Documente Academic
Documente Profesional
Documente Cultură
ScienceDirect
Article history: Reinforced concrete piles employed in earth retaining systems are typically designed with
Received 1 June 2016 symmetric reinforcement. The non-symmetric RC wall piles have recently been introduced
Accepted 2 December 2016 by the authors, obtaining savings of up to 50% in weight in longitudinal reinforcing steel
Available online compared with the traditional solutions, leading to significant financial savings while also
reducing associated environmental impacts.
Keywords: The structural behavior of this new RC member under long-term loading is studied,
Long-term loading comparing it with its symmetrical counterpart. An experimental campaign has been carried
Tension stiffening out. Full scale specimens with circular cross sections symmetrically and asymmetrically
Effective area of concrete reinforced were tested. Results have shown that asymmetrically RC pile developed a slightly
Asymmetric reinforcement higher deflection than its symmetrical counterpart. A new expression for the effective area
Serviceability of concrete in tension applicable to non-symmetrical piles is introduced. Moreover, a new
stress–strain law for cracked concrete that accounts the tension stiffening effect for long-
term loading is proposed. Finally, for non-symmetrical RC wall piles, the evolution of the
parameter that takes into account the duration of loading in deformations is presented.
Although more evidence is needed, it is shown that tension stiffening effect contribution
could be overestimated by Eurocode 2 in the case of non-symmetrically or underestimated in
case of symmetrically RC piles.
© 2016 Politechnika Wrocławska. Published by Elsevier Sp. z o.o. All rights reserved.
* Corresponding author.
E-mail addresses: emontes@ugr.es (E. Hernández-Montes), malejandrofr@ugr.es (M.A. Fernández-Ruiz),
jfcarbonell@uloyola.es (J.F. Carbonell-Márquez), mlgil@ugr.es (L.M. Gil-Martín).
http://dx.doi.org/10.1016/j.acme.2016.12.003
1644-9665/© 2016 Politechnika Wrocławska. Published by Elsevier Sp. z o.o. All rights reserved.
434 archives of civil and mechanical engineering 17 (2017) 433–445
- εy - εap - εctm
4. Computation of deformation of NSRCS and
ε= Δl/l
SRCS through the integration of the stress–strain
models of the materials (procedure 3 of Fig. 3) Ac,eff σcTS (εap)
Uncracked
Cracked
The third procedure consists in solving the equilibrium σcTS
equations to obtain the M–f diagram. M–f diagrams are used As
to obtain the deformation of elements by integrating the
flexural moment diagrams along the length of the element. N N
Approximations of M–f diagrams used to be checked experi-
fctm
mentally in four point bending tests.
Δl/2 Δl/2
Models of the behavior of steel and concrete (stress–strain l σc
models) in tension and in compression are used. The average
behavior of concrete under tension stresses in a cracked region Fig. 5 – Tension stiffening model for cracked concrete in
is represented by a stress–strain model called tension short-term loading.
stiffening formulation (Section 4.1). This formulation must
be applied to a certain region of the cross section called
effective area of concrete in tension, Ac,eff (Section 4.2). This at a crack, see Fig. 5, the average strain of the concrete prism is
area of concrete, affected by tension stiffening, surrounds the eap (apparent yield strain). The value of eap can be obtained from
tension reinforcement because this phenomenon is a conse- the equilibrium between a cracked section where the
quence of the bond interaction between concrete and reinforcement has yielded and a section between cracks
reinforcement. Finally, the M–f diagrams are obtained where the strain is equal to eap:
imposing equilibrium (Section 4.3).
f y As ¼ s s ðeap ÞAs þ s cTS ðeap ÞAc;eff (6)
4.1. Tension stiffening model where As is the area of tensioned steel and fy is the yield
strength of the steel. If the strain keeps growing, the average
Design codes as Eurocode 2 [13] and CEB-fib Model Code 2010 tensile stress in concrete decreases linearly until the average
(MC2010) [17] do not provide an explicit expression for the strain of the prism reaches ey, see Fig. 5. Afterwards, it is
tension stiffening effect of RC cross section. These codes only assumed that the tensile contribution of concrete is negligible
provide an expression to evaluate the deformation, in which [24]. The complete stress–strain expression for cracked con-
the tension stiffening effect is taken into account implicitly, crete (scr) in short-term loading (b = 1) is given by Eq. (7). This
through the interpolation between two situations that corre- expression is represented in Fig. 5.
spond to the uncracked and fully cracked states of the section
(second procedure of Fig. 3). 8
>
> 0 if e < ey
<s
cTS ðeap Þ
4.1.1. General tension stiffening model s cr ðeÞ ¼ ðe þ ey Þ if ey e < eap (7)
>
> e e
Although many different models to capture the tension : y ap
s cTS ðeÞ if eap e < ectm
stiffening effect in RC sections have been proposed in the
literature [18–22], Hernández-Montes et al. [23] proposed a In the above expression scTS(e) is the tension stiffening
model for the tension stiffening effect of concrete extracted contribution of cracked concrete, which was defined in Eq. (5).
from the implicit formulation given in EC2 by Eq. (1). In [23] an
average tensile behavior of concrete after cracking is formu- 4.1.3. Tension stiffening model in long-term loading
lated as function of the average strain e, i.e. scTS(e). Using the The long-term loading effect is included in the M–f diagram of
sign convention that tensile stresses and strains are negative the RC cross sections through the parameter b in Eq. (5). EC2
in tension, the expression proposed by [23] provides the [13] proposes a value of b = 0.5 for sustained loads or many
tension stiffening contribution of cracked concrete: cycles of repeated loading regardless of the period of time the
r
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2 load acts (months, years, etc.). Despite the complexity of the
r r
s cTS ðeÞ ¼ Es jej Es jej þ bfctm 2 ð1 þ nrÞ (5) factors that influence the value of b [16] (duration of loading,
2 2
shrinkage strain and deterioration of the bond between steel
being r the reinforcement ratio, n the coefficient of equiva- and concrete), the parameter which introduces the long-term
lence (defined as n = Es/Ec, being Es and Ec the elastic modulus effect in the response of RC members according to EC2 [13] only
of steel and concrete, respectively) and fctm the mean axial takes two possible values (1 or 0.5). Mathematically, the
tensile strength of concrete. introduction of b < 1 for the computation of long-term
deflections is equivalent to a reduction of the cracking
4.1.2. Tension stiffening model in short-term loading moment over time, being an over-simplistic way of accounting
In the case of short-term loading and according to EC2 [13] time-dependent cracking and the reduction of tension
parameter b in Eq. (2) is equal to 1 (b = 1). This expression is stiffening over time [14,25].
valid if the steel, at any crack in the member, does not reach its If there is not significant shrinkage before first loading,
tensile yield strain ey (being: ey = fy/Es). For a monotonic it is possible to compute the deflection of a RC member
increasing loading, at the instant at which the steel yields first immediately after loading using b = 1 [16] but under normal
438 archives of civil and mechanical engineering 17 (2017) 433–445
Fig. 6 – Tension force applied to the prism (a) and concrete stress–strain law (b).
circumstances, in which significant shrinkage occurs before The effect of creep is introduced in the stress–strain model
first loading, Gilbert [14] proposes a value of b = 0.7 for early ages of cracked concrete by replacing Ecm by the effective elastic
(less than 28 days) and b = 0.5 for ages greater than 6 months. modulus, Ec,eff [26], defined in Section 5. The stress–strain curve
The stress–strain model for cracked concrete given by Eq. (7) of cracked concrete in tension affected by creep is represented
(and represented in Fig. 5) is valid for short-term loading, for in Fig. 7a. The following step is the introduction of shrinkage
which b = 1. In order to consider long-term loading, parameter b effect on the model. To do this the creep-affected model in
in Eq. (7) must be converted into a value lower than 1. Fig. 7a is shifted in horizontally until zero stress corresponds
In Fig. 6a, the load–strain curve of a concrete prism with an with the shrinkage strain, esh, obtained according to EC2 [13]
embedded bar has been represented. The line o-d (dark dashed (see Fig. 7b).
line) in Fig. 6a corresponds to a bare bar whereas the other two Given all the above, the mathematical expression of the
lines (o-a-e-c-d and o-a-b-c-d) correspond to the tension stress–strain relationship for cracked concrete in long-term
stiffening model b = 1 and b = 0.5, respectively. loading is:
8
>
> 0 if e < ðey esh Þ
>
> s cTS ðeap;eff þ esh Þ s cTS ðeap;eff þ esh Þ
>
>
>
< e þ ðey esh Þ if ðey esh Þeðeap;eff esh Þ
e e
y ap;eff e e
y ap;eff
s crTS ðeÞ ¼ (8)
>
> s cTS ðe þ esh Þ if ðeap;eff esh Þ < e < ðectm;eff esh Þ
>
>
>
> f ctm
>
: ðeesh Þ if ðectm;eff esh Þeesh
ectm;eff
The portion a-b (for long-term effect, i.e. when b = 0.5) in with scTS(e) given by Eq. (5).
Fig. 6a corresponds to a sudden drop of the tension force
applied to a RC prism just at the moment in which the cracking 4.2. Effective area of concrete in tension for NSRCS and
axial load (Ncr) is reached, i.e. at the strain corresponding to SRCS
cracking, ectm.
The former explanation, related to the load–strain curve, is Reinforced concrete design codes such as EC2 [13] provide
equivalent to introduce a drop (branch a-b) in the stress–strain expressions for Ac,eff applicable only to rectangular cross
curve for cracked concrete under tension (see Fig. 6b). For a sections. Few studies have been published about effective area
given value of b, the drop a-b can be calculated using Eq. (5). of concrete in tension for circular RC cross sections.
σc σc
εy εap,eff εctm,eff εy εsh
ε ε
Affected by φ
fctm fctm
(a) (b)
Fig. 7 – Cracked concrete stress–strain relationship affected by creep (a) and shrinkage (b) in long-term loading.
archives of civil and mechanical engineering 17 (2017) 433–445 439
Recently Carbonell-Márquez et al. [27] proposed a formula- Rint is the radius of the circle defined by the centroids of the
tion to obtain the geometry of the tension stiffening area valid longitudinal bars (see Fig. 9). It is worth noting that Wiese et al.
for both SRCS and NSRCS. The value of Ac,eff proposed [27] was [28] envisioned the reinforcement to be a continuous ring
supported by short-term tests [11]. This effective area (see located at Rint. In this work, a discrete distribution of the
Fig. 8) was determined in such a way that, using the stress– reinforcing bars is considered (Fig. 9).
strain relationships for steel and uncracked concrete proposed In addition, a simpler effective area of concrete in tension
by EC2 [13] and the tension stiffening model proposed by Ac,eff for NSRCS has been considered in this work. This value of
Hernández-Montes et al. [23], the M–f relationship of the RC Ac,eff is similar to the one proposed by EC2 [13] for beams, it is
cross section obtained using procedure 2 indicated in Fig. 3 is interesting to observe that due to the steel distribution NSRCS
the same as the one obtained using procedure 3. Note that are very similar to rectangular beams: steel in the bottom part
doing so, experimental data is not needed. of the cross section. In the case of circular cross section of
The effective area proposed by Carbonell-Márquez et al. [27] NSRCS the proposed effective area corresponds to the circular
is placed on the tensile side of the cross section (below the segment represented in Fig. 10, whose height is given by:
horizontal fiber whose strain is ectm = fctm/Ec). It is a circular strip 8
< 2:5h1
whose width is hc,eff, see Fig. 8. The width hc,eff is distributed in hc;eff ¼ Min R (12)
:
two zones (interior, hc,effint, and exterior, hc,effext) around the ð2Rxcr Þ=3
circle that links the center of gravity of the bars. The
expressions proposed in [27] to calculate the geometry of Ac, In the above equation, h1 is the distance that defines the
eff are the following: position of the centroid of the tensioned reinforcement, and xcr
is the neutral fiber depth corresponding to the cracking
TSztop
hc;effint ¼ hc;eff Rint moment, Mcr [29].
TSz (9)
hc:effext ¼ hc;eff hc;effint In the present work several effective areas of concrete in
tension have been considered and compared: the one
For traditional or symmetric reinforcement: proposed by Wiese et al. [28] for SRCS, the one represented
in Fig. 10 for NSRCS and the effective area proposed by
1
hc;eff ¼ ð1:765R2 þ 11:343Rx9:375x2 Þ (10) Carbonell-Márquez et al. [27] for both, NSRCS and SRCS.
R4=3
4.3. Computation of the M–f diagrams The reinforcing steel used in the experimental campaign was
B-500-S for both specimens. In order to check if the bilinear
The M–f diagram of a particular cross section, at a certain time model of the steel is suitable, the actual stress–strain behavior
t, can be obtained from the equilibrium equations. Assuming of Ø16 reinforcing bars was obtained from tensile tests (see
that plane sections remain plane after deformation (Bernoul- Fig. 11). Due to the fact that during the experimental campaign
li's hypothesis) and that no slip of reinforcement occurs, the the strains in the steel, es, were lower than 0.01 mm/mm, the
strain on any fiber of the cross section is given by: bilinear model proposed by Eurocode 2 [13] was adopted in the
numerical analysis. The yield stress adopted in the numerical
analysis was also taken from tensile tests (Fig. 11). The effect of
eðy; ecg ; fÞ ¼ ecg þ fy (13)
the time on the mechanical characteristics of the steel was
where y is the vertical coordinate of the fiber with respect to neglected.
the centroid of the gross section and ecg is the strain at the
center of gravity of the gross section, see Fig. 8. Eq. (13) can also 5.2. Stress–strain model for uncracked concrete
be expressed in terms of the angle u (the angle between the
vertical principal axis of inertia of the section and the radius The adopted stress–strain model for uncracked concrete is the
corresponding to the fiber where the strain is evaluated – Fig. 8) one proposed by EC2 [13]. The long-term effect is introduced by
for x ≤ 2R as: considering shrinkage and creep as well as the effect of aging
in concrete.
eðu; ecg ; fÞ ¼ ecg þ fRcosu (14)
Design of RC structures is usually based on the 28-days
The sign convention chosen for this study is the following: strength of concrete. If a long-term loading is considered, EC2
bending moments which cause compression on the top fiber [13] proposes the following expression for the development of
are assumed positive, as are compression strains, stresses and the mean compressive strength of concrete with time at 20 8C:
forces.
The angle, referred to the principal vertical axis of inertia,
f cm ðtÞ ¼ bcc ðtÞf cm (17)
which defines the cracked area of the cross section is called
aectm (see Fig. 8). This angle corresponds to the fiber whose
strain is equal to the cracking strain of concrete, ectm. According σs (MPa)
to the above definition, the equilibrium of the cross section can 700
be expressed as:
Z aectm Z 600
2
N¼2 ðRsinuÞ s c ðu; ecg ; fÞdu þ s cTS ðu; ecg ; fÞdA
0 Ac;eff 500
X
þ A;j s s ðyj ; ecg ; fÞ (15) 400
j
Z 300
aectm
M¼ R3 sin2 ucosus c ðu; ecg ; fÞdu 200
0
Z X
þ s cTS ðu; ecg ; fÞydA þ A;j s s ðyj ; ecg ; fÞyj (16) 100
Ac;eff j
εs
0.00 0.02 0.04 0.06 0.08 0.10
where AØj and yj are the cross sectional area and the vertical
coordinate of the reinforcing bar j, respectively and sc, scTS and Fig. 11 – Strain–stress curves of longitudinal reinforcement
ss are the stresses in uncracked concrete, cracked concrete and used in specimens.
steel, respectively. Adapted from [11].
archives of civil and mechanical engineering 17 (2017) 433–445 441
Two pile specimens were tested, one with asymmetric and the Once the reinforcement cages were assembled (Fig. 13a and
other one with symmetric longitudinal reinforcement. Both b) they were introduced into cylindrical cardboard formworks
specimens had a length of 4 m and a circular cross section with placed vertically for concreting in. Finally, when the concrete
a diameter of 400 mm. The two specimens were subjected to 4- acquired sufficient strength, formworks were removed. The
point bending tests (Fig. 12). A constant bending moment of P load was introduced after 28 days through a lever using a
kN m in the central stretch of the specimen was applied for the transversal beam to transfer the load (a concrete cube) to the
duration of 3 months (long-term loading). two loading points of the specimen (see Fig. 14). This
Geometrical characteristics of the tested piles are summa- mechanism allowed the application of a constant load over
rized in Fig. 13. The symmetric cross section (Fig. 13a) was 100 days.
reinforced with 16 bars of Ø 16 mm (As = 3216.99 mm2) equally Linear transducers LVDT with a range of 100 mm and an
spaced. The asymmetric specimen (Fig. 13b) was reinforced accuracy of 0.001 mm were used to record deflections on both.
with 7 bars of 16 mm diameter placed at the bottom of the An ordinary Portland cement was used in the concrete
section and 3 bars of 12 mm diameter (As = 1643.06 mm2). Both mixes, with a water/cement ratio of 0.45. The concrete mean
specimens have a very similar ultimate flexural moment (see compressive strength ( fcm) of each specimen was determined
Table 1) but a saving of almost 50% of steel is achieved using from standard compression tests. These values are summa-
the non-symmetrical reinforcement. rized in the last row of Table 1.
R = 200 mm
R = 200 mm c = 77 mm s1
c = 77 mm Øs = 10 mm
R Ø R
Øs = 10 mm Ø1 = 12 mm Ø1
Ø = 16 mm nØ1 = 3
nØ = 16 s
s1 = 193.21 mm
s = 48.30 mm Ø2= 16 mm Ø2
nØ2 = 7 s2
s2 = 41 mm
c+Øs+Ø/2
c+Øs+Ø1/2
(a) (b)
M (N·mm)
6·107
5·107
4·107
3·107
EC2 [13] Eq.(3) -Procedure 2-
2·107
Ac,eff Figure 10 -Procedure 3-
7
1·10 Ac,eff Carbonell-Márquez et al. [27]
-Procedure 3-
3·107
EC2 [13] Eq.(3) -Procedure 2-
7. Deformation in time
2·107
Ac,eff Figure 10 -Procedure 3-
Theoretical deflection was computed by integration of the 1·10-6 2·10-6 3·10-6 4·10-6 5·10-6 6·10-6 7·10-6
curvature along the axis of the specimen using finite ϕ (rad/mm)
differences. The curvature at each section can be obtained
Fig. 16 – Moment–curvature diagrams of the asymmetrical
directly from the bending moment diagram because it is a
specimen corresponding to t = 120 days and b = 0.5.
determined structure.
The moment–curvature diagrams were calculated by
δ (mm)
means of two methods: (1) using the formulation proposed Experimental results
and Carbonell-Márquez et al. [27] for SRCS; and Carbonell- Ac,eff Carbonell-Márquez et al. [27], β = 0.7
4 Ac,eff Figure 10, β = 0.5
Márquez et al. [27] and the effective area of concrete in tension
Ac,eff Figure 10, β = 0.7
Ac,eff represented in Fig. 10 for NSRCS). The latter corresponds
to the procedure 3 of Fig. 3.
2
Figs. 15 and 16 show the M–f diagrams of the asymmetrical
specimen for different times (t = 28 and 120 days) and different
t (days)
values of b (1 and 0.5) respectively. The diagram corresponding
60 80 100 120 140
to b = 0.5 (Fig. 16) shows a horizontal part corresponding with
the pair Mcr–fcr. On the other hand, b = 1 provokes a M–f Fig. 17 – Estimated (solid and dashed lines) and measured
diagram without this horizontal portion (Fig. 15). (points) deflection at midspan of the asymmetric specimen,
It is evident from Figs. 15 and 16 that the M–f diagram NSRCS.
computed with the expression of Ac,eff proposed in Carbonell-
Márquez et al. [27] coincides with the one proposed by EC2 [13]
(obtained following procedure 2). For its part, the newly
introduced effective area of concrete in tension Ac,eff for NSRCS values obtained from integration of the curvature along the
represented in Fig. 10 leads to a M–f diagram that has a length of the member by finite differences (path 3 of Fig. 3).
reasonable agreement with the one proposed by EC2 [13] Two values of parameter b have been considered: the value
(Figs. 15 and 16) despite its simpler definition. proposed by EC2 [13] for sustained loads, b = 0.5, and an
Figs. 17 and 18 show the evolution of the deflection at intermediate value equal to b = 0.7. The last value of b lies
midspan of the tested specimens over time. Points represent between the values proposed by EC2 [13] and MC2010 [17] for
experimental measured values whereas lines represent the early ages – i.e. less than 28 days (for which b = 1) and for ages
archives of civil and mechanical engineering 17 (2017) 433–445 443
1 1
Ac,eff Carbonell-
0,8 Márquez et al. [27] 0,8 Ac,eff Carbonell-Márquez et al. [27]
0,6 0,6
β
β
0,4 0,4
0 0
0 20 40 60 80 100 120 0 20 40 60 80 100 120
t (days) t (days)
(a) (b)
Fig. 20 – Evolution in time of parameter b for asymmetrically (a) and symmetrically (b) reinforced specimens.
8. Conclusions