164.8 MPa, which represents a dif-
ference of the order of 0.06% with respect to the PE. In other words, the
PE estimates the loss of prestress sufficiently accurately when it is
compared with complex approaches such as the SSM method, thus
giving a certain margin of safety. The influence of reinforcing steel on
the prestress loss is studied in the next example.

Example 4: Influence of the non-prestressed reinforcement on the pres-


tress loss

In this example, one of the beams tested by Harajli and Kanj [5],
called PP3R2-3, is analyzed (Fig. 9). The authors did not carry out a
study of the prestress loss. The beam is simply supported and the span is
Fig. 8. Schematic tendon profile of the beams tested by Breckenridge and Bugg 2540 mm. It has unbonded straight tendons and was tested until failure
[44]. under symmetrical two third-point loads, Q . The beam has a rectangular
section with a width of 127 mm and a depth of 279 mm. The area of the
dp (1) (x ) = −3.5334·10−5·x 2 + 20.4605·10−2·x + 192.79for 0 ⩽ prestressing steel is Ap = 116.13 mm2, the effective depth dp = 211 mm,
and the modulus of elasticity is Ep = 206.85 GPa. The area of the non-
x < 2895.6 mm
prestressed reinforcement steel is As = 226 mm2, the effective depth
dp (1) (x ) = 488.95 for 2895.6 ⩽ x < 9296.4 mm
ds = 254 mm, and the modulus of elasticity is Es = 200 GPa. The yield
dp (1) (x ) = −3.5334·10−5·x 2 + 65.7029·10−2·x −2564.89 for 9296.4 ⩽ x ⩽ stresses of non-prestressed and prestressed steel are 564 and 1290 MPa,
12, 192 mm respectively. The effective prestress is fse = 882.56 MPa and the char-
acteristic compressive cylinder strength of concrete is fck = 42.54 MPa
(76) (Ec (t0) = 35.77 GPa).
The values of the remaining parameters used to calculate the creep
dp (2) (x ) = −1.6119·10−5·x 2 + 9.3341·10−2·x + 417.32 for 0 ⩽
coefficient and shrinkage were an RH of 70%, a mean temperature of
x < 2895.6 mm 20 °C and a strength class of cement of 42.5 N. Then, the creep coeffi-
dp (2) (x ) = 552.45 for 2895.6 ⩽ x < 9296.4 mm cient is φ (t , t0) = 1.615, the shrinkage is εsh (t , t0) = −205.76·10−6 and the
dp (2) (x ) = −1.6119·10−5·x 2 + 29.9698·10−2·x −840.49 for 9296.4 ⩽ x ⩽ aging coefficient is χ (t , t0) = 0.836.
Although the authors do not specify it, a low-relaxation prestressing
12, 192 mm
steel is considered for the example, Δσpr (t −t0) = 25.2 MPa.
(77) For a load Q = 23 kN (total load = 46 kN) applied at t0 = 28 days
According to the authors, the beams were cured for eight days and then and kept constant, the prestress loss obtained at t = 10, 000 days by
prestressed. During the test period, the average temperature was applying Eq. (25) is Δσp = 60.9 MPa, and therefore the prestressing force
15.6 °C and the average relative humidity was 72.8%. The prestress loss is Np (t ) = 97, 734 N. By applying the SSM, the prestressing force ob-
measured at 6.5 years was 22%, that is, Δσp = 150.3 MPa. tained at t = 10, 000 days is Np (t ) = 97, 353 N, representing a difference
In order to make a comparison between the different calculation of the order of 0.4%.
methods, the parameters of the concrete were calculated according to From the beam PP3R2-3 of Harajli and Kanj [5] and from the
the CEB-FIB standard [25], with corrections due to the use of a tem- considerations discussed above, the variation of the prestressing force
perature of 15.6 °C rather than 20 °C. Under these conditions, the creep was studied as a function of the amount of longitudinal non-prestressed
coefficient was φ (t , t0) = 1.915, the shrinkage was reinforcement (Fig. 10). For this purpose, the values of the prestressing
εsh (t , t0) = −385.84·10−6 and the aging coefficient was χ (t , t0) = 0.678. force at t = 10, 000 days from the PE, the equation proposed by Gauv-
The relaxation of the prestressing steel was Δσpr (t −t0) = 28.5 MPa. The reau [11], the CEB-FIB equation [18] and the exposed methods of the
values of the relaxation coefficient calculated using the AAEM, PE, FEM SSM and the AAEM were calculated. In the first instance, it was con-
and the Guo et al. equation were 0.605, 0.532, 0.610 and 0.6461, re- sidered that the area of the reinforcement steel As (1) is zero and the area
spectively. According to EC 2 [42], χt (t , t0) = 0.8. of the reinforcement steel As (2) varies from zero to 1256 mm2, which
The prestress losses obtained at t = 2, 372.5 days for the different represents a non-prestressed reinforcement ratio of ρs = 35.28·10−3 . The
approaches described in this work are summarized in Table 5. results are shown in the graph in Fig. 11.
In order to draw accurate conclusions, we must make the following For different amounts of bottom reinforcement steel, the difference
observations. In this particular case, the Gauvreau equation and the obtained between the proposed equation (PE), Eq. (25), and the SSM is
CEB-FIB equation correctly estimate the prestress loss; however, they of the order of 0.46%. For members with a small amount of bottom
do not take into account the loss of prestress due to the relaxation of reinforcement steel, such as slabs, the Gauvreau equation presents a
prestressing steel or the presence of non-prestressed reinforcement. difference of the order of 2.83–2.04% with the SSM, while for the
Given the curved profiles of the prestressed bars, the PE corresponds to proposed equation, the said difference is of the order of 0.56–0.39%.
Eq. (28), which does not take into account the reinforcing steel. If the The difference obtained between SSM and AAEM is of the order of
SSM method is applied without taking into account the reinforcing 0.01%. The loss of prestress obtained by applying the equation

Table 5
Comparison of the values of the prestress losses for the prestressed concrete I-beam of Breckenridge and Bugg [44].
Δσp : prestress loss at t = 2, 372.5 days (MPa)

Valid for unbonded prestressing Valid for bonded prestressing

Measured SSM AAEM PE (Eq. (25)) Gauvreau [11] CEB-FIB [18] FEM EC 2 [42] Guo et al. [20]

150.3 154.9 (3.06%) 150.6 (0.20%) 163.8 (8.98%) 146.0 (2.86%) 149.7 (0.40%) 149.3 (0.67%) 125.4 (16.57%) 129.7 (13.71%)

122
P.M. Páez, B. Sensale Engineering Structures 174 (2018) 111–125

Fig. 9. Geometry and cross-section of the beam PP3R2-3 tested by Harajli and Kanj [5].

Fig. 10. Study of the influence of the amount of the longitudinal reinforcing
steel on prestress loss.
Fig. 12. Prestressing force at t = 10, 000 days as a function of the area of the top
reinforcement steel As (1) , with As (2) = 200 mm2.

at t = 10, 000 days is Np (t ) = 99, 280 N, while for the Guo et al. [18]
equation it varies from Np (t ) = 99, 390 N to Np (t ) = 99, 711 N, de-
pending on the amount of reinforcement steel.
The graph in Fig. 12 shows the variation of the prestressing force at
t = 10, 000 days as a function of the area of the top reinforcement steel
As (1) and with As (2) = 200 mm2. It can be seen that as the area of the top
reinforcement steel increases, the prestress loss also increases; however,
the prestress loss obtained for As (2) = 200 mm2 is of the order of 1.7%
greater than that obtained with As (2) equal to zero.
The graph in Fig. 13 shows the variation of the midspan deflection
at 10,000 days as a function of the area of top reinforcement steel As (1)
and for an area of the bottom reinforcement steel of As (2) = 200 mm2.
As the area of the top reinforcement steel increases, the midspan de-
Fig. 11. Prestressing force at t = 10, 000 days as a function of the area of the
flection decreases. The midspan deflection obtained at t = 10, 000 days
non-prestressed reinforcement As (2) , with As (1) = 0 .
for As (1) = 0 is of the order of 36% greater than in the case with
As (1) = 200 mm2.
proposed by the CEB-FIB [18] is Δσp = 49.9 MPa, which represents a It is important to note that while the increase of the area of the top
difference of the 1.63% with respect to the Gauvreau equation. reinforcement steel between zero and 200 mm2 has a significant influ-
It can be seen that the Gauvreau [11] and CEB-FIB [18] equations ence on the beam deformation (of the order of 36%), the influence on
represent a lower limit of the prestress loss as neither the relaxation of the prestress loss is very insignificant (of the order of 1.7%).
the prestressing steel nor the influence of the non-prestressed reinfor-
cing steel are considered.
If we apply the formula proposed in EC 2 [42], the prestressing force

123
P.M. Páez, B. Sensale Engineering Structures 174 (2018) 111–125

simplified equation makes it possible to estimate in a simple manner


and with sufficient precision the prestress loss in unbonded pre-
stressing steel. However, for those cases that require greater preci-
sion, the formulation of the Step by Step Method or the Age-
Adjusted Effective Modulus Method can be applied.
2. The amount of top and bottom longitudinal non-prestressed re-
inforcement steel influences the prestress loss. As the bottom re-
inforcement steel ratio increases, the prestress loss decreases.
Therefore, even if the member does not require it, it is convenient to
incorporate bottom reinforcement steel. On the other hand, as the
top reinforcement steel ratio increases, the prestress loss also in-
creases; however, while the increase in the prestress loss has very
little significance, there is a significant reduction in long-term de-
flections.
3. If formulations based on bonding assumptions are used to calculate
Fig. 13. Midspan deflection at t = 10, 000 days as a function of the area of the the prestress loss in unbonded prestressed concrete beams, incon-
top reinforcement steel As (1) , with As (2) = 200 mm2. sistent values are obtained; in other words, these values may in
some cases be close to the measured values and in others can differ
9. Conclusions significantly. This is typically because these formulations do not
take into account the deformation of the structural member as a
In this paper, we propose an approach based on the finite element whole.
method for the analysis over time of reinforced concrete beams with 4. From the point of view of computational cost, the implementation of
unbonded internal tendons, and an improved simplified equation for the SSM has a higher cost than methods based on the age-adjusted
calculating the prestress losses in statically determinate prestressed modulus method. This is due to the fact that at each time interval of
concrete beams with the same type of prestressing. Both methods take the SSM, an iterative procedure must be carried out to obtain the
into account losses due to creep, shrinkage, the relaxation of the pre- prestressing force, which in turn includes an adjustment of the re-
stressed steel and the effect of the longitudinal non-prestressed re- laxation coefficient of the steel. In addition, obtaining the stresses in
inforcement steel. In addition, a generalization of two widely used the concrete at a given moment in time requires knowledge of the
methods for the time-based cross-sectional analysis of prestressed stresses in the concrete at all prior instants in time. However, off-
concrete members with unbonded internal tendons is presented. These setting this higher computational cost is the fact that the SSM allows
methods are the step-by-step method and the age-adjusted effective a more direct and accurate analysis when the structure is subject to
modulus method. The proposed methods were evaluated based on the loads applied at different ages, as is usual in the construction of
results of previous studies, and the main conclusions can be summar- bridges, for example.
ized as follows: 5. In general, the most important standards worldwide propose sim-
plified equations for the calculation of the prestress losses in iso-
1. For prestressed concrete members with straight tendons and con- static prestressed concrete beams with bonded internal prestressing.
stant longitudinal non-prestressed steel reinforcement, only small As has been demonstrated, the use of said equations for the case of
differences are observed among the prestress loss calculated by the unbonded internal prestressing provides values with significant er-
proposed formula, that calculated by applying the Step-by-Step rors. Therefore, it would be advisable to incorporate the proposed
Method, and the values measured in previous studies, whereas for simplified equation for the calculation of the prestress loss in iso-
members with generic tendon profiles, the values obtained with the static prestressed concrete beams with unbonded internal prestres-
proposed formula show somewhat higher differences. Therefore, for sing.
the practical purposes of structural engineering, the proposed

Appendix A. Brief description of the equations referenced in the text

A.1. CEB-FIB equation

The equation proposed by the CEB-FIB to estimate the prestress loss in unbonded prestressed beam is given by Eq. (A.1). This equation is based on
the Navier-Bernouilli assumptions and on the rate of creep method [18].
d [ΔPp (t )] δc,1,1 δc,1,0 εsh (t )·l
+ ΔPp (t ) = +
dφ δc,1,1 +δp,1,1 δc,1,1 +δp,1,1 φ (t )·(δc,1,1 +δp,1,1) (A.1)
where φ (t ) and εsh (t ) are the creep coefficient and the shrinkage at time t , respectively; δc,1,1 is the deformation of the concrete at the level of the
centroid of the unbonded prestressed steel due to a unit prestressing force; δc,1,0 is the deformation of the concrete at the level of the centroid of the
unbonded prestressed steel due to the long-term external loads and prestressing, and δp,1,1 = l/(Ep Ap ) .

A.2. Gauvreau equation

The Gauvreau equation is based on Navier-Bernoulli assumptions, linear behavior of the concrete and the assumption that all sections are
uncracked. To account for the long-term effect, the age-adjusted effective modulus method is used. Gauvreau equation is given by Eq. (A.2) [11],
1
nE ·ρp ⎡Ac φ (t ) l ∫ σc, p,0 (x ) dx + εsh (t ) Ec Ac ⎤
ΔPp (t ) = ⎣ p ⎦
Ac
1 + nE ·ρp (1 + χφ (t )) ⎡1 + ∫ ecp2 (x ) dx⎤
⎣ lp Ic
⎦ (A.2)

124
P.M. Páez, B. Sensale Engineering Structures 174 (2018) 111–125

where nE = Ep/ Ec and ρp = Ap / Ac .

A.3. Equation proposed by Guo et al.

The equation proposed by Guo et al. [20] to calculate the loss of prestress is given in Eq. (A.3). The assumptions used to develop this equation are
the plane cross-section assumption; compatibility conditions between concrete and prestressed tendons and between concrete and non-prestressed
steels; linear behavior of the concrete; and that the concrete is uncracked.

Ap̂ σc, p (τ ) φ (t , τ ) + Bp̂ Ep εsh (t , τ ) + Cp̂ Δσpr


σlp =
Dp̂ (A.3)
where σlp is the prestress loss, τ and t are the initial and the final times respectively, and the coefficients Ap̂ , Bp̂ , Cp̂ and Dp̂ are given by Eq. (A.4),

Ap̂ = nsp [1 + nsp ρs rs, s (1 + χ (t , τ ) φ (t , τ ))−nsp ρs rp, s (1 + χ (t , τ ) φ (t , τ ))]


Bp̂ = 1 + nsp ρs (1 + χ (t , τ ) φ (t , τ ))(rs, s−rp, s )
Cp̂ = 1 + nsp ρs rs, s (1 + χ (t , τ ) φ (t , τ ))
Dp̂ = 1 + nsp (1 + χ (t , τ ) φ (t , τ ))(ρs rs, s + ρp rp, p) (A.4)
where nsp = (Es + Ep)/(2·Ec (τ )) ; ρs = As / Ac , ρp = Ap / Ac ; ri, j = 1 + ei ej Ac / Ic ; i and j represent the non-prestressed steels (s) or the prestressing tendons
(p); and es and ep are their eccentricities with respect to the centroid of the concrete section, respectively.

References concrete structures – model B3. Mater Struct 1995;28:357–65.


[24] Youakim SA, Ghali A, Hida SE, Karbhari VM. Prediction of long-prestress loss. PCI J
2007:116–30.
[1] Naaman AE, Alkhairi FM. Stress at ultimate in unbonded post-tensioning tendons: [25] International Federation for Structural Concrete (FIB). Model Code 2010, vol. 1.
part 2 – proposed methodology. ACI Struct J 1991;88(6):641–51. CEB-FIP Bulletin no. 65. Federation Internationale du Béton, Lausanne,
[2] Naaman AE, Alkhairi FM. Stress at ultimate in unbonded post-tensioning tendons: Switzerland; 2012.
part 1 – evaluation of the state-of-the-art. ACI Struct J 1991;88(5):641–51. [26] Bazant ZP. Mathematical models for creep and shrinkage of concrete. In: BaZant ZP,
[3] Mattock AH, Yamazaki J, Kattula BT. Comparative study of prestressed concrete Wittmann FH, editors. Creep and shrinkage in concrete structures. Wiley: London;
beams, with and without bond. PCI J 1971;116:25. 1982. p. 163–256.
[4] Do G, Tao X. Ultimate stress of unbonded tendons in partially prestressed concrete [27] Neville AM. Creep of concrete: plain reinforced and prestressed. North-Holland
beams. PCI J 1985:72–91. Publishing Company; 1970.
[5] Harajli MH, Kanj MY. Ultimate flexural strength of concrete members prestressed [28] McHenry DA. A new aspect of creep in concrete and its application to design. ASTM
with unbonded tendons. ACI Struct J 1991;88(6):663–73. Proc 1943:43.
[6] Gauvreau DP. Load test of concrete girders prestressed with unbonded tendons. [29] Ross AD. Creep of concrete under variable stress. J Am Concr Inst 1958:54.
Birkhauser Verlag Basel; 1992. [30] Dilger W, Neville AM. Method of creep analysis of structural members. ACI SP27-17
[7] Alkhairi FM, Naaman AE. Analysis of beams prestressed with unbonded internal or 1971;17:349–72.
external tendons. J Struct Eng 1993;119(9):2680–700. [31] Bazant ZP. Prediction of concrete creep effects using age-adjusted effective modulus
[8] Pisani MA, Nicoli E. Beams prestressed with unbonded tendons at ultimate. Can J method. ACI J 1972;69:212–7.
Civ Eng 1993;23:1220–30. [32] Neville AM, Dilger WH, Brooks JJ. Creep of plain and structural concrete. London:
[9] He ZQ, Liu Z. Stresses in external and internal unbonded tendons: unified metho- Construction Press (Longman Group Ltd); 1983.
dology and design equations. ASCE, J Struct Eng 2010;136(9):1055–65. [33] Bazant ZP, Hubler MH, Jirásek M. Improved estimation of long-term relaxation
[10] Harajli MH. Proposed modification of AASHTO-LRFD for computing stress in un- function from compliance function of aging concrete. J Eng Mech
bonded tendons at ultimate. ASCE J Bridge Eng 2011;16(6):828–38. 2013;139:146–52.
[11] Gauvreau DP. Ultimate limit state of concrete girders prestressed with unbonded [34] Galhi A, Favre R, Eldbadry M. Concrete structures: stresses and deformation. 3rd ed.
tendons. Birkhauser Verlag Basel; 1993. New York: Spon Press; 2002.
[12] Ariyawardena N, Ghali A. Prestressing with unbonded internal or external tendons: [35] Menn C. Prestressed concrete bridges. Berlín: Birkhauser Verlag AG Basel; 1990.
analysis and computer model. ASCE J Struct Eng 2002;128(12):1493–501. [36] Gilbert RI, Ranzi G. Time-dependent behavior of concrete structures. New York:
[13] Harajli MH. Tendon stress at ultimate in continuous unbonded post-tensioned Spon Press; 2011.
members: proposed modification of ACI 318 Eq. (18-4) and (18-5). ACI Struct J [37] Oñante, E. Structural analysis with the finite element method. Linear statics.
2012;109(2):183–92. Volume 1: basis and solids. Springer; 2009.
[14] Lee DH, Kim KS. Flexural strength of prestressed concrete members with unbonded [38] Oñante, E. Structural analysis with the finite element method. Linear statics.
tendons. Struct Eng Mech 2011;38(5):1–20. Volume 2: beams, plates and shells. Springer; 2009.
[15] Lou T, Lopes S, Lopes AV. Nonlinear and time-dependent analysis of continuous [39] Lan C, Zhou Z, Ou J. Monitoring of structural prestress loss in RC beams by inner
unbonded prestressed concrete beams. Comput Struct 2013;119:166–76. distributed Brillouin and fiber Bragg grating sensors on a single optical fiber. Struct
[16] Du JS, Au F, Chan E, Liu L. Deflection of unbonded partially prestressed concrete Control Health Monit. Wiley Online Library 2013. https://doi.org/10.1002/stc.
continuous beams. Eng Struct 2016;118:89–96. 1563.
[17] Vu NA, Castel A, Francois R. Response of post-tensioned concrete beams with un- [40] American Concrete Institute (ACI). ACI 209.2R-08. Guide for modeling and calcu-
bonded tendons including serviceability and ultimate state. Eng Struct lating shrinkage and creep in hardened concrete. Farmington Hill, USA: ACI com-
2010;32(2):556–69. mittee 209; 2008.
[18] International Federation for Structural Concrete (FIB). Structural concrete: textbook [41] American Concrete Institute (ACI). ACI 211.1-91: Standard practice for selecting
on behaviour, design and performance, vol. 2. CEB-FIP Bulletin no. 52. Federation proportions for normal, heavyweight, and mass concrete reapproved Farmington
Internationale du Béton, Lausanne, Switzerland; 2010. Hill, USA: ACI committee 211; 2009.
[19] Harajli MH, Kanj MY. Service load behavior of concrete members prestressed with [42] UNE-EN 1992-1-1:2013/A1:2015. Eurocode 2: Design of concrete structures – Part
unbonded tendons. ASCE, J Struct Eng 1992;118(9):2569–89. 1-1: General rules and rules for buildings. CEN; 2016.
[20] Guo T, Chen Z, Lu S, Yao R. Monitoring and analysis of long-term prestress losses in [43] American Concrete Institute (ACI). ACI 211.4R-08: Guide for selecting proportions
post-tensioned concrete beams. Measurement 2017. https://doi.org/10.1016/j. for high-strength concrete using Portland cement and other cementitious materials.
measurement.2017.07.057. Farmington Hill, USA: ACI committee 211; 2008.
[21] Glanville WH. Studies in reinforced concrete – III, The creep or flow of concrete [44] Breckenridge RA, Bugg SL. Effect of long-time loads on prestressed concrete beams.
under load. Building research technical paper no. 12 1930. PCI J 1964;75–89.
[22] Whitney CS. Plain and reinforced concrete arches. ACI J 1932;28:479–519.
[23] Bazant Z, Baweja S. Creep and shrinkage prediction model for analysis and design of

125

You might also like