Sunteți pe pagina 1din 15

Original Research

Advances in Structural Engineering

2019, Vol. 22(10) 2250–2263
Effect of matrix ductility on the Ó The Author(s) 2019
Article reuse guidelines:
compression behavior of
DOI: 10.1177/1369433219837388

steel-reinforced engineered

cementitious composite columns

Fang Yuan and Mengcheng Chen

Engineered cementitious composites have the characteristics of tensile strain hardening and multiple cracking. Substituting concrete
with engineered cementitious composites can effectively avoid the cracking and durability problems induced by concrete brittleness.
In this study, four steel-reinforced columns with various matrix types and load eccentricities were tested under eccentric compres-
sion. Test results indicated that steel-reinforced engineered cementitious composite columns exhibited a larger load-carrying capacity,
higher ductility, better crack control ability, and damage tolerance, compared to reinforced concrete columns. All columns finally failed
in compression as manifested by matrix crushing. However, the failure patterns between steel-reinforced engineered cementitious
composite columns and reinforced concrete columns were extremely different. Significant concrete spalling appeared in the reinforced
concrete columns, while no sign of engineered cementitious composite spalling was observed in the steel-reinforced engineered
cementitious composite columns owing to the fiber bridging effect of engineered cementitious composite. The maximum crack width
in reinforced concrete column increases almost linearly with the applied load and reaches 2 mm at most just prior to attaining the ulti-
mate strength, while the maximum crack width in the steel-reinforced engineered cementitious composite column first increases and
thereafter remains constant at approximately 60 mm with an increasing compression load. In addition to experimental work, a finite
element model was proposed to predict the load–deformation response of the steel-reinforced engineered cementitious composite
column. The prediction results are in close agreement with the test data. Finally, parametric studies were conducted to further illus-
trate the effects of matrix ductility and load eccentricity on the load–deformation curves, strain contour, and moment–load interac-
tion curves of the columns.

column, crack control, damage tolerance, ductility, eccentric compression, engineered cementitious composites, finite element, load-
carrying capacity

Introduction reduce the damage tolerance of these structure types

(Fischer and Li, 2002a, 2002b).
Concrete cracking owing to tensile brittleness causes Over the past several decades, a type of high-
severe durability problems in reinforced concrete (RC) performance fiber-reinforced cementitious composite,
structures, manifested by serious steel reinforcement known as engineered cementitious composites (ECCs)
corrosion in RC structures in offshore or chemical with tensile strain hardening behavior and multiple
treatment environments. Replacing the steel reinforce-
ment with noncorroding fiber-reinforced polymer
(FRP) can potentially solve the corrosion problem. Department of Civil Engineering and Architecture, East China Jiaotong
However, the extensive use of FRP reinforcement is University, Nanchang, China
limited owing to the two inherent drawbacks of FRP Corresponding authors:
materials, namely, the low elasticity modulus and the Fang Yuan, Department of Civil Engineering and Architecture, East China
lack of ductility (Vijay and GangaRao, 2001). In RC Jiaotong University, Nanchang 330013, China.
structures, concrete brittleness also causes other major Email:
problems in terms of concrete spalling, bond splitting, Mengcheng Chen, Department of Civil Engineering and Architecture,
and loss of composite action between the steel reinfor- East China Jiaotong University, Nanchang 330013, China.
cement and concrete, which weaken the ductility and Email:
Yuan and Chen 2251

cracking properties has been developed and applied in In this study, several R/ECC columns are tested
infrastructure engineering (Kim et al., 2004; Lepech under eccentric compression. The influence of matrix
and Li, 2009, 2010; Li, 2003; Li and Kanda, 1998; Li types and load eccentricities on the column’s ultimate
and Leung, 1992). ECC and concrete exhibit similar strength, ductility, and damage tolerance—among
ranges of tensile strength (4–6 MPa) and compressive other factors—are evaluated. Subsequently, a finite
strength (30–80 MPa) but deform in a totally different element model is proposed to predict the load versus
manner under tension. Conventional concrete fails in a deformation response of the R/ECC column. Finally,
brittle manner once the first crack occurs. However, in an extensive parametric study was conducted to fur-
an ECC member, after the first cracking, the stress still ther evaluate the effect of matrix ductility on the ulti-
increase with increasing deformation until final crack mate strength of R/ECC members.
localization occurs, accompanied by multiple cracks
along the tested member. Typically, mechanical soften-
ing only begins at a tensile strain of 3%–5%, with a Experimental program
crack spacing of 3–6 mm, and a crack width of approx- Specimen preparation
imately 60 mm (Zhang et al., 2011). In uniaxial com-
pression, the ECC strain at the peak stress is nearly Four steel-reinforced columns were tested under com-
twice that of concrete (Kesner et al., 2003; Zhou et al., pression in this work. All columns have the same
2014). Existing research results indicate that the cross-sections of breadth (b) 3 depth (d) = 200 mm
mechanical properties of ECC material in shear are 3 250 mm, and lengths (L) of 1200 mm. The column
similar to those in tension (Kanda et al., 1998). FRP- specimens in this experiment can be divided into two
reinforced ECC beams without transverse steel reinfor- series. Series I consists of two columns tested under a
cement demonstrate superior mechanical performance load eccentricity of 40 mm, designed to fail in matrix
to concrete beams with closely spaced stirrups, thus crushing without reinforcement yield in tension. Series
indicating that the elimination of shear reinforcement II includes two columns tested under a load eccentri-
is feasible when concrete is replaced with ECC (Li and city of 120 mm, designed to fail in matrix crushing
Wang, 2002). A previous study also indicated that beyond reinforcement yield in tension. The corre-
ECC deforms in a compatible manner with steel rein- sponding load–eccentricity ratios (e0/r) in Series I and
forcement, thereby resulting in decreased interfacial II are 0.32 and 0.96, respectively, where r is given by d/
bond stresses and in the elimination of bond splitting 2. Each series includes one RC column and one R/
cracks and cover splitting (Fischer and Li, 2002a). ECC column. For each specimen, four steel bars with
Maalej and Li (1995) pointed out that steel-reinforced the same diameters of 16 mm were used as longitudinal
ECC-concrete composite beams with an ECC layer on reinforcements. For all specimens, ribbed steel stirrups
the tension side exhibited higher flexural strength and with diameters of 8 mm and a spacing of 100 mm were
finer cracks prior to their final failure than correspond- used as transverse reinforcements. Table 1 provides the
ing RC beams. Recently, Chen et al. (2018) proposed details of each specimen. Different specimen designa-
the use of high-strength ECC for the flexural repair of tions are employed to distinguish these columns.
RC structures with significant steel corrosion. With Specimen designations starting with the letter ‘‘E’’
this novel technique, the flexural strength of the corro- denote ECC specimens, while those starting with the
sive beam was retrieved, and the splicing of additional letter ‘‘C’’ represent concrete specimens. The Arabic
reinforcements or the removal of a large amount of numerals following the first hyphen indicate the longi-
concrete could be avoided. Experimental observations tudinal reinforcement diameter. The listed Arabic
of the cyclic response of steel-reinforced ECC (R/ECC) numerals following the second hyphen are used to dis-
flexural members also demonstrated that the energy tinguish specimens with different load eccentricities.
dissipation capacity can be improved significantly and For example, the specimen ‘‘E–16–120’’ represents an
member integrity can be better maintained when con- R/ECC column with a longitudinal reinforcement dia-
crete is replaced with ECC (Canbolat et al., 2005; meter of 16 mm and a load eccentricity of 120 mm.
Fischer and Li, 2002b; Yuan et al., 2013). Obviously,
ECC can operate in conjunction with steel reinforce-
ment to improve ductility and durability for R/ECC
Material properties
flexural members, owing to their unique properties. An increased replacement volume (80%) of cement
Previous studies have focused on the flexural beha- with fly ash was employed in the ECC composition in
vior of R/ECC members, but few experimental studies order to improve the environmental sustainability. The
have investigated the eccentric compression behavior matrix compositions of the ECC and concrete are dis-
of R/ECC members. This gives rise to the need for the played in Table 2. The fiber is made of polyvinyl alco-
conduct of additional studies on this issue. hol (PVA) fiber. The specific performance indexes of
2252 Advances in Structural Engineering 22(10)

Table 1. Summary of specimen information.

Series Specimen ID e0 e0/r Longitudinal Shear Matrix fcu Peak load Deflection at
(mm) reinforcement reinforcement type (MPa) (kN) peak load (mm)
(mm) (mm)

I C–16–40 40 0.32 4f16 f8@100 Concrete 32.28 1040 4.08

E–16–40 40 0.32 4f16 f8@100 ECC 46.08 1410 3.69
II C–16–120 120 0.96 4f16 f8@100 Concrete 32.28 560 5.68
E–16–120 120 0.96 4f20 f8@100 ECC 46.08 738 5.68

ECC: engineered cementitious composite.

Table 2. Mixture proportions.

Matrix Cement Fly ash Sand Coarse Water High-range PVA fiber volume
designation aggregate water-reducing fraction, %
admixture, %

ECC 0.2 0.8 0.2 – 0.22 0.8 2.0

Concrete 1.0 – 1.5 2.5 0.35 0.3 –

ECC: engineered cementitious composite; PVA: polyvinyl alcohol.

Table 3. Specific performance indexes of PVA fiber.

Length (mm) Diameter (mm) Tensile strength (MPa) Elongation (%) Elastic modulus (GPa) Density (g/cm3)

12 39 1620 7 42.8 1.3

PVA fiber are shown in Table 3. The volume of the

fiber is 2%, and the fiber number per square meter is
about 1.40 3 109. To evaluate the ECC tensile ducti-
lity, tensile tests were carried out on specimens with
dimensions of 350 mm 3 50 mm 3 15 mm. A typi-
cal ECC tensile stress–strain curve is illustrated in
Figure 1. It was found that the tensile strength
exceeded 5 MPa and the ultimate tensile strain
approached 4%. A number of ECC and concrete
cubes with dimensions of 100 mm 3 100 mm 3 100
mm were also prepared and tested in axial compres-
sion at the same time as the column tests. The mea-
sured cube strengths (fcu) of the ECC and concrete
were 46.08 and 32.28 MPa, respectively. The mechani-
cal properties of the steel reinforcement are shown in
Table 4.

Loading configuration Figure 1. Typical tensile stress–strain relationship of ECC.

The specimens were tested under compression at dif-
ferent load eccentricities. The compression load was
applied by a testing machine with a 5000 kN capacity. 6 mm were machined on each steel cap so that the load
Plate hinges were arranged at both ends of the speci- eccentricity could be controlled precisely. The load
men, and steel caps were installed at both ends of each was applied through V-shaped edges connected to the
specimen prior to loading. Grooves with depths of corresponding groove of each specimen. Three
Yuan and Chen 2253

Table 4. Material properties of steel reinforcement. data were collected automatically by a data logger dur-
ing the tests.
Diameter Yield Ultimate Elasticity
(mm) strength strength modulus
fy (MPa) fsu (MPa) Es (GPa) Test results and discussions
8 359 525 195 Failure characteristics and crack patterns
12 534 620 203
16 527 621 201 Figure 3 illustrates the specimen failure modes. Matrix
20 506 607 211 crushing was observed for all tested specimens and
local crushing of the ECC or concrete occurred near
the top ends for most specimens owing to the effects of
the end conditions. All columns finally failed in matrix
displacement transducers were arranged uniformly
crushing, but the failure patterns of the R/ECC col-
along the specimens to monitor the column deflec-
umns were extremely different from those of the RC
tions. Two displacement transducers were installed at
columns. Significant concrete spalling occurred in the
the specimen end to measure longitudinal shortening
RC columns, while no sign of ECC spalling was
during the tests. Numerous strain gauges were observed in the R/ECC columns because of the ECC
attached to the longitudinal tensile bars at an equidi- fiber bridging effect. This means that the ECC can
stant spacing of 100 mm and on the concrete surface provide more efficient confinement to the longitudinal
at mid-height at a spacing of 50 mm (Figure 2). The reinforcement compared to concrete. As a result, steel
test setup and measurement configurations are dis- buckling can be delayed and the damage tolerance of
played in Figure 2. A graded loading program was the members can be improved.
adopted for the test, and a load interval equal to one- Figure 4 illustrates the crack patterns for all tested
fifteenth of the estimated load capacity was employed specimens. For specimens with a load eccentricity of
until the peak load was reached. Each load interval 40 mm, the cracks initially occurred at approximately
was maintained for approximately 2 min. The loading 30% of the maximum load. In contrast, initial cracks
was applied slowly and continually following the peak were observed at approximately 15% of the peak load
load, until final failure occurred. All of the measured for specimens with a load eccentricity of 120 mm.

Figure 2. Details of test columns and setup (units: mm): (a) specimen dimensions and (b) measurement arrangement.
2254 Advances in Structural Engineering 22(10)

Figure 3. Failure modes for each specimen: (a) C–16–40, (b) E–16–40, (c) C–16–120, and (d) E–16–120.

Figure 5. Maximum crack width versus mid-height deflection

curves prior to peak load.

cracks with a small crack spacing and width were

observed along the column span for the R/ECC speci-
men, while only less than 10 evident flexural cracks
with a large crack width were found in the case of the
Figure 4. Final crack patterns of tested specimens. RC specimens. The results indicate that substitution of
concrete with ECC in the column can significantly
diminish the crack width, resulting in improved dur-
With a larger load eccentricity, the applied moment ability of the member if repairs cannot be carried out
becomes larger and the matrix cracking strength is immediately following a severe accidental event.
thus more likely to be reached. The flexural cracks ini- Figure 5 illustrates the maximum crack width versus
tially appeared on the side far from the loading point mid-height deflection (f) curves of specimens prior to
and extended toward the opposite side at increasing the peak strength. For the RC column, the crack width
loads. Crack localization occurred beyond the peak increased almost linearly with the applied load and
load for the R/ECC columns with a load eccentricity reached a value of 2 mm at most at peak load. For the
of 120 mm, while it was not observed for ECC col- R/ECC column, the crack width first increased and
umns with a load eccentricity of 40 mm during the then remained constant at approximately 60 mm with
entire loading process. It was also found from the test increases in the applied load. The significant difference
observations that at the ultimate stage, dozens of tiny in the development of the crack width was derived
Yuan and Chen 2255

can be observed that the deflections increase rapidly at

increasing loads prior to the peak load (Nu). As
expected, the flexural stiffness decreased when the load
eccentricity increased. Subsequently, the compression
load decreased at increased deflections owing to crush-
ing of the concrete or ECC. The decreasing trend was
more pronounced for the RC columns than the R/
ECC columns. This is a result of the fact that the ECC
strain at the ultimate compressive strength is nearly
twice as high as that of concrete (Kesner et al., 2003;
Zhou et al., 2014), which benefits the ductility of flex-
ural members that fail in matrix crushing. It can also
be observed from Figure 6 that the ultimate strengths
of the R/ECC columns are higher than the corre-
sponding RC columns. Compared with the RC speci-
mens, the ultimate strength of the R/ECC specimens
Figure 6. Load versus mid-height deflection curves for increased by 35.6% when the load eccentricity was
specimens with different load eccentricities. 40 mm and by 31.8% when the load eccentricity was
120 mm. Owing to the higher measured compressive
strength of the ECC, it is difficult to determine
from the different cracking processes of the concrete and whether there are other attributing factors apart from
ECC. A tension-softening process occurred in conven- the higher compressive strength of the matrix that lead
tional concrete once its tensile strength was obtained, to the higher ultimate strength of the R/ECC columns.
followed by a rapid increase in crack width. However, in This will be discussed in detail in the following section
ECC materials, following initial cracking, the tensile on numerical simulations.
load continued to increase with strain hardening beha-
vior, accompanied by the formation of multiple cracks.
Each individual crack tended to open steadily up to a Strain analysis
certain crack width, and the increasing deformation Figure 7 illustrates the strain distributions along the
resulted in the formation of an additional crack. Based tensile longitudinal reinforcement at different load lev-
on the same cracking mechanism, ECC member crack- els prior to Nu for specimens C–16–120 and E–16–120
ing can reach a saturated state with small crack spacing, of series II. The longitudinal reinforcement rebar strain
until the localization of a random single crack occurs. fluctuation in the R/ECC column was less pronounced
than that in the corresponding RC column. For C–16–
120, once cracking occurred, the tensile force at the
Load versus deformation curves cracked section was almost entirely maintained by the
Figure 6 illustrates the compression load (N) versus steel reinforcement, thus leading to strain fluctuations
the mid-height deflection (f) curves of the specimens. It along the longitudinal reinforcement. In contrast, for

Figure 7. Strain distributions along tensile longitudinal reinforcement: (a) C–16–120 and (b) E–16–120.
2256 Advances in Structural Engineering 22(10)

(a) (b)

(a) (b)
Figure 8. Stress–strain relationships of (a) ECC under uniaxial tension, (b) ECC under uniaxial compression, (c) steel
reinforcement, and (d) concrete.

E–16–120, ECC still provided a tensile load at the software which cannot effectively define the perfor-
cracked section through the stress transfer of the PVA mance of ECC materials, ATENA can easily give the
fiber, thus resulting in a highly uniform distribution of performance definition of ECC materials. In addition,
the longitudinal reinforcement strain. At the same load the software has its own unique ‘‘professional’’ in
value, the strains along the longitudinal reinforcement simulating concrete cracking, failure, and yielding of
in the RC column were also significantly higher than steel bars. The typical stress–strain curves of ECC
those in the R/ECC column. For example, the maxi- obtained from uniaxial tension and compression tests
mum strains of the longitudinal reinforcement for C– are shown as dotted lines in Figure 8 (Yuan et al.,
16–120 at 0.54 Nu and 0.88 Nu were 1190 and 2149 me, 2017; Zhou et al., 2014). To simplify numerical model-
respectively, compared to 704 me for E–16–120 at the ing, the following assumptions were made for the con-
same load levels. These special characteristics of the R/ stitutive relationships of materials: (1) the stress–strain
ECC column lead to a significant reduction in the inter- relationship of ECC in tension and compression can
facial bond stresses that subsequently benefits its be, respectively, described by a bilinear (Figure 8(a))
eccentric compression response. and a polyline curve (Figure 8(b)), (2) the steel reinfor-
cement is described by a bilinear curve with strain
hardening (Figure 8(c)), (3) for concrete in tension, the
Finite element implementation stress increases linearly with strain up to the first
cracking followed by a linear descending part pointing
Stress–strain relationship
to the zero stress point (Figure 8(d)), and (4) for con-
Numerical simulations of this study were carried out crete in compression, the stress–strain relationship pro-
using the nonlinear finite element software ATENA. It posed by Hognestad et al. (1955) was adopted to
was developed by Czech Cervenka Consulting model the ascending part followed by a linear descend-
Company and is mainly used to solve RC structure ing part pointing toward the zero stress point, as indi-
problems. Compared to other finite element analysis cated in Figure 8(d).
Yuan and Chen 2257

(a) (b)

(c) (d)
Figure 9. Comparison of the predicted load versus mid-height deflection curves with measured results for specimens: (a) C–16–
40, (b) E–16–40, (c) C–16–120, and (d) E–16–120.

Three-dimensional finite element models were property parameters were the same as those described
adopted to simulate the specimens in the experiments. in section ‘‘Experimental program.’’Figure 9 shows the
In this analysis, the mesh size was selected to be 20 mm comparison of the predicted moment versus the deflec-
according to the convergence study. The brick solid ele- tion curves at the mid-height section with the mea-
ments were used for concrete and ECC. The truss ele- sured results. The total moment at the mid-height
ments were adopted for the steel reinforcement. The section is the sum of bending moments caused by the
spring elements were used to consider the bond beha- initial eccentricity and lateral deflection, expressed by
vior between the steel bars and the matrix (concrete or
ECC). The bond-slip model proposed by Wu and Zhao M m = N ðe0 + f Þ ð1Þ
(2012) was employed to model the steel–concrete inter-
It can be observed that the simulated initial stiffness
face, while the model proposed by Bandelt and
almost coincides with the measured stiffness for each
Billington (2016) was used to model the steel–ECC
specimen. The predicted ultimate strengths were
interface. The columns were loaded by displacement slightly smaller than the corresponding measured val-
control during the entire loading process. The ues, while the predicted deflections at the ultimate
‘‘Newton–Raphson’’ iterative procedure was selected as strengths were slightly larger than the measured deflec-
the solution method. Both displacement and residual tions. The values of the experimental and numerical
convergence criteria were adopted in the computation peak loads are given and compared quantitatively in
and the error tolerance was set to be 0.005. Table 5. The average ratio of the predicted to the mea-
sured ultimate moment was 0.97, and the average ratio
of the predicted to the measured deflection at peak
Verification of finite element model strength was 1.08. The larger simulated deflection was
To verify the validity of the proposed model, the tested attributed to the fact that the numerical simulation is
columns were selected for analysis. The material an ideal condition where the maximum deflection and
2258 Advances in Structural Engineering 22(10)

Table 5. Proposed material parameters of concrete and ECC in numerical simulations.

Specimen ID Measured ultimate Predicted ultimate Measured ultimate Predicted ultimate

compression load (kN) compression load (kN) moment (kN m) moment (kN m)

C–16–40 1040.0 1020.9 45.84 44.72

E–16–40 1410.0 1481.4 61.60 60.18
C–16–120 560.0 540.3 70.38 68.43
E–16–120 738.0 721.9 92.75 90.95

Table 6. Proposed material parameters of concrete and ECC in numerical simulations.

Concrete Compressive Compressive strain Tensile strength ft Cracking strain etc# Ultimate strain ecu#
strength fc at peak stress ec0#

230 MPa 20.002 3.5 0.00015 20.006

ECC Cracking strain etc Cracking stress stc Peak stress st0/sc0 Strain at peak stress et0/ec0 Ultimate strain etu/ecu

0.00021 3 MPa 4.5 MPa/–30 MPa 0.03/–0.004 0.045/–0.012

the most severe damage occur at the mid-height sec-

tion. By contrast, the measured maximum deflection
section may deviate from the mid-height section owing
to the geometric imperfections or loading location
errors, which results in a smaller mid-height deflection.
After the peak load point, the simulated curves deviate
from the measured curves to some extent. Increased
nonlinearity in terms of steel bar buckling and con-
crete crushing occurred in the softening stage, which
explains the simulating difficulty in this stage. In gen-
eral, a reasonable agreement is observed between the
test and numerical results.

Parametric studies
A parametric study was conducted in order to gain a
comprehensive understanding of the eccentric com- Figure 10. Comparison of moment–deflection responses
pression behavior of R/ECC columns. The control col- between RC and R/ECC columns at a load eccentricity of
umn had a length of L = 1600 mm, a cross-section of 180 mm.
b 3 d = 250 mm 3 250 mm, a concrete cover
thickness of 40 mm to the longitudinal reinforcement
center, and an initial load eccentricity of 180 mm. For Load versus deformation analysis
the control column, four steel bars with the same dia- Figure 10 shows the comparison of the mid-height sec-
meter of 16 mm were used as the longitudinal reinfor- tional moment–deflection curves of the RC and the R/
cement of the control column. Steel stirrups with ECC columns. It is observed that the moment capacity
diameters of 10 mm and spacings of 100 mm were of the R/ECC column is significantly larger than that
used as transverse reinforcements. The proposed mate- of the RC column with the same geometric dimension
rial parameters of the concrete and ECC are displayed and reinforcement layout. The peak moment of the R/
in Table 6. The elastic modulus, yield strength, ulti- ECC column is 95.68 kN m, which is 1.38 times that of
mate strength, and ultimate strain of the steel reinfor- the RC column (69.55 kN m). The larger load-carrying
cement, were assumed to equal 200 GPa, 460 MPa, capacity of the R/ECC column is mainly owing to the
610 MPa, and 0.1, respectively. ultra-high-tensile ductility characteristic of the ECC
Yuan and Chen 2259

Figure 11. Overall deformations and strain contours of RC and R/ECC columns at peak load: (a) principle tensile strain and (b)
principle compressive strain (deformation magnified 10 times).

material. Because the ultimate tensile strain of the ECC section. It can be observed from Figure 11(a) that at
material can be as high as 3%, the ECC in the tension peak load, the maximum tensile strain of concrete and
zone is typically in the tensile strain hardening stage at ECC are 3.472 3 1023 and 4.373 3 1023, respec-
the peak load and can still contribute to the tensile tively. The assumed strains at the peak stress of concrete
stress resultant. However, concrete has already cracked and ECC are 0.00015 and 0.03, respectively. This indi-
before the peak load is reached and the tensile capacity cates that cracks appear on the tensile side of the RC
of concrete is almost negligible. The participation of column at peak load, and the tension contribution of
ECC in the tensile stress resultant is equivalent to the concrete is negligible. However, the extreme tensile
increase in the longitudinal reinforcement ratio of the strain of R/ECC column is far from reaching the locali-
column members. The higher the reinforcement ratio zation strain of ECC, and ECC can thus provide stable
is, the larger the bending moment capacity is. The mid- tensile force that ultimately has a positive effect on the
height deflection at the peak load of the R/ECC col- load-carrying capacity of the R/ECC column. Figure
umn is approximately 1.64 times that of the RC col- 11(b) shows the principle compressive strains of the RC
umn, as shown in Figure 10. The ultimate strain of and R/ECC columns at peak load. The maximum prin-
ECC is much greater than that of concrete. As a result, ciple compressive strains of concrete and ECC are
the R/ECC column exhibits superior deformability 21.693 3 1023 and 23.615 3 1023, respectively. The
than the RC column when they fail by matrix crushing. maximum principal compressive strain of ECC in the R/
It is also worth mentioning that the initial stiffness of ECC column at peak load is more than two times that
the R/ECC column is very close to that of the RC col- of concrete in the RC column, thus resulting in signifi-
umn although the ECC elastic modulus is only half to cantly larger mid-height deflection of the R/ECC col-
two-fifth of that of concrete. This is mainly owing to umn compared to that of the RC column.
the excellent tensile ductility of ECC, which ensures Figure 12 shows the overall deformation and strain
that ECC always contributes to the tensile force of the contour of RC and R/ECC columns at the ultimate state.
R/ECC column. Herein, the ultimate state is defined as the point where
the applied load drops to 85% of the load-carrying capac-
ity. It can be found from Figure 12(a) that the maximum
Strain contour analysis principal tensile strains of concrete and ECC at the ulti-
Figure 11 shows the overall deformation and strain mate state are 2.090 3 1022 and 3.121 3 1022, respec-
contours of RC and R/ECC columns at peak load. tively, indicating that both the concrete and ECC at
The deformations of the columns are magnified 10 extreme tensile faces enter the tensile softening stage.
times. The loading line at the upper plate coincides However, the ECC close to the neutral axis still provides
with the constraint line at the lower plate at the verti- tensile resistance. It can be observed from Figure 12(b)
cal projection, thus resulting in the occurrence of the that the maximum principal compressive strains of con-
maximum lateral deformation at the mid-height crete and ECC at the ultimate state are 21.146 3 1022
2260 Advances in Structural Engineering 22(10)

Figure 12. Overall deformations and strain contours of RC and R/ECC columns at ultimate state: (a) principle tensile strain and (b)
principle compressive strain (deformation magnified 10 times).

Figure 13. Comparison of moment versus mid-height deflection curves between RC and R/ECC columns at various eccentricities:
(a) e0 = 45 mm, (b) e0 = 90 mm, (c) e0 = 135 mm, (d) e0 = 225 mm, (e) e0 = 270 mm and (f) e0 = 315 mm.

and 22.071 3 1022, respectively, which are evidently Effect of load eccentricities
larger than the strains of concrete and ECC at peak To study the influence of the load eccentricity (e0), a
stress. This indicates that the concrete and ECC enter the parametric study was carried out by changing the load
compression softening stage, and both the RC and R/
eccentricity e0 from 45 to 315 mm. Figure 13 shows
ECC columns fail by matrix crushing. The larger maxi-
the mid-height moment–deflection curves of the RC
mum principal compressive strain of ECC leads to larger
and R/ECC columns at each load–eccentricity level. It
mid-height curvature and thus greater deflection of the
can be observed that both of the peak moments and
R/ECC column at the ultimate state.
Yuan and Chen 2261

Figure 14. Comparison of load versus moment interaction curves between RC and R/ECC columns: (a) columns with steel bar
diameter of 16 mm and (b) columns with steel bar diameter of 20 mm.

the corresponding deflections of the R/ECC columns constant. The numerical load–moment interaction
are larger than those of the RC columns. To better curve is shown in Figure 14(b). The same variation
analyze the ultimate strength, the compression load trend can be observed. That is, the load-carrying
capacity and moment capacity at the mid-height sec- capacity of the R/ECC columns is very close to that of
tion at each load–eccentricity level are shown together the RC columns when load–eccentricity ratio is low,
in Figure 14, also known as the load–moment interac- while it is evidently larger than that of the RC columns
tion curve. It is clearly found from Figure 14(a) that when the load–eccentricity ratio is high.
the load–moment interaction curve of the RC column
is surrounded by that of the R/ECC column, thus indi-
cating that the replacement of concrete with ECC leads
to a larger load-carrying capacity of the column mem- The eccentric compression behaviors of R/ECC col-
ber under eccentric compression. However, the ulti- umns were systematically studied in this work. First,
mate strength improvement is greatly dependent on four columns with various matrix types and load
the load–eccentricity ratio. The strength improvement eccentricities were tested under eccentric compression.
is more pronounced for columns with higher load– The failure modes, crack patterns, and load versus
eccentricity ratios. The mechanism can be explained as deformation responses of R/ECC columns were ana-
follows. When the load–eccentricity ratio is low, the lyzed and compared with those of RC columns.
bending moment to the axial force ratio is relatively Subsequently, a finite element model was proposed
small and most of the section areas are therefore under and verified by the comparison of the simulation with
compression. The contribution of the ECC tensile the measured results. Based on the finite element
strength to the total load capacity is negligible. As the model, a systematic parametric study was carried out
load–eccentricity ratio increases, the bending moment to further evaluate the effects of matrix ductility on
to axial force ratio increases, and the tension area the strain contour distribution as well as the load–
becomes increasingly larger. In this situation, the moment interaction curves of column members. In
ultra-high-tensile ductility of the ECC causes it to play summary, the following conclusions can be drawn
an important role in the resultant tensile force. It is as from this study:
if the tensile reinforcement ratio of the R/ECC column
is significantly higher than that of the RC column. It is 1. The failure modes of all columns were compres-
well known that a higher tensile reinforcement ratio sion failures manifested by matrix crushing;
leads to a larger compression stress resultant and thus however, significant concrete spalling occurred
to a higher flexural strength. That is, the R/ECC col- in the RC columns, while no sign of ECC spal-
umn exhibits a superior mechanical performance com- ling was observed in the R/ECC columns owing
pared to the RC column when they are subjected to to the fiber bridging effect of ECC.
eccentric compression at an increased load–eccentricity 2. The maximum crack width in RC columns
ratio. To further verify this observation, an additional increased almost linearly with the applied load
case with longitudinal steel bar diameter of 20 mm and reached values as high as 2 mm prior to
was also simulated. The other parameters were kept the ultimate strength, while the maximum crack
2262 Advances in Structural Engineering 22(10)

width in the R/ECC column first increased and Hognestad E, Hanson NW and McHenry D (1955) Concrete
thereafter remained constant at approximately stress distribution in ultimate strength design. ACI Journal
60 mm as a function of increasing compression 52(12): 455–479.
loads. Kanda T, Watanabe S and Li VC (1998) Application of
3. The deformability of R/ECC columns is super- pseudo strain hardening cementitious composites to shear
ior to that of RC columns at each load– resistant structural elements. In: Proceedings of the 3rd
international conference on fracture mechanics of concrete
eccentricity level.
and concrete structures FRAMCOS-3, Freiburg, October,
4. Replacement of concrete with ECC leads to a
pp. 1477–1490. Available at:
larger load-carrying capacity of column mem- 2027.42/84675
bers under eccentric compression. The ultimate Kesner KE, Billington SL and Douglas KS (2003) Cyclic
strength improvement is greatly dependent on response of highly ductile fiber-reinforced cement-based
the load–eccentricity ratio. The strength composites. ACI Materials Journal 100(5): 381–390.
improvement is more pronounced for columns Kim YY, Fischer G and Li VC (2004) Performance of bridge
with higher load–eccentricity ratios. deck link slabs designed with ductile engineered cementi-
tious composite. ACI Structural Journal 101(6): 792–801.
Lepech MD and Li VC (2009) Application of ECC for
Declaration of Conflicting Interests bridge deck link slabs. Materials and Structures 42(9):
The author(s) declared no potential conflicts of interest with 1185–1195.
respect to the research, authorship, and/or publication of this Lepech MD and Li VC (2010) Sustainable pavement over-
article. lays using engineered cementitious composites. Interna-
tional Journal of Pavement Research and Technology 3(5):
Funding Li VC (2003) On engineered cementitious composites (ECC).
The author(s) disclosed receipt of the following financial sup- Journal of Advanced Concrete Technology 1(3): 215–230.
port for the research, authorship, and/or publication of this Li VC and Kanda T (1998) Innovations forum: engineered
article: The work described in this study was fully supported cementitious composites for structural applications. Jour-
by a grant from the National Natural Science Foundation of nal of Materials in Civil Engineering 10(2): 66–69.
China (Grant No.: 51608199). Li VC and Leung CK (1992) Steady-state and multiple crack-
ing of short random fiber composites. Journal of Engineer-
ing Mechanics 118(11): 2246–2264.
ORCID iD Li VC and Wang S (2002) Flexural behaviors of glass fiber-
Fang Yuan reinforced polymer (GFRP) reinforced engineered cementi-
tious composite beams. ACI Materials Journal 99(1): 11–21.
Maalej M and Li VC (1995) Introduction of strain-hardening
References engineered cementitious composites in design of rein-
Bandelt MJ and Billington SL (2016) Bond behavior of steel forced concrete flexural members for improved durability.
reinforcement in high-performance fiber-reinforced ACI Structural Journal 92(2): 167–176.
cementitious composite flexural members. Materials and Vijay PV and GangaRao HVS (2001) Bending behavior and
Structures 49(1–2): 71–86. deformability of glass fiber-reinforced polymer reinforced
Canbolat BA, Parra-Montesinos GJ and Wight JK (2005) concrete members. ACI Structural Journal 98(6): 834–842.
Experimental study on the seismic behavior of high- Wu YF and Zhao XM (2012) Unified bond stress–slip model
performance fiber reinforced cement composite coupling for reinforced concrete. Journal of Structural Engineering
beams. ACI Structural Journal 102(1): 159–166. 139(11): 1951–1962.
Chen Y, Yu J and Leung CK (2018) Use of high strength Yuan F, Pan J and Leung CK (2017) Elastoplastic time his-
strain-hardening cementitious composites for flexural tory analysis of reinforced engineered cementitious com-
repair of concrete structures with significant steel cor- posite or engineered cementitious composite–concrete
rosion. Construction and Building Materials 167: composite frame under earthquake action. Advances in
325–337. Structural Engineering 20(4): 491–503.
Fischer G and Li VC (2002a) Influence of matrix ductility on Yuan F, Pan J, Dong L, et al. (2013) Mechanical behaviors
tension-stiffening behavior of steel reinforced engineered of steel reinforced ECC or ECC/concrete composite
cementitious composites. ACI Structural Journal 99(1): beams under reversed cyclic loading. Journal of Materials
104–111. in Civil Engineering 26(8): 04014047.
Fischer G and Li VC (2002b) Effect of matrix ductility on Zhang J, Leung CK and Gao Y (2011) Simulation of crack
deformation behavior of steel reinforced ECC flexural propagation of fiber reinforced cementitious composite
members under reversed cyclic loading condition. ACI under direct tension. Engineering Fracture Mechanics
Structural Journal 99(6): 781–790. 78(12): 2439–2454.
Yuan and Chen 2263

Zhou J, Pan J and Leung CK (2014) Mechanical behavior of Mu ultimate bending moment of column
fiber-reinforced engineered cementitious composites in N eccentric load
uniaxial compression. Journal of Materials in Civil Engi- Nu ultimate eccentric load of column
neering 27(1): 04014111.
e strain
ec0 compressive strain of ECC at peak stress
Appendix 1 ec0# compressive strain of concrete at peak
Notation stress
ecu ultimate compressive strain of ECC
b cross-section breadth ecu# ultimate compressive strain of concrete
d cross-section depth esu ultimate strain of steel reinforcement
e0 load eccentricity etc# cracking strain of concrete
Es elastic modulus of steel reinforcement etc first cracking strain of ECC
f deflection at mid-height section etu ultimate tensile strain of ECC
fc compressive cylinder strength of concrete et0 tensile strain of concrete at peak stress
fcu compressive cube strength of concrete s stress
fsu ultimate strength of steel reinforcement sc0 compressive strength of ECC
ft tensile strength of concrete stc first cracking stress of ECC
fy yield strength of steel reinforcement st0 tensile strength of ECC
L column length
Mm moment at mid-height section
Copyright of Advances in Structural Engineering is the property of Sage Publications Inc. and
its content may not be copied or emailed to multiple sites or posted to a listserv without the
copyright holder's express written permission. However, users may print, download, or email
articles for individual use.