Professional Documents
Culture Documents
ECC Composite Beams
ECC Composite Beams
ECC Composite Beams
Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct
a
Beijing Engineering Research Center of Steel and Concrete Composite Structures, Tsinghua University, Beijing 100084, China
b
Dept. of Civil Engineering, Tsinghua University, Beijing 100084, China
c
Key Lab. of Civil Engineering Safety and Durability of China Education Ministry, Dept. of Civil Engineering, Tsinghua University, Beijing 100084, China
Keywords: Due to the strain-hardening and multi-cracking properties, engineered cementitious composite (ECC) is a new
Reinforced engineered cementitious composite solution to the cracking issue in the negative bending moment region of steel–concrete composite beams. This
Steel–ECC composite beam paper presents an experimental and analytical study on the flexural performance of composite steel–ECC beams
Negative bending moment subjected to negative bending moments as well as the tension stiffening behaviour of reinforced ECC (R/ECC)
Tension stiffening
flange slabs. Three beams with different slab materials and reinforcement ratios were tested under a hogging
Crack width
Analytical model
moment. Experimental results demonstrated significant enhancement in stiffness and crack resistance for the
steel–ECC composite beams. A practical four-parameter fibre-bridging model was established to describe the
strain-hardening behaviour of different ECC materials. Then, a modified analytical tension-stiffening model for
R/ECC was formulated considering the strain-hardening behaviour of ECC and rebar–ECC bond–slip interaction
based on the conventional tension-stiffening model for reinforced concrete. This modified model was verified by
several direct tensile tests of R/ECC members. In addition, by applying the model to the traditional fibre
beam–column element model, the mechanical performance and crack opening of general R/ECC structures were
derived. The simulation results of the steel–ECC composite beams demonstrated satisfactory accuracy compared
with the test results. Finally, a parametric study based on the new model was conducted to identify the influence
of several important material and structural parameters on the flexural performance of steel–ECC composite
beams under negative moments.
Corresponding author.
⁎
E-mail addresses: fanjsh@mail.tsinghua.edu.cn (J. Fan), gskthu@163.com (S. Gou), dingran@mail.tsinghua.edu.cn (R. Ding),
junz@mail.tsinghua.edu.cn (J. Zhang), shi-zj08@139.com (Z. Shi).
https://doi.org/10.1016/j.engstruct.2020.110309
Received 31 July 2019; Received in revised form 3 January 2020; Accepted 29 January 2020
Available online 19 February 2020
0141-0296/ © 2020 Elsevier Ltd. All rights reserved.
J. Fan, et al. Engineering Structures 210 (2020) 110309
test results showed that ECC link slabs exhibited a large strain capacity of 8 mm, a height of 45 mm, and a spacing of 50 mm, which were
and durability. Afefy et al. [19] and Zhang et al. [20] studied the designed to achieve sufficient shear connection between the steel beam
flexural performance of layered ECC–concrete composite beams, which and concrete/ECC slab. Transverse reinforcement included 6 mm
exhibited higher flexural strength, ductility, and cracking resistance. In HPB235 bars with a spacing of 200 mm at top of the slab, and 8 or
addition, several numerical [21] and theoretical studies [22] have been 10 mm HPB235 bars with a spacing of 70 mm at bottom of the slab. The
conducted to simulate the bending performance of ECC beams. How- longitudinal reinforcement of SCB-1 and SEB-1 included eight HRB335
ever, there were few experimental and analytical studies on the bending longitudinal bars with a diameter of 12 mm. The longitudinal re-
behaviour of steel–ECC composite beams. inforcement of SEB-2 contained eight HRB335 bars with a diameter of
The tension-stiffening effect of reinforced ECC (R/ECC) flange slabs 8 mm. The basic parameters of the test specimens are summarized in
in steel–ECC composite beams under negative moments is closely re- Table 1.
lated to the stiffness and crack width of composite beams. Significant The tensile strength of steel plates used in the I-shaped beams and
tension-stiffening effects were observed in several typical direct tension steel bars used in the slabs are shown in Table 2. The initial loading
tests of R/ECC members [23]. Fischer and Li [24] revealed that the strain rate was 30 με/s. After yielding, the loading rate was increased to
strain difference between ECC and rebar was negligible in direct tension around 500 με/s. For steel–concrete composite beams, three
test. Kang et al. [25] revealed that the increase of the reinforcement 150 mm × 150 mm × 150 mm concrete cubes were cast and cured
ratio improved the overall deformation capacity of the test members; under the same environmental conditions. The measured average
hence, a minimum reinforcement ratio should be required to ensure compressive strength was 55.4 MPa. The PVA-ECC material was pro-
adequate ductility. Tension stiffening effect is determined by crack duced with PVA fibres that had a tensile strength of 1620 MPa, a length
spacing, bond–slip behaviour, and the crack–bridging constitutive law. of 12 mm, and a volume fraction of 1.7%. The measured properties of
So far, many efforts have been devoted to study the bond behaviour the ECC are shown in Table 3.
[26–28] and crack–bridging constitutive law [29–33] of R/ECC mem-
bers. While most research focused on the experimental results of R/ECC 2.2. Loading apparatus and measurement
members, few theoretical approaches have been developed.
Therefore, this study focuses on the negative flexural performance As shown in Fig. 2, the composite beams were simply supported on
of steel–ECC composite beams and the tension stiffening behaviour of top of the slab with a span of 3 m. Four-point bending tests were applied
R/ECC flange slabs through experimental and analytical methods. An with a 600 mm-long pure bending region. A transverse distributive
experimental program including two steel–ECC and one steel–concrete girder was employed between the actuator and the beam to guarantee a
composite beams was designed to examine the failure mode, loading uniform distribution of the force. Stiffening ribs were added to the
capacity, stiffness, ductility, and crack propagation under negative support and the loading point of the steel beam to avoid local bucking
moments. Then, a modified analytical tension-stiffening model for R/ caused by stress concentration. The concrete/ECC slab was constructed
ECC was formulated considering the strain hardening of ECC and on the propped steel beam, and the tests were carried out two months
rebar–ECC bond–slip interaction. This model was based on the con- later. The loading history consisted of a force-controlled stage followed
ventional tension-stiffening model for reinforced concrete. By applying by a displacement-controlled stage after the flange of the steel beam
the model to the traditional fibre beam–column element model, the yielded. The load level increased by 5 kN during the force control stage
flexural performance of steel–ECC composite beams was thoroughly to help carefully observe the deformation and crack propagation of the
investigated and crack width was specifically calculated. test specimens.
The measurement scheme is shown in Fig. 3. Linear variable dif-
ferential transformers (LVDT) were employed to measure the dis-
2. Experimental programme placement at the middle and at the supports of the beams. Ex-
tensometers were employed to measure slippage between the steel
2.1. Specimen design and material properties beam and the concrete/ECC slab. The strain distribution along the
height of three critical sections was recorded by strain gauges.
Based on the focus of the study, three specimens were designed to
experimentally evaluate the flexural performance of steel–ECC com- 2.3. Experimental results and analysis
posite beams under hogging moments. Fig. 1 shows the design details of
the cross sections including dimensions and reinforcement details. Di- 2.3.1. Observed phenomenon
mensions were identical for all specimens with a length of 3200 mm, a For specimen SCB-1 with a concrete slab under negative loading, the
height of 250 mm and a width of 600 mm. The steel beam had an I- first flexural crack propagated at a load of 10 kN. The crack width
shaped cross section with a height of 180 mm and a width of 100 mm. reached 0.2 and 0.5 mm at a load of 60 and 145 kN, respectively. As the
The thickness of the web and the flange were 6 and 8 mm, respectively. load increased towards the peak load, buckling of the bottom flange
The width and thickness of the concrete/ECC slab were 600 and 70 mm, was observed. In addition, splitting cracks were observed on the con-
respectively. The shear studs were placed in two rows with a diameter crete slab parallel to the direction of the longitudinal reinforcement.
The failure mode is shown in Fig. 4 (a).
For specimen SEB-1 with an ECC slab, the first visible flexural crack
was observed at a load of 55 kN. The crack width increased slowly
within the pre-yielding stage, while the number of cracks increased
rapidly. The peak load Pp denotes the maximum load during the loading
process. When the load reached Pp, a large number of fine cracks were
generated at the top of the ECC slab. The bottom steel plates and bars
yielded and buckling was clearly observed at the bottom flange of the
steel beam. The failure mode of SEB-2 was similar to that of SEB-1
except for a higher crack width and larger crack spacing. Details of the
strain development and crack propagation will be described later in this
paper.
After failure of the negative loading specimens, a comparison of the
Fig. 1. Cross-section of test specimens. top slabs was carried out and shown in Fig. 4 (d). Compared with
2
J. Fan, et al. Engineering Structures 210 (2020) 110309
Table 1
Basic parameters of test specimens.
Specimen Slab material Steel beam dimensions (mm) Slab dimensions (mm) Longitudinal reinforcement Reinforcement ratio
Table 2
Tensile strength of steel plates and reinforcement bars.
Material Thickness/ Yield strength Ultimate Elongation ratio
Diameter (mm) fy (MPa) strength fu
(MPa)
Table 3
Material properties of the ECC.
ECC for SEB-1 & SEB-2 Concrete for SCB-1
Compressive strength fcu,E (MPa) Tensile strength ft,E (MPa) Elastic modulus EE (MPa) Poisson's ratio Compressive strength fcu (MPa)
3
J. Fan, et al. Engineering Structures 210 (2020) 110309
(a) (b)
SEB-2
SCB-1
SEB-1
(c) (d)
Fig. 4. Failure mode of (a) SCB-1 (b) SEB-1 (c) SEB-2 and (d) top slabs of the negative loading specimens.
4
J. Fan, et al. Engineering Structures 210 (2020) 110309
Table 4
Summary of test results.
Specimen Cracking load Pcr (kN) Yielding load Peak load Ultimate load δu /δy
Fig. 6. Strain distribution at mid-span section of specimens (a) SCB-1, (b) SEB-1 and (c) SEB-2.
5
J. Fan, et al. Engineering Structures 210 (2020) 110309
Fig. 8. Strain comparison of rebar and ECC/concrete in specimens (a) SCB-1, (b) SEB-1 and (c) SEB-2.
100 45 45
30
90 25 55 50 100 75
50
18 50 25 130
100 160 25
(a)
100 75 110
Side Span Middle Span Side Span
110 120
120 120
75
120 110 120 120
75 120
130 100 55
75 70 80 65 90
120 90 90
120 100 110 70
180 75 75 110
75
(b)
60
70
Side Span 90 Middle Span Side Span
120 90 80 70 90
60 60
130 120 110 70
100 80 80 75 90
75
120 100 55 110 110
60 80
80
130 70 75 60 65 80 110
80 130 70 55 55 80
70
(c)
Fig. 10. Crack development at top of the slab in specimens (a) SCB-1, (b) SEB-1 and (c) SEB-2.
6
J. Fan, et al. Engineering Structures 210 (2020) 110309
Fig. 11. Stress distribution of (a) a typical negative bending composite beam in (b) elastic method, (c) plastic method neglecting the tensile strength of ECC, and (d)
plastic method considering the tensile strength of ECC.
Table 5
Comparison between theoretical and experimental flexural strength.
Specimen Test value Elastic Plastic analysis Plastic analysis
(kN⋅m) analysis (neglecting ECC) (considering ECC)
(kN⋅m) (kN⋅m) (kN⋅m)
x B: Bonding region
PVA fibres ECC B CC B C: Cracking region
Rebar
lB lC lC lB
Lm
7
J. Fan, et al. Engineering Structures 210 (2020) 110309
Fiber stress
ECC stress
Matrix stress
(a) (b)
Fig. 14. Ideal distribution of stresses between cracks in the (a) separate model
and (b) integral model of R/ECC members. Fig. 15. A simplified bridging stress–crack width model for reinforced ECC in
structural calculation.
and fibres slide out of the matrix. In region C, the tension force of the
structural analysis.
matrix transfers to both rebar and fibres. Thus, the bridging stress of
A simplified four-parameter crack bridging model (Eq. (2)) is pro-
fibres is now lower than the stress of the matrix. The tension behaviour
posed to facilitate the application in numerical calculation, and the
of the matrix and fibres are considered separately in many studies on
accuracy is verified later in this paper. The bridging model consists of
unreinforced ECC materials, and many crack–bridging relations are
proposed through this separate model. However, considerable para- two quadratic curves. And the first derivative of the curve is con-
tinuous. As shown in Fig. 15, cracking stress σc denotes the cracking
meters of fibres and matrix bring difficulties in simulating large-scale
R/ECC structures. Another way to quantify the tensile contribution of stress of the matrix. Ultimate stress σu and ultimate crack width ωu
denote the maximum bridging stress and the corresponding crack width
ECC in R/ECC members is to consider fibres and matrix as a whole
during the loading process, respectively. Failure width ω0 denotes the
material. As shown in Fig. 14 (b), the integral model consists of only
crack width when the bridging stress drops to zero. The ultimate crack
two materials. Numerous research on crack–bridging relations makes it
width ωu could be negative depending on the balance of fibre and
possible to employ the integral model in numerical analyses of R/ECC
matrix. In this situation, the fibre reinforced matrix exhibits a tension-
structures.
softening effect instead of a strain-hardening effect. Each parameter
The crack bridging relationship is key to the strain-hardening effect
represents an explicit physical characteristic and all parameters are
in R/ECC integral models. So far, many analytical models have been
easy to determine in tension tests. Besides, the simplified model can
proposed to describe the fibre-bridging law based on different as-
simulate most crack bridging curves in Table 6 with sufficient accuracy
sumptions and conditions. Table 6 summarizes part of the typical fibre-
in structural simulations. Ultimate stress σu and cracking stress σc can
bridging models developed since 1992. Li et al. [7] established the first
be derived from the tensile strength and the crack strength during ECC
model predicting the complete bridging stress–crack width curve for
direct tension test. As suggested by several uniaxial tension tests
ECC. This model considers the randomly distributed fibres and the
[30,32], the recommended ranges for ωu and ω0 are 0.1–0.2 mm and
snubbing effect. Since then, more and more mechanisms have been
0.8–1.0 mm, respectively, for ECC materials with stable multiple
added to extend the adaptability of the original model. Yang et al. [30]
cracking and strain hardening behaviour.
proposed a numerical fibre-bridging constitutive law considering al-
most all fibre-matrix mechanisms including fibre rupture, chemical c u u
2
u
( 2
bonding, slip-hardening, matrix spalling, Cook-Gordon effect and two- 2 u) + u for ( u) ( u
u 0 )( c u)
=
way pull-out effects. Several years later, Huang et al. [31] introduced a u( c u)
2
( 0 )2 for ( u) > u u
simplified bridging model with explicit expressions. This model ignores ( u 2
0) ( c u) + u
2
u ( u 0 )( c u)
several effects with relatively less impact including matrix spalling and (2)
Cook-Gordon effect. Even so, dozens of micromechanical parameters
and complex computational processes make it difficult to apply in
structural simulations. Some of these parameters are difficult to obtain 4.1.5. Tension stiffening analysis procedure for R/ECC members
even in structural labs. A simplified crack bridging model with para- The analytical model for R/ECC members subjected to uniaxial
meters that are easy to obtain through material tests is needed in tension is derived based on the tension model for RC members proposed
Table 6
Current analytical fibre-bridging models for ECC materials.
Theoretical model by Proposed year Snubbing effect Fibre rupture Chemical bonding Slip-hardening Matrix spalling Two-way pullout Fatigue
8
J. Fan, et al. Engineering Structures 210 (2020) 110309
Fig. 16. Ideal stress distribution of the R/ECC model considering strain-hardening property.
by Xu et al. [41]. The ideal stress distribution of the R/ECC tension derived. The average bond stress ¯ is also calculated based on Eq. (1).
model is shown in Fig. 16. At T0, ECC stress is lower than the cracking Then, the stress–strain distribution (σe(x) and εe(x)) of uncracked ECC
stress and thus no crack forms. At T1, a crack forms at region C and a in region B can be obtained. The stress–strain distribution (σr(x) and εr
bridging stress of the ECC material is generated between cracks. After (x)) of rebar can be derived from the constitutive law and the equation
the maximum load is reached, the strain hardening and tension stif- of equilibrium. Finally, the overall load P, average stress ¯ , and the
fening effects start to decrease. The function σ(x) denotes the dis- corresponding average strain ¯ are obtained. In this way, a unified R/
tribution of sectional stress along crack spacing. The function τ(x) de- ECC tension model combining reinforcement and ECC has been estab-
notes the distribution of interface shear stress between the lished. During the elastic stage, the average stress–strain of the R/ECC
reinforcement bar and matrix, which can be derived from Eq. (1) by members can be derived as follows:
assuming that the global slip equals half the crack width and that the
(L m ) r (L m )2
bond stress is uniform. Stress functions σe(x) and σr(x) denote the stress ¯= + e( 2) + ( 2)
distribution of the ECC and reinforcement bar, respectively. The Lm Ee L m Ee L m d r (9)
average stress ¯ and average strain ¯ of R/ECC members can be defined
r (L m )2
as follows: ¯e = e( 2) + ( 2)
Lm dr (10)
Lm
e (x )· A e + r (x )· Ar 2 2 e (x )· A e+ r (x )· Ar ¯e + r ¯r
¯ = = dx =
A e + Ar Lm 0 A e + Ar (1 + r) Er E (L ) r Er (L m )2
¯r = + r m e( 2) + ( 2)
(3) Lm Ee L m Ee L m dr (11)
Lm
2 2 2 lB r Er r Er (L m ) + Ee L m
¯= r (x ) dx = + e (x ) dx ¯ = + e( )
Lm 0 Lm Lm 0 (4) (1 + r ) Lm (1 + r ) Ee Lm
where Ae and Ar denote the area of ECC and reinforcement bars, re- r ( r Er + Ee )(L m )2
+ ( 2)
spectively; ρr denotes the reinforcement ratio of the R/ECC section, (1 + r ) Ee L m d r (12)
calculated as Ar/(Ar + Ae)≈Ar/Ae; ω denotes the crack width. Eq. (4)
where Ee and Er denote the elastic modulus of ECC and reinforcement,
also represents the deformation compatibility relationship of the R/ECC
respectively; ¯e , and ¯r denote the average stress of ECC and rebar,
member. The equations of equilibrium in section B can be derived as
respectively. The derivation of the tension model is shown in the
follows:
Appendix A. This tension model can provide both the constitutive law
Lm
2 (x )·Cr and the corresponding crack width for R/ECC members. In ECC struc-
e (x ) = + dx
e0
x Ae (5) tures, the mechanical behaviour of R/ECC members can be pre-ob-
tained through the R/ECC tension model. By applying the behaviour of
R/ECC members to the fibre beam model, the mechanical performance
Lm
2 (x )· Cr
r (x ) = dx
r0
x Ar (6) and cracking behaviour of common ECC structures can be obtained.
where Cr denotes the sum of the circumferences of all steel bars; σe0 and
σr0 denote the stress of ECC and rebar at the cracking region. The ECC 4.2. Verification of the tension model
stress σe0 can be derived from the crack bridging relations in Eq. (2).
The constitutive law of reinforcement bars and uncracked ECC mate- 4.2.1. Verification of the crack bridging model
rials are defined as follows: For validation of the simplified four-parameter crack bridging
model, the analytical fibre–bridging models in Table 6 are employed.
e = e ( e) (7) All the previous analytical models are obtained directly from the lit-
eratures shown in Table 6. All the ECC crack bridging models are im-
r = r ( r) (8)
plemented in the tension model to predict the mechanical behaviour of
where all constitutive laws can be input through arbitrary stress–strain reinforced ECC specimens. By comparing the simulation results from
curves. In this study, the response of ECC before cracking is assumed to the simplified model and the previous analytical models, the accuracy
be elastic and the reinforcement constitutive law employs the measured of the proposed model is evaluated. A reinforced ECC specimen with a
stress–strain curve. 120 mm square section and a reinforcement ratio of 3.0% is adopted in
The analysis procedure for R/ECC members in uniaxial tension is the analysis. The yield strength of the rebar is 400 MPa. The simulation
shown in Fig. 17. First of all, a crack width and an average crack spa- results are shown in Fig. 18. Despite various expressions in all the
cing are selected and input into the analysis program. Then, the brid- analytical models, the simplified crack bridging model has sufficient
ging stress σe0 and the regional division of the R/ECC model can be accuracy in structural simulation.
9
J. Fan, et al. Engineering Structures 210 (2020) 110309
Fig. 17. Analysis procedure for the proposed R/ECC tension model.
4.2.2. Direct tension tests In simulation of Kang’s tests, σc, σu, ωu, and ω0 equal to 1.5 MPa,
For validation of the proposed R/ECC tension model, several R/ECC 3.0 MPa, 0.2 mm, and 1.0 mm, respectively. The corresponding para-
members tested in previous studies [23,25] were employed. In the meters in the simulation of Moreno’s tests are 2.7 MPa, 3.0 MPa,
tension behaviour analysis, the observed crack spacing and the mea- 0.2 mm, and 1.0 mm, respectively. The crack spacing equals 6 mm. As
sured reinforcement constitutive laws are employed. The cracking can be seen in the figure, the simulated responses by the proposed
stress and ultimate stress of ECC were measured in the original material tension model of R/ECC members show good agreement with the test
tests and the critical crack widths are chosen as the recommendations. results. It should be noted that the initial stiffness of several specimens
Fig. 19 shows comparison of the tensile behaviour of simulations and are slightly higher in simulations. The disparity is caused by the elastic
experimental tests. assumption of the uncracked ECC material. The difference increases
Kang et al. [25] conducted several tension tests to study the influ- with the increase of the cross section area. The observed crack widths of
ence of reinforcement ratio on the behaviour of R/ECC members. all specimens were less than 100 μm before yielding, which was con-
Moreno et al. [23] studied the tension stiffening effect of several re- sistent with the simulation results.
inforced HPFRCC members and those with ECC materials are employed.
Load P (kN)
Load P (kN)
Load P (kN)
Load P (kN)
Load P (kN)
Load P (kN)
Fig. 18. Verification of the simplified crack bridging model with analytical fibre–bridging models proposed by (a) Li, 1992 [7], (b) Maalej, 1995 [37], (c) Lin, 1997
[38], (d) Kanda, 1999 [39], (e) Lin, 1999 [29], (f) Yang, 2008 [30], (g) Huang, 2015 [31], and (h) Qiu, 2016 [40].
10
J. Fan, et al. Engineering Structures 210 (2020) 110309
90 90 120
ECC: 90 mm square ECC: 120 mm square ECC: 150 mm square
75 Rebar: φ13 75 Rebar: φ13 Rebar: φ13
Ratio: 1.64 % Ratio: 0.92 % 90 Ratio: 0.59 %
60 60
Load P (kN)
Load P (kN)
Load P (kN)
45 Steel bar
45 60
Test
30 Simulation 30
Steel bar 30 Steel bar
15 15 Test Test
Simulation Simulation
0 0 0
0.00 0.15 0.30 0.45 0.60 0.75 0.00 0.15 0.30 0.45 0.60 0.75 0.00 0.15 0.30 0.45 0.60 0.75
Strain ε (%) Strain ε (%) Strain ε (%)
(a) (b) (c)
150 200 150
ECC: 150 mm square ECC: 150 mm square ECC: 127 mm square
Rebar: φ20 Rebar: φ16
120 Rebar: φ16 160 120
Ratio: 0.89 % Ratio: 1.40 % Ratio: 1.25 %
Load P (kN)
Load P (kN)
Load P (kN)
90 120 90
60 80 60
Steel bar Steel bar #5 bar
30 Test 40 Test 30 Test
Simulation Simulation Simulation
0 0 0
0.00 0.15 0.30 0.45 0.60 0.75 0.00 0.15 0.30 0.45 0.60 0.75 0.0 0.1 0.2 0.3 0.4 0.5
Strain ε (%) Strain ε (%) Strain ε (%)
(d) (e) (f)
Fig. 19. Verification of the proposed R/ECC tension model with test results by (a)–(e) Kang et al. [25] and (f) Moreno et al. [23].
11
J. Fan, et al. Engineering Structures 210 (2020) 110309
Table 7
Deflection comparison of SEB-1 and SEB-2 between current method and R/ECC tension model.
Load (kN) SEB-1, 8Φ12, 2% SEB-2, 8Φ8, 1%
Test (mm) Current method Simulation Test (mm) Current method Simulation
0.3 Pp 3.36 5.14 53% 3.55 6% 2.91 5.66 94% 3.36 15%
0.4 Pp 4.88 6.86 41% 5.12 5% 4.38 7.54 72% 4.49 2%
0.5 Pp 6.79 8.57 26% 6.73 −1% 6.19 9.43 52% 5.77 −7%
0.6 Pp 8.55 10.31 21% 8.36 −2% 8.31 11.32 36% 7.38 −11%
0.7 Pp 10.97 12.06 10% 10.21 −7% 10.63 13.30 25% 9.96 −6%
80 Lf
= 1+ (1 )2
SEB-1 test result 0 2d f (14)
SEB-1 simulation
where β denotes the nondimensional hardening parameter; denotes
40 SEB-2 test result
the normalized crack opening; σ0 denotes the normalized critical
SEB-2 simulation
bridging stress determined by other fibre parameters. The crack brid-
SCB-1 test result
ging relations with different slip-hardening strengths are shown in
0 Fig. 23 (a). The corresponding load–displacement curves and crack
0.0 0.1 0.2 0.3 0.4 developments are shown in Fig. 23 (b) and (c), respectively. As can be
Crack width ω (mm) seen from the figure, the flexural strength and ductility increase with
Fig. 21. Comparison of measured and simulated load–crack width curves for the hardening factor β. As β increases from 0.0 to 5.0, the ultimate
negative loading specimens. flexural strength increases from 140 to 159 kN. The corresponding ul-
timate displacement increases from 23 to 35 mm. Steel–ECC specimens
with higher hardening factors tend to have higher strength and ducti-
diversification of ECC materials provides the possibility to achieve the
lity. The crack width is also relatively smaller compared to specimens
integrated design of materials and structures. The key is to obtain the
with lower β.
influence of different material properties on the performance and crack
Although chemical bonding between fibre and matrix could increase
resistance of ECC infrastructures. In this section, several material
the bridging stress, excessive chemical bonding can lead to fibre rupture
parameters are selected to study their effects on the mechanical beha-
and decrease of the ductility. There are several methods, such as oil
viour in steel–ECC composite beams. Different ECC fibre–bridging
coating [8], to control the bonding strength. Lin et al. [29] proposed the
models can be obtained with different material parameters. By repla-
maximum embedded length of a single fibre which guarantees full pull-
cing Eq. (16) with these fibre–bridging models, the mechanical per-
out without rupture, and the critical embedded length is calculated as
formance and crack propagation of ECC structures can be derived.
follows:
Specimen SEB-2 is employed in the parametric study. The effects of
fibre volume fraction, chemical bonding strength, and fibre slip-hard- lc = Lc0 e (f + f )
2 (15)
ening behaviour are studied.
Leung and Li [7] showed that the crack bridging relationship could where γ denotes the chemical bonding parameter; Lc0 denotes the
be predicted as follows: normalized critical embedded length; f denotes the snubbing coefficient
between fibre and matrix; f’ denotes the strength reduction coefficient
4Vf 2 L f 2 cos
( )= P ( ) p ( ) p (z ) dz d considering the effect of fibre alignment; lc and denote the critical
d f2 =0 z=0 (13) embedded length and the orientation angle of the single fibre, respec-
where Lf and df denote the length and diameter of the fibre, respec- tively. By integrating the tension contribution of all unbroken fibres,
tively; Vf denotes the fibre volume fraction; P ( ) denotes the fibre the crack bridging relationship can be obtained as shown in Fig. 24 (a).
pullout load; and z denote the orientation angle and centroidal dis- The corresponding load–displacement curves and crack developments
tance of fibres from the crack plane, respectively; p ( ) and p (z ) denote are shown in Fig. 24 (b) and (c), respectively. As can be seen from the
the corresponding probability-density functions, respectively. As shown figure, although the stiffness in steel–ECC composite beams increases
in the equation, the bridging stress is in linear relation with Vf. with high bonding strength, the corresponding ultimate displacement is
Therefore, for ECC with low fracture toughness, the crack bridging significantly reduced. As γ/Lc0 increases from 0.00 to 0.80, the secant
curves with different fibre volume fractions are shown in Fig. 22 (a). stiffness increases from 7.32 to 11.64 kN/mm at a load level of 80 kN.
The corresponding load–displacement curves and crack propagations However, the corresponding ultimate displacement drops from 30 to
are shown in Fig. 22 (b) and (c), respectively. As can be seen in the 22 mm. Beams with a high value of γ tend to have a smaller crack width
figure, the stiffening effect of ECC becomes stronger with the increase of before failure. However, the ductility is significantly reduced. A ba-
fibre content. As the fibre volume fraction increases from 0.33% to lanced performance between ductility and crack opening should be
2.00%, the secant stiffness of the composite beam increases from 7.03 achieved in material design.
to 9.23 kN/mm at a load level of 80 kN. The corresponding crack width
decreases from 0.07 to 0.04 mm. This result is consistent with experi- 4.3.2. Structural parameters
mental observations. Influences of more structural parameters are studied in this section
12
J. Fan, et al. Engineering Structures 210 (2020) 110309
Fig. 22. Influence of fibre volume fraction on (a) crack bridging relations of ECC material, (b) load–displacement curves and (c) crack propagations of steel–ECC
composite beams.
4 180 180
0.0 Steel-ECC beam
1.0 Ratio: 0.54 %
Bridging stress σ (MPa)
3 2.0
3.0 120 120
Load P (kN)
Load P (kN)
4.0 Rebar
2 5.0 0.0 0.0
1.0 Increase of fiber 1.0
60 2.0 60 slip-hardening 2.0
1 3.0 3.0
Increase of fiber
4.0 4.0
slip-hardening
5.0 5.0
0 0 0
0.0 0.2 0.4 0.6 0.8 0 10 20 30 40 0.0 0.1 0.2 0.3 0.4 0.5
Crack width ω (mm) Displacement δ (mm) Crack width ω (mm)
(a) (b) (c)
Fig. 23. Influence of fibre slip hardening on (a) crack bridging relations of ECC material, (b) load–displacement curves and (c) crack propagation of steel–ECC
composite beams.
including slab area, rebar strength, steel plate strength, slab re- load–displacement curves is relatively small.
inforcement ratio, top flange thickness, bottom flange thickness, web The influence of these parameters on crack propagation of
thickness, and crack spacing. Specimen SEB-2 is employed in the steel–ECC composite beams are shown in Fig. 26. As can be seen in the
parametric study. The analysis results provide valid references for the figure, increase of ECC slab area can significantly reduce the crack
design and crack control of steel–ECC composite beams. width development before yield of the composite beams. Increase of
The influence of these parameters on load–displacement curves are slab reinforcement ratio, steel plate thickness and decreasing of crack
shown in Fig. 25. As can be seen in the figure, increase of ECC slab area spacing can also help reduce the crack width at initial loading stage.
and bottom flange thickness can significantly improve the initial stiff- The effect of rebar strength, steel strength, and web thickness is rela-
ness of the composite beams. Increase of steel plate thickness, steel tively small. After the composite beam yields, beams with higher flex-
strength, and reinforcement ratio can help improve the negative ulti- ural strengths demonstrate stronger crack width control capability.
mate flexural strength. Decrease of crack spacing can delay ECC strain Beams with different slab areas show the same crack width develop-
hardening process, which can improve the negative flexural strength to ment at this stage.
some extent. The effect of reinforcement strength on the
4 160 160
0.00 Steel-ECC beam
0.16 Ratio: 0.54 %
Bridging stress σ (MPa)
Load P (kN)
0.64 Rebar
2 0.80 80 0.00 80 0.00
0.16 0.16
0.32 0.32
1 40 0.48 40 0.48
0.64 Increase of chemical bond 0.64
Increase of chemical bond 0.80 0.80
0 0 0
0.0 0.1 0.2 0.3 0 10 20 30 40 0.0 0.1 0.2 0.3
Crack width ω (mm) Displacement δ (mm) Crack width ω (mm)
(a) (b) (c)
Fig. 24. Influence of chemical bonding on (a) post-debonding crack bridging relations of ECC material, (b) load–displacement curves and (c) crack propagation of
steel–ECC composite beams.
13
J. Fan, et al. Engineering Structures 210 (2020) 110309
Load P (kN)
Load P (kN)
Load P (kN)
Load P (kN)
350MPa→450MPa
40%
80 80 300 80 350 80 0.4%
ECC slab area 60%
320 370 0.6%
40%→140% 80%
340 390 0.8%
100%
40 40 360 40 410 40 Slab Reinforcement Ratio 1.0%
120%
380 430 0.4%→1.4% 1.2%
140%
400 450 1.4%
0 0 0 0
0 10 20 30 40 0 10 20 30 40 0 10 20 30 40 0 10 20 30 40
Displacement δ (mm) Displacement δ (mm) Displacement δ (mm) Displacement δ (mm)
(a) (b) (c) (d)
160 160 160 160
Basic top flange Basic bottom flange Basic web
Thickness: 8 mm Thickness: 8 mm thickness: 6 mm
120 120 120 120
Crack spacing
Load P (kN)
Load P (kN)
Load P (kN)
Load P (kN)
40%→140%
80 40% 80 40% 80 40% 80 40%
60% 60% 60% 60%
80% Bottom flange thickness 80% Web thickness 80% 80%
Top flange thickness
40 100% 40 40%→140% 100% 40 40%→140% 100% 40 100%
40%→140%
120% 120% 120% 120%
140% 140% 140% Basic crack spacing: 50mm 140%
0 0 0 0
0 10 20 30 40 0 10 20 30 40 0 10 20 30 40 0 10 20 30 40
Displacement δ (mm) Displacement δ (mm) Displacement δ (mm) Displacement δ (mm)
(e) (f) (g) (h)
Fig. 25. Influence of (a) slab area, (b) rebar strength, (c) steel plate strength, (d) slab reinforcement ratio, (e) top flange thickness, (f) bottom flange thickness, (g)
web thickness, and (h) crack spacing on the load–displacement curves of steel–ECC composite beams.
Load P (kN)
Load P (kN)
Load P (kN)
350MPa→450MPa
40% 300
80 80 80 350 80 0.4%
60% 320
370 Slab Reinforcement Ratio 0.6%
ECC slab area 80% 340
390 0.4%→1.4% 0.8%
40%→140% 100% 360
40 40 40 410 40 1.0%
120% 380
430 1.2%
140% 400
450 1.4%
0 0 0 0
0.0 0.1 0.2 0.3 0.0 0.1 0.2 0.3 0.0 0.1 0.2 0.3 0.0 0.1 0.2 0.3
Crack width ω (mm) Crack width ω (mm) Crack width ω (mm) Displacement δ (mm)
(a) (b) (c) (d)
160 160 160 160
Basic top flange Basic bottom flange Basic web
Thickness: 8 mm Thickness: 8 mm thickness: 6 mm
120 120 120 120
Load P (kN)
Load P (kN)
Load P (kN)
Load P (kN)
14
J. Fan, et al. Engineering Structures 210 (2020) 110309
R/ECC structures, more efforts should be made to experimentally study Declaration of Competing Interest
the crack spacing pattern of R/ECC members. Furthermore, experi-
mental study [42] showed that the shrinkage of concrete could affect The authors declare that they have no known competing financial
the tension behaviour of RC members. Thus more experimental and interests or personal relationships that could have appeared to influ-
theoretical studies should also be carried out to study the effect of ence the work reported in this paper.
shrinkage on ECC structures in the future.
Jiansheng Fan: Conceptualization, Methodology, Funding acqui- The authors gratefully acknowledge the financial support provided
sition. Shuangke Gou: Data curation, Software, Writing - original draft. by the National Science Fund of China (Grant No. 51890901 and
Ran Ding: Validation, Writing - review & editing, Supervision, Funding 51978378).
acquisition. Jun Zhang: Resources. Zhengjie Shi: Investigation.
Appendix A
The derivations of the theoretical solution are based on following assumptions. ECC is elastic before cracking. Ee represents the Young's modulus
of ECC.
For a certain crack width ω, according to crack bridging relationship of ECC, the nominal ECC stress at region C (σe0) is given by
2
c u 2 u u
2 ( u) + u for ( u) ( u
u 0 )( c u)
e0 = e1 = e( ) = 2
u( c u) 2 for ( u u
( 0) u) >
( u 2
0) ( c u) + u
2
u ( u 0 )( c u) (16)
After ECC cracks, the steel bar slips out of ECC. Assume that the steel bar slides the same distance from both sides of the crack. Then the slip
distance of the steel bar s is derived by
s=
2 (17)
According to the bond–slip behaviour for steel bar embedded in ECC, the average bond strength at region B is given by
1 2s1 for 2 s1
1 for s1 < 2 s2
= ( 2) =
2 +(1 2 )(s3 2) (s3 s2 ) for s2 < 2 s3
2 for s3 < 2 (18)
The bond strength at region C is given by
=0 (19)
The crack spacing Lm can be derived through defects and hardening curve analysis. The ECC stress distribution is shown in Fig. 16. ECC stress σe
(x) decreases as x increases because of the bond load τ, and the stress at the embedded end σe2 (x = 0) can be derived as
Lm
2 (x )·Cr ·n d r L m 2 r (L m )
e2 = e0 + dx = e0 + · = e( )+ ( 2)
0 Ae n dr2 2 dr
4 r (20)
where n denotes the number of reinforcement bars.
The average stress of ECC e is defined as the average value of ECC stress along Lm. e is given by
Lm
2 2 2 e1 + e2 Lm r (L m )2
¯e = e (x ) dx = e0· + · = e( )+ ( 2)
Lm 0 Lm 2 2 2 Lm dr (21)
The average ECC strain can be derived as
Lm Lm
2 2 2 2
e = e (x ) dx = e (x ) dx +
Lm 0 Lm 0 2 (22)
According to the elastic assumption, the average ECC strain can be rewritten as
Lm
2 e (x ) 2 e1 + e2 L m
¯e= Lm
+ Lm 0
2
Ee
dx = Lm
+ L m 2Ee 2
(L m ) r (L m )2
= + e( )+ ( 2)
Lm Ee Lm Ee L m dr (23)
According to the deformation compatibility of ECC and rebar, the average ECC strain is equal to the average reinforcement strain. The average
strain of rebar is given by
r = e = (24)
Assume the reinforcement stress at crack zone equals σr0. The stress distribution of reinforcement bar can be derived as:
15
J. Fan, et al. Engineering Structures 210 (2020) 110309
Lm Lm
2 (x )· Cr 2 (x )·n d r
r (x ) = r0 dx = r0 dx
x Ar x n dr 2
4 (25)
According to the shear stress distribution shown in Eq. (18) and Eq. (19), the stress distribution can be rewritten as
Lm
4 ( 2) Lm
r0 dr
x|x 2 for 0 x < 2
r (x ) =
Lm Lm
r0 for 2
x 2 (26)
Then the critical rebar stress σr1 and σr2 are derived as
r1 = r0 (27)
2(L m )
r2 = r0 ( 2)
dr (28)
The average rebar stress is given by
Lm
2 2 2 r1 + r2 Lm (L m )2
¯r = r (x ) dx = r0· + · = r0 ( 2)
Lm 0 Lm 2 2 2 Lm dr (29)
According to the constitutive law of the reinforcement bar and the stress distribution (Eq. (26)), the strain distribution of the rebar is given by
r = r ( r (x )) (30)
Then the average strain of the reinforcement bar can be derived as
Lm
2 2
r = r ( r (x )) dx
Lm 0 (31)
The critical rebar stress σr0 can be then derived through Eqs. (31) and (24). The solve process needs numerical iteration because the constitutive
law of the reinforcement uses an implicit expression. The stress and strain distribution of the reinforcement bar can be derived. Finally, the average
stress of the R/ECC specimen can be derived as:
Lm
2 2 e (x )· A e + r (x )· Ar ¯e + r ¯r
¯ = dx =
Lm 0 A e + Ar (1 + r) (32)
The average stress , the average strain , and the corresponding crack width ω are derived through equations above.
Furthermore, if the reinforcement is ideally elastoplastic with a Young's modulus of Er and a yield strength of fr, explicit expressions of the
solution can be derived. During elastic stage of the rebar, Eq. (31) can be rewritten as
Lm Lm
2 r (x ) 2 ¯r (L m )2
¯r = 2
dx = 2
r (x ) dx = = r0 ( 2)
Lm 0 Er Er L m 0 Er Er Er L m d r (33)
Using Eqs. (23), (24) and (33), the expression of the critical rebar stress σr0 and σr1 are given by
Er E (L ) r Er (L m )2 (L m )2
r1 = r0 = + r m e( )+ ( 2) + ( 2)
Lm Ee L m Ee L m d r Lm dr (34)
The rebar stress at the embedded end is given by
Er E (L ) r Er ( L m )2 Lm2 2
r2 = + r m e( )+ ( 2) ( 2)
Lm Ee L m Ee L m d r Lm dr (35)
The average rebar stress is given by
Er E (L ) r Er (L m )2
¯r = + r m e( )+ ( 2) = Er ¯
Lm Ee L m Ee L m dr (36)
And the average stress of the R/ECC specimen is given by
r Er r Er (Lm ) + Ee Lm r ( r Er + Ee )(Lm )2
= + e( )+ ( 2)
(1 + r ) Lm (1 + r ) Ee Lm (1 + r ) Ee Lm dr (37)
Finally, when the rebar at crack zone yields, i.e. r1 fr , the critical rebar stress σr0 and σr1 can be derived as
r1 = r0 = fr (38)
The rebar stress at the embedded end is given by
2(L m ) *
r2 = fr ( 2)
dr (39)
where ω* denotes the critical crack width when r1 = fr . ω* can be derived from
16
J. Fan, et al. Engineering Structures 210 (2020) 110309
( r Er + Ee )(L m * )2
Er * E (L *)
r1 = + r m e(
*) + ( * 2) = fr
Lm Ee L m Ee L m dr (40)
The average rebar stress is given by
(L m )2 *
¯r = fr ( 2)
Lm dr (41)
And the average stress of the R/ECC specimen is given by
e( )+ r fr
¯ =
1+ r (42)
Explicit expressions of the average stress , the average strain , and the corresponding crack width ω are derived through equations above.
17