Construction and Building Materials 188 (2018) 101–118

Contents lists available at ScienceDirect

Construction and Building Materials

journal homepage: www.elsevier.com/locate/conbuildmat

Experimental and numerical studies on bonded prestressed concrete

beams at low temperatures
Jian Xie, Xueqi Zhao, Jia-Bao Yan ⇑
School of Civil Engineering, Tianjin University, Tianjin 300350, China
Key Laboratory of Coast Civil Structure Safety of Ministry of Education, Tianjin University, Ministry of Education, Tianjin 300350, China

h i g h l i g h t s

Low temperature improves cracking and ultimate resistances of bonded PC beams.

Proposed theoretical models predict well resistances of bonded PC beams at low temperatures.

Proposed FEM predict well behaviours of bonded PC beams at low temperatures.

Strength and elastic modules of concrete decrease after freeze-thaw cycles.

Behaviours of PC beams were changed by low temperatures and freeze-thaw cycles.

a r t i c l e i n f o a b s t r a c t

Article history: This manuscript studied the structural performances of bonded prestressed concrete (PC) beams at dif-
Received 23 June 2018 ferent low temperatures (+20 °C to 100 °C). The main parameters of analysis and discussion were tem-
Received in revised form 17 August 2018 perature and prestress levels. Theoretical models were proposed to predict the cracking and ultimate
Accepted 17 August 2018
resistances of the bonded PC beams at low temperatures. Numerical model was also proposed to simulate
Available online 23 August 2018
behaviours of PC beams at low temperatures. The accuracies of these theoretical and numerical models
were validated against the test results. Compressive tests on concrete after freeze-thaw cycles were also
Keywords:
carried out to obtain the corresponding mechanical properties. With these test results as input informa-
Low temperatures
Prestressed concrete beams
tion, numerical parametric studies were performed to study the behaviours of bonded PC beams at low
Concrete strength temperatures or after freeze-thaw cycles.
Freeze-thaw cycles Ó 2018 Elsevier Ltd. All rights reserved.

1. Introduction the external concrete members may suffer low temperature of

Prestressed concrete (PC) members are widely used in civil
engineering constructions, e.g., long-span structures and bridges
[1]. Due to their improvements on the initial stiffness and cracking
resistance, PC beams are also applicable to constructions suffering
harsh environments. In order to meet the needs of economic devel-
opment and resource explorations or storage, engineering con-
structions with PC beams were built in cold regions or exposed
to low-temperature environments, e.g., infrastructures in cold
regions, liquefied natural gas (LNG) containers, the Arctic onshore
and offshore platforms. For cold regions in northern China or Tibet,
the recorded lowest temperature could drop to 60 °C [2] whilst
the lowest temperature in the Arctic was about 70 °C [3,4]. For
the structures exposed to low temperatures, e.g., LNG containers,
⇑ Corresponding author.
E-mail address: ceeyanj@163.com (J.-B. Yan).
about 165 °C in the scenario of leakage of LNG [5]. Thus, PC
beams used in these structures exposed to low temperatures and
freeze-thaw cycles. All these would result in impacts on their
structural behaviours that need to be carefully studied and consid-
ered in the designs.
Previous studies showed that the mechanical behaviours of
constructional materials changed with the decrease of tempera-
ture, e.g., concrete and steel. As the temperature decreased from
20 °C to 160 °C, the tensile strength, compressive strength and
elastic modulus of concrete, the yield and ultimate strengths of
steel all increased [6,7]. Therefore, these changes of material prop-
erties improved the structural behaviours of reinforced concrete
beams and prestressed concrete beams made of such materials.
Liu et al. [8] tested six reinforced concrete (RC) beams at different
temperature levels (+20 °C to 180 °C). It showed that with the
decrease of temperature, the cracking load, yield resistance, and
ultimate resistance all increased. Yan et al. [9] studied the struc-
tural performance of twelve RC beams at temperatures of 20 °C,

https://doi.org/10.1016/j.conbuildmat.2018.08.117
0950-0618/Ó 2018 Elsevier Ltd. All rights reserved.
102 J. Xie et al. / Construction and Building Materials 188 (2018) 101–118

Nomenclature

Ap section area of prestressing strand f pyT yield strength of the prestressing strand at temperature
A0 cross section area of concrete beams T
Aps area of prestressing strand f puT ultimate strength of the prestressing strand at tempera-
As area of nonprestressed tension reinforcement ture T
As 0 area of compression reinforcement f ccD ultimate compressive strength of the concrete after
EcT elastic modulus of the concrete at temperature T freeze-thaw cycles
EccD elastic modulus of the concrete after freeze-thaw cycles f tT tensile strength of the concrete at temperature T
I0 the moment of inertia of the section h0 distance from extreme compression fibre to centroid of
La shear span of the beam nonprestressed longitudinal tension reinforcement
McrT bending moment at temperature T x equivalent height of the compressive concrete
Pcr crack-initiation resistances y0 distance from centroidal axis to the tensile edge of the
Pu ultimate resistance beams
T temperature e0T compressive strain of the concrete against the ultimate
W0 elastic resistance moment compressive strength f cT at temperature T
ap distance from extreme compression fibre to centroid of eccD strain against f ccD of the concrete after freeze-thaw cy-
prestressing strand cles
as 0 distance from extreme compression fibre to centroid of q reinforcement ratio of nonprestressed flexural rein-
longitudinal compression reinforcement forcement
b width of the concrete section q0 reinforcement ratio of compression reinforcement
ep0 distance from centroidal axis to preloading point qb reinforcement ratio of prestressing tendons
0
f cT specified compressive strength of concrete cylinder at scr , scr bond stress of cracking and its corresponding slip
temperature T su , su peak bond stress (bond strength) and its corresponding
f cT compressive strength of the concrete at temperature T slip
f yT yield strength of nonprestressed longitudinal tension sr , su residual bond stress and its corresponding slip
reinforcement at temperature T rpc prepressing stress at tensile edge of the concrete
f yT
0
yield strength of compression reinforcement at temper- rpe effective tensile stress in the prestressing strand
ature T
f ps stress in prestressing reinforcement at nominal flexural
strength

40 °C, 70 °C and 100 °C. It was observed that low temperature
can improve resistances of RC beams and a nonlinear finite ele-
ment model (FEM) was developed to predict the behaviours of
tested beams at low temperatures. DeRosa et al. [10] reported per-
formance of four RC beams at room temperature and 20 °C. Four-
point loads at service levels were applied to the tested beams for
48 h before the beams were loaded to failure. It showed that the
crack width decreased and ultimate resistance increased at low
temperature whilst the effect of temperature on stiffness was mar-
ginal. Mirzazadeh et al. [11] studied the performance of four large-
scale beams at 15 °C and 25 °C. They found that both their ductil-
ity and ultimate resistance were increased with the decrease of
temperature. The cracking resistance of beams increased while
the number and the depth of cracks were reduced at 25 °C.
Recently, Xie et al. [12] studied the behaviours of unbonded PC
beams within temperature ranges of +20 °C to 100 °C. Their test
results showed that the cracking load, yield and ultimate resis-
tances, and elastic stiffness all increased almost linearly with the
decreasing temperature.
The numerical analysis method has also been widely used to
investigate structural behaviours of PC beams. Mercan et al. [13]
presented a nonlinear 3-D FEM for PC spandrel beams to investi-
gate their sensitivities to different parameters, e.g., tension stiffen-
ing, finite element type and fracture energy. Yapar et al. [14]
developed a nonlinear FEM to predict the structural behaviours
of bonded PC beams especially on the damage behaviour of con-
crete and bond-slip behaviour between the concrete and pre-
stressing strand. Ayoub et al. [15] developed a FEM to study the
nonlinear responses of bonded PC girders, and an analytical
method was proposed to give an accurate prediction for the pre-
stressing operation. Arab et al. [16] presented two numerical meth-
ods to simulate the bonded concrete members and concluded that
embedment technique could accurately predict the behaviours of
bonded concrete members in comparison with the extrusion
method. Abdelatif et al. [17] developed a FEM to simulate the
transferring mechanism of prestress in PC elements. Post-
cracking behaviour and shrinkage of concrete were considered in
this model. A parametric study was performed to study the effect
of the concrete strength, diameter of prestressing steel and initial
prestress on prestress transfer between concrete and strand.
From these previous studies, it can be found that most of exper-
imental studies focused on the behaviours of RC beams at low tem-
peratures and there are still few studies focused on PC beams at
low temperatures especially on bonded PC beams. The numerical
investigations on PC beams mainly focused on their structural
behaviours in ambient temperature environment. The numerical
studies on structural behaviours of PC beams at low temperatures
have not been thoroughly studied. Thus, it is necessary to experi-
mentally investigate the structural performance of PC beams at
low temperatures and develop numerical model to predict their
behaviours at low temperatures.
This manuscript firstly reported 12 quasi-static tests on bonded
PC beams at low temperatures. The main investigated parameters
were prestress levels and temperatures (+20 °C to 100 °C). The
effects of these parameters on the behaviours of the bonded PC
beams were analysed and discussed. Theoretical models that can
predict the crack-initiation resistances and ultimate resistances
of bonded PC beams at low temperatures were developed. A
detailed numerical model on the behaviours of the bonded PC
beams at low temperatures was developed. The accuracies of these
theoretical and numerical models were both validated against the
test results. Based on the experimental results, the stress-strain
curves equations of concrete after freeze-thaw cycles were devel-
oped. Finally, parametric studies were performed to study the
 J. Xie et al. / Construction and Building Materials 188 (2018) 101–118 103

behaviours of bonded PC beams at low temperatures and after
freeze-thaw cycles.
2. Structural performance of bonded PC beams at low
temperatures
2.1. Specimens
Twelve bonded PC beams in total were prepared in this test pro-
gram. The investigated parameters contained the prestress levels
(0 and 0.75fpu) and the temperature levels (20 °C, 40 °C, 70 °C
and 100 °C). Thus, these twelve specimens can be divided into
four groups depending on different temperatures. Table 1 shows
the details of all tested specimens. All beams had the same cross
section size of 65 mm  120 mm (width  depth) and the length
of 1600 mm. The shear span of the beams was 350 mm and actual
span was 1400 mm. The vertical stirrups were steel bars with
diameters of 3.5 mm, which distributed in the shear zone of the
beams. The bottom and top longitudinal tendons were set only in
the shear zone. Each PC beam was pressed by a 3-wire prestressing
strand at the middle-bottom of the cross section. Its nominal diam-
eter equalled to 6.2 mm. The material properties of prestressing
strands were obtained by direct tensile tests and listed in
Table 1. Normal weight concrete (NWC) was used in all the tested
beams, and its mix proportions of cement: water: stand: stone:
fly ash: mineral powder: water-reducing additive were
1:0.44:1.88:0.22:0.28:0.04. The tensile and compressive strengths
of concrete at different temperatures were obtained by splitting
tensile and compression tests as listed in Table 1. The reinforce-
ments and dimensions of beams are shown in Fig. 1.
2.2. Test setup and loading procedures
Test setup of the bonded PC beams at different low temperature
levels is shown in Fig. 1. The cold storage was used to cool the
tested PC beams to the design temperatures. Then the beams were
moved into cooling chamber for quasi-static tests. During the test-
ing process, liquid nitrogen was sprayed into the cooling chamber
through an electromagnetic valve to maintain the target testing
temperature of the PC beams. The environmental temperature in
the cooling chamber was monitored by three thermocouples that
were fixed at left, middle and right regions of the cooling chamber.
As shown in Fig. 1, four thermocouples were installed on each side
of the surface of tested beam. And two thermocouples were
embedded into top and bottom regions of mid-span of each beam.
Once the target temperatures were achieved, followed quasi-static
loads were applied. All the PC beams were under two-point loads
in a displacement-controlled mode as shown in Fig. 1. The dis-
placement load from a hydraulic actuator was applied to the PC
beams through a spread beam. The reaction forces at different
loading levels were measured by a load cell that attached to the
actuator. Linear varying displacement transducers (LVDTs) were
installed underneath the beams to monitor the vertical deforma-
tions during the tests. Settlements of the supports were also mon-
itored by LVDTs. Five linear strain gauges were fixed at the mid-
span of the beams along the depth of the cross section with a ver-
tical spacing of 30 mm to measure the strains of the concrete. Two
linear strain gauges were also attached to the middle of prestress-
ing strands to record the developed strains. The lowest operating
temperature of linear strain gauge is 269 °C, which meets the
tests requirements.
2.3. Test results
2.3.1. General behaviours, failure modes and cracking distribution
Fig. 2 depicts the measured load versus mid-span deflection
curves of tested PC beams. Fig. 3 depicts the load versus strain
curves of representative prestressing strands at the mid-span of
the PC beams. These figures showed that the behaviours of the
beams at low temperatures were very close to those of ambient
temperature with three loading stages. At the first stage, PC beams
showed elastic deformation until the first crack appeared. The first
crack can be observed at the bottom portion of the cross section at
mid-span. Load versus deflection curves of the PC beams at the first
stage exhibited a linear manner. After achieving the cracking load,
the decrease in the slope of the load versus deflection curves
implied their reduced stiffness. Meanwhile, decreasing stiffness
in the load-strain curves was also observed due to the strain redis-
tribution from the concrete to the prestressing strand after con-
crete cracking. More cracks can be observed in the pure bending
regions of the PC beams and the cracks propagated vertically along
the depth of the cross section. At this stage, the mid-span deflec-
tions of the PC beams and the strains in prestressing strands
increased almost linearly with the increase of reaction forces. At
the third stage, the cracks were observed on the top fibres of con-
crete and the concrete gradually crushed with the increase of the
loading displacement. PC beams showed plastic behaviours with
plastic hinges developed at loading positions. The mid-span deflec-
tions of the beams rapidly increased and the crack widths also
increased due to the softening behaviour of the beams. Finally,
concrete crushing, excessive crack widths, or excessive flexural
deflections occurred to the PC beams as shown in Fig. 4. These
observations indicated that all the tested beams failed in flexure
modes.

Table 1
Details of the bonded PC beams.

Item T (°C) fcT0 (MPa) ftT (MPa) fpyT (MPa) fpuT (MPa) rcon (MPa) rpe (MPa)
CB1 20 56.55 5.89 1856.4 2042.9 0 0
PCB1-1 20 56.55 5.89 1856.4 2042.9 0.75fpuT 1026
PCB1-2 20 56.55 5.89 1856.4 2042.9 0.75fpuT 1026
CB2 40 79.29 7.69 1898.8 2084.9 0 0
PCB2-1 40 79.29 7.69 1898.8 2084.9 0.75fpuT 1026
PCB2-2 40 79.29 7.69 1898.8 2084.9 0.75fpuT 1026
CB3 70 90.66 8.05 1957.6 2168 0 0
PCB3-1 70 90.66 8.05 1957.6 2168 0.75fpuT 1095
PCB3-2 70 90.66 8.05 1957.6 2168 0.75fpuT 1095
CB4 100 102.03 8.25 2027.8 2165.5 0 0
PCB4-1 100 102.03 8.25 2027.8 2165.5 0.75fpuT 1095
PCB4-2 100 102.03 8.25 2027.8 2165.5 0.75fpuT 1095

T denotes testing temperature; fcT0 , ftT denote specified compressive strength of concrete cylinder and split tensile strength of concrete at temperature T, respectively; fpyT, fpuT
denote yield and ultimate strength of prestressing strand at temperature T, respectively; rcon denotes tension control stress; rpe denotes the effective prestress of prestressing
strands.
104 J. Xie et al. / Construction and Building Materials 188 (2018) 101–118

Fig. 1. Test setup and details of the bonded PC beams at low temperatures.

Fig. 2. Load-deflection curves of the bonded PC beams at different low temperatures.

Fig. 5 shows the cracks of tested PC beams at different low tem-
peratures. It can be observed that the effects of temperature and
prestress levels on distribution of cracks were marginal. But both
the low temperature and prestress levels exhibited improvements
on first cracking resistances. In addition, as the temperature
decreased, the brittleness of concrete increased. Damage of
concrete was caused by frost heaving at low temperatures. Thus,
the height and widths of crack increased with the decreasing
temperature.
The cracking loads, Pcr, and ultimate loads, Pu, were determined
from the load versus deflection curves of the tested specimens and
listed in Table 2.
 J. Xie et al. / Construction and Building Materials 188 (2018) 101–118 105

Fig. 3. Load-strain curves of the prestressing strand at mid-span.

(a) Concrete crushing in specimen (b) Excessive crack widths in specimen

PCB2

PCB3

PCB4

(c) Flexural failure mode of specimens PCB2, PCB3, PCB4 after test
Fig. 4. Failure mode of the bonded PC beams.

2.3.2. Strain distribution along the depth of cross section
Fig. 6 depicts representative horizontal strain distribution
curves along the depth of mid-span cross section at low tempera-
ture of 100 °C. It shows that horizontal strain of concrete at mid-
span was linearly distributed along its height at different loading
levels at 100 °C. Thus, the assumption of plane section was still
applicable to the bonded PC beams at low temperatures.
As shown in Fig. 6, the depth of the neutral axis at mid-span at
different load levels can also be observed. Typically, after the
prestressing force was applied to strands, PC beams appeared in
the inverted arch, and the position for the neutral axis of the cross
section moved from the middle to its top tensile fibre. As the
applied force increased, inverted arch effect was reduced and the
downward deflection of PC beams gradually increased. Once the
crack initiated, the neutral axis quickly moved towards the com-
pressive fibre of the concrete.
2.3.3. Effect of low temperature
The load-deflection curves at mid-span of tested PC beams at
different low temperatures of +20 °C, 40 °C, 70 °C and 100 °C
as shown in Fig. 2. With the decrease of temperature, the ultimate
resistance P u and initial stiffness of the PC beams increased. As the
temperature decreased from 20 °C to 40 °C, 70 °C, 100 °C, the
ultimate resistance of tested PC beams was averagely increased by
9%, 15% and 33%, respectively. Fig. 7 depicts the effect of the low
temperature on the crack-initiation resistance Pcr . As the tempera-
106 J. Xie et al. / Construction and Building Materials 188 (2018) 101–118

Fig. 5. Cracks distribution of the bonded PC beams in the bending regions at low temperatures.

ture decreased from 20 °C to 40 °C, 70 °C, 100 °C, the P cr for
tested beams with prestress levels of 0 and 0.75fpu were linearly
increased by 51% (24%), 83% (49%) and 108% (56%), respectively.
It can be observed that compared with Pu , low temperature
showed more significant effect on Pcr . These improvements of Pcr
and Pu by low temperature were due to the increased tensile and
compressive strengths of concrete, and yield and ultimate
strengths of steel strands at low temperatures. The load-strain
curves of prestressing strands at ambient and different low tem-
peratures are shown in Fig. 3. It can be seen that temperature
had quite limited effect on the load-strain behaviours of prestress-
ing strands.
seen that prestress levels had marginal effect on initial stiffness
and ultimate resistance P u , but exhibited significant influence
on crack-initiation resistances Pcr of the PC beams. As the
prestress level increased from 0 to 0.75fpu, the Pcr for the tested
beams at different low temperatures of 20 °C,40 °C, 70 °C and
100 °C, were averagely increased by 29.5%, 6.9%, 5.2% and 5.0%,
respectively.
3. Theoretical analysis on structural performance of bonded PC
beams at low temperatures
3.1. Cracking resistance

2.3.4. Effect of prestress level The cracking moment of the bonded PC beams can be
As shown in Fig. 2, the load-deflection curves of representative estimated by following equations in Chinese code GB50010-2010
tested beams at different prestress levels of 0 and 0.75fpu. It can be [18]:
 J. Xie et al. / Construction and Building Materials 188 (2018) 101–118 107

Table 2
Test results, analytical and numerical predictions of bonded PC beams at low temperatures.

Item Pcr (kN) Pu (kN) Pcr,a (kN) Pcr,a/Pcr Pu,a (kN) Pu,a/Pu Pcr,e (kN) Pcr,e/Pcr Pu,e (kN) Pu,e/Pu
CB1 5.9 19.1 8.1 1.37 19.4 1.02 8.1 1.37 23.5 1.23
PCB1-1 6.5 20.2 11.3 1.73 19.0 0.94 8.6 1.31 20.4 1.01
PCB1-2 8.9 20.0 11.3 1.27 19.0 0.95 8.6 0.94 20.4 1.13
CB2 8.9 21.0 6.9 0.77 20.0 0.95 8.3 1.07 22.6 1.12
PCB2-1 9.8 21.3 12.9 1.32 19.8 0.93 9.6 0.98 23.5 1.10
PCB2-2 9.3 21.9 12.9 1.39 19.8 0.90 9.6 0.94 23.5 1.12
CB3 10.9 22.0 7.2 0.66 20.9 0.95 8.8 1.02 24.5 1.08
PCB3-1 10.9 22.4 13.6 1.25 20.7 0.92 11.1 1.01 23.8 1.06
PCB3-2 11.9 22.5 13.6 1.14 20.7 0.92 11.1 0.95 23.8 1.22
CB4 12.3 26.9 7.4 0.60 20.9 0.78 11.3 0.91 27.3 1.02
PCB4-1 10.4 24.6 13.8 1.33 20.8 0.84 11.5 1.11 25.3 1.03
PCB4-2 13.6 25.5 13.8 1.01 20.8 0.81 11.5 0.85 25.3 0.99
Mean 1.15 0.91 1.04 1.09
Cov 0.33 0.07 0.16 0.08

Pcr, Pu denote resistances corresponding to first crack and ultimate resistance of PC beam, respectively; Pcr,a, Pu,a denote resistances corresponding to first crack and ultimate
resistance of PC beams predicted by theoretical models, respectively; Pcr,e, Pu,e denote resistances corresponding to first crack and ultimate resistance of PC beam predicted by
FE model, respectively.

T, which can be calculated by the following equation suggested by

Wang [19]:
 
f tT ¼ 1:45  1:02T60 f ta  160  C 6 T 6 20  C ð2Þ

where f ta denotes the tensile strength of concrete at ambient

temperature.
The prepressing stress on the tensile edge of the concrete is
expressed as rpc that can be determined as the following [18]:

rpe Ap rpe Ap ep0

rpc ¼  y0 ð3Þ
A0 I0
where rpe denotes effective tensile stress in the prestressing strand;
Ap denotes section area of prestressing strand; A0 denotes cross sec-
tion area of concrete beams and includes the equivalent concrete
area of the prestressing strand; ep0 denotes distance from centroidal
axis to preloading point; I0 is the moment of inertia of the section;
y0 denotes distance from centroidal axis to the tensile edge of the
Fig. 6. Strain distribution along the depth of the cross section. beam.
The cracking load of the PC beams can be calculated as the
following:
2M crT
Pcr ¼ ð4Þ
La
where La is the shear span of the beam.

3.2. Ultimate resistance

The ultimate bending moment of the bonded PC beams can be

estimated by the equations in ACI318-14 [20] as the following:
0 0
0:85f cT bx ¼ Aps f ps þ As f yT  As 0 f yT ð5Þ

 x

0 0
MuT ¼ 0:85f cT bx h0  þ As 0 f yT ðh0  as 0 Þ  Aps f ps h0  ap ð6Þ
2
where As , As 0 and Aps are the area of nonprestressed tension rein-
forcements, compression reinforcements and prestressing strand,
Fig. 7. The increment of the resistance corresponding to the crack initiation versus respectively; h0 , as 0 and ap are the distance from extreme compres-
the temperatures. sion fibre to centroid of nonprestressed tension reinforcements, lon-
gitudinal compression reinforcements and strand, respectively; b is

 the width of the cross section; x is the equivalent height of the com-
M crT ¼ rpc þ cf tT W 0 ð1Þ 0
pressive concrete and can be obtained by Eq. (5); f cT is the specified
where W 0 denotes elastic resistance moment; c denotes a coeffi- compressive strength of concrete cylinder at temperature T; f yT and
0
cient for W 0 that considers the effect of plastic deformation of the f yT are the yield strengths of nonprestressed tension reinforce-
concrete; f tT denotes tensile strength of concrete at temperature ments and compression reinforcements at temperature T,
108 J. Xie et al. / Construction and Building Materials 188 (2018) 101–118

respectively; f ps is the stress in prestressing reinforcement at nom- bonded PC beams at different low temperatures. ABAQUS/Standard
inal flexural strength. If all the prestressing strands are in the tensile type of implicit solver was adopted for solutions.
zone and f ps P 0:5f puT , the f ps for bonded PC beams can be calcu-
lated as the following: 4.2. Elements
  
cp f h
f ps ¼ f puT 1  qp puT0 þ 0 ðq  q0 Þ ð7Þ As shown in Fig. 8, the typical FEM for the bonded PC beams
b1 f cT ap that contained the concrete block, loading platen, support,
prestressing strand, non-prestressed reinforcements, and shear
where f puT denotes ultimate strength of the prestressing strand at
reinforcements. The concrete, support and loading platen were
temperature T, which is obtained from material tests at different
modelled using 3-D 8-node continuum elements in ABAQUS ele-
low temperatures and listed in Table 1; q, q0 and qb denote the rein-
ment library with reduced hourglass control and integration point
forcing ratios for nonprestressed tension reinforcements, compres-
(C3D8R). This element consists of one integration point and three
sion reinforcements and prestressing strand, respectively; cp
translation degrees of freedom at each node. The mesh sizes for
denotes a factor related to the strength of prestressing strand and the concrete, loading platen and support were selected to be
is obtained as the following: 10  10  10 mm3, 5  5  5 mm3 and 5  5  5 mm3 as shown
8
< 0:55 0:80 6 f pyT =f puT < 0:85
> in Fig. 8. Prestressing strand, non-prestressed reinforcements,
and shear reinforcements in PC beams were simulated using
cp ¼ 0:40 0:85 6 f pyT =f puT < 0:90 ð8Þ
>
: two-node three-dimensional truss element (T3D2). Perfect bond
0:28 f pyT =f puT P 0:90 between reinforcements (non-prestressed reinforcements and
b1 denotes a factor related to the strength of the concrete that can shear reinforcements) and concrete were assumed in the FEM.
be obtained as the following: For the bonded prestressing strand, the effect of bond-slip on PC
( 0
beams was considered at low temperatures. Thus, nonlinear spring
0:85 f cT 6 27:6 MPa element was introduced to simulate bond-slip between prestress-
b1 ¼ 0
27:6 0
ð9Þ
0:85  0:05 f cT 6:9 P 0:65 f cT > 27:6 MPa ing strand and the surrounding concrete.

4.3. Material models

3.3. Validations of theoretical analysis
The predicted values of P cr and P u of the beams are compared
with the experimental results in Table 2. It shows that the pre-
dicted P cr values are averagely 11% larger than the test data, and
the coefficient of variation (COV) is 0.34. Compared with P cr , the
theoretical models can give more accurate predictions on Pu . The
average test-to-prediction for Pu is 0.91 with a COV of 0.07. This
might be explained by the fact that accumulative discreteness of
the material performance and inaccuracy of the prestress loss mea-
surement in the tests. Thus, it can be concluded that the theoretical
models proposed in this study can give reasonable predictions on
Pcr and P u .
4. Finite element analysis on structural performance of bonded
PC beams at low temperatures
4.1. General
A detailed 3-D finite element model (FEM) was developed
adopting general commercial software package ABAQUS [21] for
4.3.1. Concrete
In ABAQUS material library, concrete damage plasticity model
(CDPM) was employed to simulate the behaviours of the concrete
in PC beams. CDPM adopts isotropic compression and tension plas-
ticity principles to describe its inelastic behaviours of concrete
with isotropic damage. In the CDPM, the yield function, proposed
by Lubliner et al. [22] and modified by Lee and Fenves [23], is
adopted to describe the evolution laws of strength of the concrete
under tension and compression. The CDPM also assumes the non-
associated potential flow rule to describe the relationship between
the stress and incremental plastic strain. A hardening law shows
that yield criterion and the flow rule change with the increase of
plastic strain. Isotropic damage is also considered in the CDPM to
represent degradation of stiffness.
In this model, the uniaxial compressive and tensile stress-strain
(r  e) behaviours need to be defined. The compressive r  e
curves of concrete used in this FEM at different low temperatures
were obtained by regression analysis as shown in Fig. 9. It showed
that the material properties of concrete at low temperatures were
quite different from those at ambient temperature. The equations

Support
Loading
Longitudinal Concrete platen
reinforcement

Prestressing strand

Shear
reinforcement

Fig. 8. FEM for bonded PC beam.

 J. Xie et al. / Construction and Building Materials 188 (2018) 101–118 109

In this manuscript, mechanical properties of concrete at ambi-

ent temperature can be obtained from tests. Thus, with the test
results at ambient temperature, the generalized stress-strain
curves for concrete at low temperatures finally used in FE model
were determined and shown in Fig. 9.
The tensile strength of the concrete at different low tempera-
tures can be obtained by Eq. (2). The tensile behaviour of the con-
crete was simulated by fracture energy cracking model. The
fracture energy used in the FE model can be calculated by the
following formula suggested by the CEB-FIP [24]:

0:7
f cT
Gf ¼ Gf 0 ð14Þ
10
where Gf 0 varies with the change of the coarse aggregate of the con-
crete, Gf 0 = 0.058 Nmm/mm2 for the d = 32 mm, Gf 0 = 0.030 Nmm/
mm2 for the d = 16 mm, Gf 0 = 0.025 Nmm/mm2 for the d = 8 mm; d
denotes the diameter of the coarse aggregate in the concrete; f cT
denotes ultimate compressive strength of the concrete at
Fig. 9. Compressive stress-strain curves used in the FE analysis.
temperature T.
There were additional plasticity material parameters need to be
of r  e curves suggested by Xie et al. [7] were used to predict the defined in the CDPM. Dilation angle that represented volumetric
material properties of concrete at different low temperatures with change beyond its elastic limit was determined to be 25° by
strain softening as the following: sensitivity analysis. The flow potential eccentricity and ratio of
8 h i the biaxial/uniaxial compressive strength adopted 0.1 and 1.16.
>
>
> f cT Aðec =e0T Þ þ ð3  2AÞðec =e0T Þ2 þ ðA  2Þðec =e0T Þ3 0
< 4.3.2. Steel
rc ¼ 6 ec =e0T 6 1
>
> h i1 Prestressing strands and nonprestressed reinforcements in the
>
: f ðe =e Þ Bðe =e  1Þ2 þ ðe =e Þ e =e > 1
cT c 0T c 0T c 0T c 0T PC beams were simulated by the nonlinear isotropic model that
adopted von-Mises yield criterion to define isotropic yielding. Typ-
ð10Þ ical bi-linear r  e curves with strain hardening were adopted in
this model shown in Fig. 10. The values of the yield and ultimate
where f cT is the compressive strength of the concrete at tempera-
strengths, elastic modulus and Poisson’s ratio of prestressing
ture T and can be obtained by Eq. (12); e0T is compressive strain
strands at different low temperatures can be obtained from mate-
against the ultimate compressive strength f cT ; rc and ec are com-
rial tests as listed in Table 1. The values of the yield and ultimate
pressive stress and strain of the concrete, respectively; A and B
strengths of the nonprestressed reinforcements at different low
are constants that can be determined by regression model as listed
temperatures can be calculated as the following [6]:
in Table 3.
(
Previous test results [9] showed that the low temperature sig- f ya e0:0012ðT 0 T Þ for HRB335
nificantly influenced mechanical properties of concrete. Empirical f yT ¼ ð15Þ
f ya e0:0007ðT 0 T Þ for HRB400
prediction formulae proposed by Yan and Xie [9] were used to
incorporate these influences of low temperatures on elastic modu- (
lus, ultimate compressive strength and strain at ultimate compres- f ua e0:0009ðT 0 T Þ for HRB335
f uT ¼ ð16Þ
sive strength of concrete as the following: f ua e0:0007ðT 0 T Þ for HRB400
EcT where f ya and f yT are the yield strength of reinforcement at ambient
IEc ¼ ¼ 0:0011T þ 1:08  160  C 6 T 6 20  C ð11Þ
Eca temperature and temperature T; f ua and f uT are the ultimate
strength of reinforcement at ambient temperature and temperature
f cT T; T0 denotes the ambient temperature, 20 °C. In this manuscript,
Ifc ¼ ¼ 0:0027T þ 1:036  160  C 6 T 6 20  C ð12Þ
f ca the yield strength (f ya ) and ultimate strength (f ua ) of the nonpre-

e0T
I e0 ¼ ¼ 0:0012T þ 0:88  160  C 6 T 6 20  C ð13Þ
e0a
where f ca , e0a and Eca denote ultimate compressive strength, strain
against f ca and elastic modulus at ambient temperature, respec-
tively; f cT , e0T and EcT denote ultimate compressive strength, strain
against f cT and elastic modulus at temperature T, respectively; Ifc , Ie0
and IEc denote increasing factor of f c , ec and Ec at different low
temperatures, respectively.

Table 3
Coefficients of A and B for concrete at different low temperatures.

Coefficient 20 °C 0 °C 40 °C 80 °C 120 °C 160 °C

A 2.7 2.7 2.2 1.8 1.6 1.5
B 0.7 1.3 1.7 2.0 5.0 6.0
Fig. 10. Bi-linear curve of steel used in the FE analysis.
110 J. Xie et al. / Construction and Building Materials 188 (2018) 101–118

stressed reinforcements at ambient temperature were 342 MPa and Table 4

455 MPa, respectively, which were obtained by material tests. Parameters of bond-slip curve between concrete and steel.

Parameters Cracking values Peak values Residual values

4.4. Simulation on bonded prestressing strand Stress 2.5ftT 3ftT ftT
Slip 0.025d 0.04d 0.55d
For the bonded PC beams, the effect of bond-slip on the
mechanical properties of beams needs to be well considered espe-
cially for the beams at low temperatures. The bond-slip model sug-
gested by GB50010-2010 [18] was chosen to predict the interfacial
behaviours between bonded prestressing strand and concrete,
which consists of four stages as shown in Fig. 11 and the equations
can be expressed as the following:
8 0 6 s 6 scr
> k1 s
>
< s þ k ðs  s Þ s 6 s 6 s
s¼ cr 2 cr cr u
ð17Þ
>
>
: us þ k 3 ð s  s u Þ s u 6 s 6 s r
sr s > sr

scr
k1 ¼
scr
su  scr
k2 ¼
su  scr
sr  su
k3 ¼
sr  su
where s and s denote the bond stress and its corresponding slip; scr
and scr denote the bond stress of cracking and its corresponding
slip; su and su denote the peak bond stress (bond strength) and its
corresponding slip; sr and sr denote the residual bond stress and
its corresponding slip. As shown in Table 4, these parameters can
be determined according to the diameter of prestressing
strand and the tensile strength of the concrete at different low
temperatures.
The nonlinear spring element was used to simulate the bond
characteristic between the strand and concrete at different
low temperatures. This element was adopted to integrate the
bond-slip model into FE analysis for describing the strand-
concrete interface behaviours. And the tangential stiffness kt and
normal stiffness kn of the spring element need to be defined. As
shown in Fig. 12, nodes associated with the concrete elements
were connected with nodes associated with the strand elements
by nonlinear spring elements.
ð18Þ
Fig. 12. Configuration of the concrete-steel bond.
ð19Þ
The force assigned to each spring element was related to the
number of defined spring elements and the spacing between them.
ð20Þ It was assumed that nonlinear spring elements were set with equal
spacing along the prestressing strand. The relations between force
and slip need to be input for the nonlinear spring elements.
The total force for each spring element can be calculated as
following:
F ¼ sA ð21Þ
s ¼ pdl ð22Þ
where s denotes the bond stress; F denotes the force of single
spring element; A denotes the contact area between strand and con-
crete; l denotes the spacing of each spring element and d denotes
the diameter of the prestressing strand.
4.5. Interactions, boundary conditions and loading
Contact type of surface-to-surface was adopted to define the
interactions between the support and concrete and between the
loading platen and concrete. A finite sliding algorithm that allows

Fig. 11. The bond-slip curve between concrete and steel used in the FE analysis.
 J. Xie et al. / Construction and Building Materials 188 (2018) 101–118 111

large deformations was used to describe the interfacial contact.
Contact surfaces of support and loading platen were taken as mas-
ter surfaces because of the relative higher stiffness and strength.
The surface-to-surface contact can obtain a smoother contour
and improve stability at the edges and corners. The risk of unde-
tected, large penetration of master surface into the slave surface
can be reduced via this approach. Penalty friction type and hard
contact of contact algorithms were adopted to describe the
interactions in the tangential and normal directions, respectively.
In the normal direction, hard contact can transfer the pressure
between the two contacting surfaces, and zero pressure will be
transferred once they are separated. In the tangential direction,
an isotropic Coulomb law was adopted to describe the interfacial
frictional effect. In this FE model, the friction coefficient between
beam and support as well as the friction coefficient between the
beam and loading platen were taken as 0.2.

Fig. 13. Comparisons of the FE predicted load-central deflection curves with the experimental curves.
112 J. Xie et al. / Construction and Building Materials 188 (2018) 101–118

In the tests, PC beams were simply supported and loaded by
two-point concentrated loads. The supports of the beams were
restrained against moving in all directions. In order to improve
the efficiency of calculation, two reference points were coupled
to the loading platens, and the displacement loads were then
applied to these two reference points.
4.6. Validations of FE analysis
The FE analysis predicted load versus deflection curves at mid-
span of the bonded PC beams are compared with test results in
Fig. 13. It shows that the predicted load versus mid-span deflection
curves are in good agreements with the tests. The Pcr and P u pre-
dicted by the FEM are compared with the corresponding test
results in Table 2. It shows that most of the predictions fall within
the scope of 10% error. The P cr of PC beams is averagely 4%
overestimated by the FEM with a COV of 0.16. The average
test-to-prediction for Pu is 1.09 with a COV of 0.08. The errors of
predictions may be caused by scatter of the strength of the
concrete and prestressing strand at different low temperatures.
The FE predicted load versus mid-span strain curves of prestressing
strand are compared with experimental results in Fig. 14. The
cracking distribution observed during the tests is compared with

Fig. 14. Comparisons of the FE predicted load-strain curves with the experimental curves.

Fig. 15. Comparisons of failure mode for beam at 100 °C between FE simulations and tests.
 J. Xie et al. / Construction and Building Materials 188 (2018) 101–118 113

Fig. 16. Details of the PC beams used in parametric studies.

Table 5
Details and results of different cases in FE parametric studies.

Item T (°C) fcu (MPa) ft (MPa) q (%) rpe (Mpa) Pcr (kN) Pu (kN)
A-1-1 20 C60 2.85 0.83% 800 6.61 28.74
A-1-2 20 C60 2.85 1.47% 800 6.89 34.66
A-1-3 20 C60 2.85 2.30% 800 7.01 43.43
A-2-1 40 C60 2.85 0.83% 800 7.34 31.24
A-2-2 40 C60 2.85 1.47% 800 7.68 37.56
A-2-3 40 C60 2.85 2.30% 800 7.75 46.99
A-3-1 80 C60 2.85 0.83% 800 7.57 32.38
A-3-2 80 C60 2.85 1.47% 800 7.87 39.26
A-3-3 80 C60 2.85 2.30% 800 7.99 49.48
A-4-1 120 C60 2.85 0.83% 800 7.79 35.24
A-4-2 120 C60 2.85 1.47% 800 7.95 43.21
A-4-3 120 C60 2.85 2.30% 800 8.21 53.95
B-1-1 20 C60 2.85 1.47% 400 4.74 32.16
B-1-2 20 C60 2.85 1.47% 800 6.61 34.66
B-1-3 20 C60 2.85 1.47% 1200 8.07 36.85
B-2-1 40 C60 2.85 1.47% 400 5.58 35.54
B-2-2 40 C60 2.85 1.47% 800 7.34 37.56
B-2-3 40 C60 2.85 1.47% 1200 9.44 39.89
B-3-1 80 C60 2.85 1.47% 400 5.82 37.61
B-3-2 80 C60 2.85 1.47% 800 7.57 39.26
B-3-3 80 C60 2.85 1.47% 1200 9.68 42.14
B-4-1 120 C60 2.85 1.47% 400 6.06 41.75
B-4-2 120 C60 2.85 1.47% 800 7.79 43.21
B-4-3 120 C60 2.85 1.47% 1200 9.92 45.73
C-1-1 20 C30 2.01 1.47% 800 5.70 31.24
C-1-2 20 C45 2.51 1.47% 800 6.23 32.98
C-1-3 20 C60 2.85 1.47% 800 6.61 34.66
C-2-1 40 C30 2.01 1.47% 800 6.29 33.68
C-2-2 40 C45 2.51 1.47% 800 6.90 35.79
C-2-3 40 C60 2.85 1.47% 800 7.34 37.56
C-3-1 80 C30 2.01 1.47% 800 6.48 35.34
C-3-2 80 C45 2.51 1.47% 800 7.12 37.74
C-3-3 80 C60 2.85 1.47% 800 7.57 39.26
C-4-1 120 C30 2.01 1.47% 800 6.66 38.67
C-4-2 120 C45 2.51 1.47% 800 7.32 41.08
C-4-3 120 C60 2.85 1.47% 800 7.79 43.21

fcu and ft denote cubic compressive strength and split tensile strength of concrete at ambient temperature, respectively; q denotes nonprestressed reinforcing ratio; rpe
denotes the effective prestress of prestressing strand; Pcr and Pu denote the resistance corresponding to the crack initiation and ultimate resistance of the bonded PC beams,
respectively.
114 J. Xie et al. / Construction and Building Materials 188 (2018) 101–118

FE analysis as shown in Fig. 15. It can be observed that the FE pre- behaviours of bonded PC beams at low temperatures in terms
dicted results agree well with the experimental results. of load-central deflection curves, load versus strain curves for
From the above comparisons, it can be confirmed that the prestressing strand, crack-initiation resistance and ultimate
proposed FEM offered reasonable predictions on structural resistance.

Fig. 17. Load-deflection curves of the bonded PC beams in parametric study.

 J. Xie et al. / Construction and Building Materials 188 (2018) 101–118 115

5. Parametric studies on structural performance of bonded PC
beams at low temperatures and after freeze-thaw cycles
5.1. Parametric studies on PC beams at low temperatures
Based on the validated FEM, the parametric studies were per-
formed. Fig. 16 shows the details of the bonded PC beams in the
parametric studies. The main parameter, i.e., low temperature con-
sidered four levels of 20 °C, 40 °C, 80 °C and 120 °C. The rest
parameters investigated at each low temperature levels were as
the following:
(a) Reinforcing ratio of 0.83%, 1.47% and 2.3% were chosen to
study the effect of nonprestressed reinforcement on beha-
viours of bonded PC beams at different low temperatures.
(b) Different effective tensile stress in the strands were selected,
i.e., 400 MPa, 800 MPa and 1200 MPa.
(c) Different grades of concrete, i.e., C30, C45 and C60 were
chosen in this study.
The details of different cases in this parametric study are listed
in Table 5.
Fig. 17 plots the load versus deflection mid-span curves of
bonded PC beams with different nonprestressed reinforcing ratio,
effective prestress and strength of the concrete at different low
temperatures. Table 5 lists the P cr and Pu of the beams obtained
from the parametric studies. These figures and table show that as
the nonprestressed reinforcing ratio increased from 0.83% to
1.47% and 2.3%, the Pu of the PC beams at different low tempera-
tures was averagely increased by 21% and 52%, respectively. The
influence of nonprestressed reinforcing ratio on P cr was quite lim-
ited and can be neglected. As the temperature decreased from
20 °C to 120 °C, the P u of PC beams with nonprestressed reinforc-
ing ratios of 0.83%, 1.47%, and 2.3%, were increased by 22%, 24% and
24%, respectively. As the temperature decreased from 20 °C to
120 °C, P cr (or P u ) for the PC beams with effective tensile stress
of 400 MPa, 800 MPa and 1200 MPa, were increased by 28%
(30%), 18% (25%), and 13% (24%), respectively. And it can be
observed that as the effective prestress increased from 400 MPa
to 800 MPa and 1200 MPa, P cr for the PC beams with different
low temperatures were averagely increased by 32% and 67% whilst
Pu increased slightly. With the increase of concrete strength and
decrease of temperature, P cr and P u of beams were both increased.
As the temperature decreased from 20 °C to 120 °C, Pcr (or Pu ) for
the PC beams with the concrete strength of C30, C45, and C60,
were increased by 17% (24%), 17% (25%), and 18% (25%),
respectively.
With the decrease of temperature, the strengths of concrete and
steel increased. Thus, as the nonprestressed reinforcing ratio and
strengths of the concrete and steel increased, the neutral axis of
the cross section moved up to the compressive fibre of the section.
This leads to the increased bending resistance of the cross section,
and finally increased P u of the bonded PC beams. As the effective
prestress and tensile strength of the concrete increased, the Pcr of
the bonded PC beams were increased.
5.2. Parametric studies on PC beams after freeze-thaw cycles
5.2.1. Stress-strain curves of the concrete after freeze-thaw cycles
With the validated FE model, these parametric studies on PC
beams after freeze-thaw cycles were carried out. Fig. 16 shows
the geometric details of the bonded PC beams in this parametric
study. Freeze-thaw cycles widely occurred to the PC members
located in the cold regions. And this issue also brought great dam-
age to these PC structures that compromised their structural per-
formance and service life. The freeze-thaw cycles have much
more significant influence on the material properties of concrete
than that of steel. Thus, an accurate simulation on the structural
performance of the PC beams after freeze-thaw cycles required
the accurate constitutive model of the concrete after freeze-thaw
cycles. The selection of steel models ignores the influences of
freeze-thaw cycles.
Twenty-one concrete prisms in total measuring
100 mm  100 mm  300 mm were prepared for freeze-thaw
cycles tests in compliance with Chinese code GB50081-2002 [25].
All the specimens were de-molded 24 h after casting, and main-
tained for 28 days in compliance with Chinese code GB50081-
2002 [25] (temperature 20  2°, relative humidity above 95%).

Table 6
Test and predicted results of concrete after freeze-thaw cycles.

Item TL (°C) N fccD (MPa) fccD,a (MPa) fccD,a/fccD eccD (l) eccD,a (l) eccD,a/eccD EccD (MPa) EccD,a (MPa) fccD,a/fccD
FT1-0-1 20 0 42.28 41.19 1.03 1830 2093 0.87 58,790 53,721 1.09
FT1-0-2 20 0 39.58 41.19 0.96 2100 2093 1.00 44,987 53,721 0.84
FT1-0-3 20 0 41.71 41.19 1.01 2350 2093 1.12 57,386 53,721 1.07
FT2-15-1 40 15 37.43 37.13 1.01 2310 2071 1.12 47,767 41,631 1.15
FT2-15-2 40 15 38.02 37.13 1.02 2040 2071 0.98 37,954 41,631 0.91
FT2-15-3 40 15 37.53 37.13 1.01 1930 2071 0.93 45,290 41,631 1.09
FT2-25-1 40 25 33.72 34.77 0.97 2680 2242 1.20 31,057 32,356 0.96
FT2-25-2 40 25 33.93 34.77 0.98 1950 2242 0.87 30,670 32,356 0.95
FT2-25-3 40 25 35.35 34.77 1.02 2090 2242 0.93 41,239 32,356 1.27
FT3-15-1 80 15 35.09 35.32 0.99 2270 2073 1.09 31,615 35,511 0.89
FT3-15-2 80 15 39.14 35.32 1.11 1830 2073 0.88 37,276 35,511 1.05
FT3-15-3 80 15 34.08 35.32 0.96 2190 2073 1.06 31,629 35,511 0.89
FT3-25-1 80 25 32.91 33.08 0.99 2260 2245 1.01 24,132 27,599 0.87
FT3-25-2 80 25 31.36 33.08 0.95 2380 2245 1.06 22,666 27,599 0.82
FT3-25-3 80 25 33.93 33.08 1.03 2010 2245 0.90 31,772 27,599 1.15
FT4-6-1 120 6 37.32 37.80 0.99 1780 1800 0.99 43,109 47,605 0.91
FT4-6-2 120 6 34.81 37.80 0.92 1660 1800 0.92 40,605 47,605 0.85
FT4-6-3 120 6 38.21 37.80 1.01 1830 1800 1.02 45,575 47,605 0.96
FT5-6-1 160 6 40.05 35.96 1.11 1670 1802 0.93 44,958 40,606 1.11
FT5-6-2 160 6 35.3 35.96 0.98 1890 1802 1.05 43,516 40,606 1.07
FT5-6-3 160 6 34.11 35.96 0.95 1940 1802 1.08 46,583 40,606 1.15
Mean 1.00 1.00 1.00
Cov 0.05 0.09 0.13

TL denotes the the lowest temperature in the freezing and thawing tests; N denotes the the number of freeze-thaw cycles; fccD, eccD and EccD denote the ultimate compressive
strength, strain against fccD and elastic modulus after freeze-thaw cycles, respectively; fccD,a, eccD,a and EccD,a denote the ultimate compressive strength, strain against fccD,a and
elastic modulus after freeze-thaw cycles predicted by theoretical models, respectively.
116 J. Xie et al. / Construction and Building Materials 188 (2018) 101–118

And the specimens adopted 28-day nominal design strength grade number of freeze-thaw cycles increased. The lowest temperature
of C60. The weight ratio of cement: water: river stand: stone was 1: in the freezing and thawing tests T L exhibited similar influence
0.44: 1.31: 2.78. Fine aggregate was adopted river sand (fineness on compressive r  e curves to that of the number of freeze-
modulus 2.8, medium sand) and coarse aggregate was adopted thaw cycles N. Based on the experimental results, empirical models
gravel with a diameter of less than 20 mm. Before tests, the spec- were established to predict the mechanical properties of the con-
imens were immersed in water at ambient temperature for 7 days crete after freeze-thaw cycles as the following:
to make the concrete saturated. The main investigated parameters
f ccD
included the lowest temperature in the freezing and thawing Ifc ¼ ¼ 1:342e0:00125T L N0:128  160  C 6 T L 6 20  C ð23Þ
f c0
tests T L and the number of freeze-thaw cycles N. All the tested
specimens were subjected to uni-axial compression after freeze-
eccD
thaw cycles and the r  e curves of the concrete were recorded. I e0 ¼ ¼ 0:649e2:58e5T L N0:155  160  C 6 T L 6 20  C ð24Þ
The details of the tests are listed in Table 6.
ec0
The representative compressive r  e curves of the concrete
EccD
after freeze-thaw cycles are shown in Fig. 18. The ultimate com- IEc ¼ ¼ 3:456e0:00398T L N0:493  160  C 6 T L 6 20  C ð25Þ
Ec0
pressive strengthf ccD , strain at ultimate compressive strength eccD
and elastic modulus EccD of concrete obtained from experimental where f c0 , ec0 and Ec0 are the ultimate compressive strength, strain
r  e curves and listed in Table 6. From Fig. 18, it can be observed against f c0 and elastic modulus before freeze-thaw cycles, respec-
that f ccD and EccD decreased with the increase of the number of tively; f ccD , eccD and EccD are the ultimate compressive strength,
freeze-thaw cycles N. However, eccD increased slightly with the strain against f ccD and elastic modulus after freeze-thaw cycles,

Fig. 18. Compressive stress-strain curves of concrete after freeze-thaw cycles.

 J. Xie et al. / Construction and Building Materials 188 (2018) 101–118 117

respectively; Ifc , Ie0 and IEc are decreasing factor of f c0 , ec0 and Ec0 In addition, since the empirical formulae developed in this
after freeze-thaw cycles, respectively. manuscript were developed based on limit experimental values
Table 6 compares the predicted f ccD , eccD and EccD of the concrete and further verifications were needed. Furthermore, the empirical
after freeze-thaw cycles with those test data. It can be observed formulae were applicable to the concrete with temperature inter-
that the analysis models can give reasonable predictions on f ccD , val of 20 °C to 160 °C and the number of freeze-thaw cycles of
eccD and EccD . The average test-to-prediction ratio for f ccD , eccD and 0 to 25.
EccD equal to 1.0. The coefficient of variations (COV) for f ccD , eccD
and EccD equal to 5%, 9% and 13%, respectively. 5.2.2. Discussions
In order to predict compressive r  e curves of the concrete Fig. 19 plots the load versus mid-span deflection curves of the
after freeze-thaw cycles, the constitutive model proposed by Car- bonded PC beams of the FE analysis with different number of
reira and Chu [26] was introduced in this manuscript as the freeze-thaw cycles N and lowest temperature in the freezing and
following: thawing tests T L . The ultimate resistance of the bonded PC beams
e
after freeze-thaw cycles are listed in Table 8. It can be observed
r ða þ 1Þ eccD that as the number of freeze-thaw cycles increased from 5 to 25
¼  b ð26Þ
f ccD e and the lowest temperature in the freezing and thawing tests T L
a þ eccD
decreased from 40 °C to 120 °C, the ultimate resistance was
decreased by 8.5%. As shown in Fig. 19, the bending stiffness of
f ccD the bonded PC beam also decreased.
b¼ þ 1:55 ð27Þ
32:4
where r and e denote compressive stress and strain of concrete, 6. Conclusions
respectively; f ccD and eccD in Eq. (26) can be obtained by Eqd. (23)
and (24); a is constant that can be determined by regression analy- In this manuscript, twelve bonded PC beams were firstly tested
sis as listed in Table 7. As shown in Fig. 18, the analytical results at different temperature levels (+20 °C, 40 °C, 70 °C and
agree well with test data. 100 °C). Then, theoretical models were proposed to predict the
behaviours of bonded PC beams at different low temperatures. A
3-D nonlinear FE model has been proposed for bonded PC beams
Table 7 with nonlinear material properties of both concrete and steel at
Coefficients of a for concrete after freeze-thaw cycles.
low temperatures. Numerical parametric studies were performed
Coefficient 20 °C 40 °C 80 °C 120 °C 160 °C with this developed FEM. Based on experimental results, theoreti-
N=0 1.43 1.35 1.30 1.25 1.10 cal and numerical analysis, the following conclusions are drawn:
N = 15 1.62 1.72 1.79 1.82 1.93
N = 25 1.74 1.84 2.20 2.21 2.48 1) The 12 quasi-static tests on bonded PC beams showed that
the low temperatures increased the cracking Pcr and ulti-
mate resistances Pu of the bonded PC beams. As the temper-
ature decreased from 20 °C to 40 °C, 70 °C and 100 °C,
the ultimate resistance of the bonded PC beams was
increased by 9%, 15% and 33%, respectively. As temperature
decreased from 20 °C to 40 °C, 70 °C, 100 °C, the Pcr
for tested beams with prestress levels of 0 and 0.75fpu were
linearly increased by 51% (24%), 83% (49%) and 108% (56%),
respectively.
2) Prestress levels showed significant influence on the P cr of the
PC beams, but exhibited marginal influences on the elastic
stiffness and ultimate resistance of bonded PC beams at
low temperatures. As prestress level increased from 0 to
0.75fpu, the Pcr for the tested beams at different low temper-
atures of 20 °C,-40 °C, 70 °C and 100 °C, were averagely
increased by 29.5%, 6.9%, 5.2% and 5.0%, respectively.
3) A FEM was proposed to simulate the structural behaviours of
bonded PC beams at low temperatures. This FEM was devel-
oped on the basis of using material models at low tempera-
Fig. 19. Load-deflection curves of the bonded PC beams after freeze-thaw cycles. tures and interaction between the concrete and prestressing

Table 8
Details and results of different cases for freeze-thaw cycles in FE parametric studies.

Item N TL (°C) a b fccD (MPa) eccD (l) EccD (MPa) rpe (MPa) Pu (kN)
D-1-1 5 40 1.54 2.78 40.00 1492 47,989 800 38.00
D-1-2 15 40 1.72 2.62 34.75 1769 27,921 800 36.94
D-1-3 25 40 1.84 2.59 33.82 1915 21,705 800 36.14
D-2-1 5 80 1.55 2.61 34.34 1494 40,927 800 37.61
D-2-2 15 80 1.79 2.45 29.03 1771 23,811 800 36.77
D-2-3 25 80 2.20 2.40 27.62 1917 18,510 800 36.51
D-3-1 5 120 1.35 2.39 27.29 1495 34,903 800 36.06
D-3-2 15 120 1.82 2.25 22.56 1773 20,307 800 34.95
D-3-3 25 120 2.21 2.19 20.88 1919 15,786 800 34.79
118 J. Xie et al. / Construction and Building Materials 188 (2018) 101–118
strand at low temperatures. The accuracy of the FEM was
validated by test results, and it proved to be capable of sim-
ulating load-central deflection curves, cracking resistance,
and ultimate resistance of the PC beams at low tempera-
tures. The FE model averagely overestimated the Pcr of the
PC beams at low temperatures by 4% with the COV of 0.16
for 12 predictions whilst the model slightly overestimated
Pu by 9% with a COV of 7% for 12 predictions.
4) Based on the validated FE model, parametric studies were
performed and the results showed that as the temperature
decreased, nonprestressed reinforcing ratio increased and
the concrete strength increased, the ultimate resistance of
the bonded PC beams increased. As the temperature
decreased, effective tensile stress increased and concrete
strength increased, the cracking resistance of bonded PC
beams increased.
5) Compressive tests were carried out on concrete after freeze-
thaw cycles to obtain the corresponding material properties.
The test data showed that the modulus of elasticity and ulti-
mate compressive strength of concrete decreased with the
increasing freeze-thaw cycles and the decreasing of lowest
temperature. Based on the experimental results, regression
models were proposed to predict compressive r  e curves
of the concrete after freeze-thaw cycles. The FE parametric
studies on bonded PC beams after freeze-thaw cycles were
performed based on these material models after freeze-
thaw cycles. The studies showed that with the increase of
the number of freeze-thaw cycles and decrease of the lowest
temperature in the freezing and thawing tests, the ultimate
resistance and stiffness of the bonded PC beams were
decreased.
Acknowledgments
This work was financially funded by the National Natural
Science Foundation of China under project No. 51478309 and
51608358 and by Tianjin University under project No. 2018XRG-
0019. The authors gratefully express their gratitude for these finan-
cial supports.
Compliance with ethical standards
Funding: This study was funded by National Natural Science
Foundation of China (grant number 51478309 and 51608358).
Conflict of interest: The authors declare that they have no con-
flict of interest.
References
[1] T. Koh, Y. Bae, M. Shin, S. Hwang, Experimental study on transfer length of an
eco-friendly prestressed concrete sleeper, Constr. Build. Mater. 109 (2016) 25–
33.
[2] Y. Qiao, H.F. Wang, L.C. Cai, W. Zhang, B.H. Yang, Influence of low temperature
on dynamic behavior of concrete, Constr. Build. Mater. 115 (2016) 214–220.
[3] D.L. Gautier, K.J. Bird, R.R. Charpentier, et al., Assessment of undiscovered oil
and gas in the arctic, Science 324 (5931) (2009) 1175–1179.
[4] A. Palmer, K. Croasdale, Arctic Offshore Engineering, World Scientific
Publishing, Singapore, 2013.
[5] B.B. Kanbur, L.M. Xiang, S. Dubey, F.H. Choo, F. Duan, Cold utilization systems
of LNG: a review, Renewable Sustainable Energy Rev. 79 (2017) 1171–1188.
[6] J.B. Yan, J. Xie, Experimental studies on mechanical properties of steel
reinforcements under cryogenic temperatures, Constr. Build. Mater. 151
(2016) 661–672.
[7] J. Xie, X. Li, H. Wu, Experimental studies on the axial-compression performance
of concrete at cryogenic temperatures, Constr. Build. Mater. 72 (2014) 380–
388.
[8] S. Liu, Q.H. Huang, X.L. Gu, W.P. Zhang, Study on bending bearing capacity of
reinforced concrete beams at super-low temperature, J. Build. Struct. S 2
(2009) 86–91.
[9] J.B. Yan, J. Xie, Behaviours of reinforced concrete beams under low
temperatures, Constr. Build. Mater. 141 (2017) 410–425.
[10] D. DeRosa, N.A. Hoult, M.F. Green, Effects of varying temperature on the
performance of reinforced concrete, Mater. Struct. 48 (2015) 1109–1123.
[11] M.M. Mirzazadeh, M. Noel, M.F. Green, Effects of low temperature on the static
behaviour of reinforced concrete beams with temperature differentials, Constr.
Build. Mater. 112 (2016) 191–201.
[12] J. Xie, X. Chen, J.B. Yan, G.C. Lei, L.M. Zhu, Ultimate strength behavior of
prestressed concrete beams at cryogenic temperatures, Mater. Struct. 50
(2017) 81.
[13] B. Mercan, A.E. Schultz, H.K. Stolarski, Finite element modeling of prestressed
concrete spandrel beams, Eng. Struct. 32 (9) (2010) 2804–2813.
[14] O. Yapar, P.K. Basu, N. Nordendale, Accurate finite element modeling of
pretensioned prestressed concrete beams, Eng. Struct. 101 (2015) 163–178.
[15] A. Ayoub, F.C. Filippou, Finite-element model for pretensioned prestressed
concrete girders, J. Struct. Eng. 136 (4) (2010) 401–409.
[16] A.A. Arab, S.S. Badie, M.T. Manzari, A methodological approach for finite
element modeling of pretensioned concrete members at the release of
pretensioning, Eng. Strut. 33 (6) (2011) 1918–1929.
[17] A.O. Abdelatif, J.S. Owen, M.F.M. Hussein, Modelling the prestress transfer in
pre-tensioned concrete elements, Finite Elem. Anal. Des. 94 (2015) 47–63.
[18] 50010–2010, Code,, for design of concrete structures, China Standards Press,
Beijing, 2010.
[19] C.X. Wang. In: Experimental research on properties of concrete under low-
temperature. Tianjin China: Tianjin University, 2010 (Chinese).
[20] American Concrete Institution 318-14 (ACI, 2014) Building code requirements
for structural concrete (ACI 318-14) and commentary, American Concrete
Institute, Farmington Hills.
[21] ABAQUS. ABAQUS standard user’s manual, version 6.12. Providence, RI(USA):
Dassault Systemes Corp, 2012.
[22] J. Lubliner, J. Oliver, S. Oller, E. Onate, A plastic-damage model for concrete, Int.
J. Solids Struct. 25 (3) (1989) 299–326.
[23] J. Lee, G.L. Fenves, Plastic-damage model for cyclic loading of concrete
structures, J. Eng. Mech. ASCE 124 (8) (1998) 892–900.
[24] CEB-FIP. CEB-FIP model code 1990, Trowbridge, Comité Euro-International Du
Béton, Redwood Books, Wiltshire, UK, 1993.
[25] GB/T 50081-2002, Standard for test method of mechanical properties on
ordinary concrete, China Standards Press, Beijing, 2002.
[26] D. Carreira, K. Chu, Stress-strain relationship for plain concrete in compression,
J. ACI. Struct. 82 (11) (1985) 797–804.