See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/323907889

Cast-in-place Headed Anchor Groups Under Shear: Experimental and

Numerical Modelling

Article · March 2018

DOI: 10.1016/j.istruc.2018.03.008

CITATIONS READS

0 161

5 authors, including:

Trung Bui Ali Limam

Institut National des Sciences Appliquées de Lyon Institut National des Sciences Appliquées de Lyon
25 PUBLICATIONS   122 CITATIONS    153 PUBLICATIONS   984 CITATIONS   

SEE PROFILE SEE PROFILE

Wendpanga Serge Auguste Nana

Institut National des Sciences Appliquées de Lyon
7 PUBLICATIONS   11 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Generating Guided Waves for Detection of Transverse Type-Defects in Rails View project

Rockfall Protection Structures View project

All content following this page was uploaded by Trung Bui on 23 March 2018.

The user has requested enhancement of the downloaded file.

 Structures 14 (2018) 178–196

Contents lists available at ScienceDirect

Structures
journal homepage: www.elsevier.com/locate/structures

Cast-in-place Headed Anchor Groups Under Shear: Experimental and T

Numerical Modelling
⁎ ⁎ ⁎
T.T. Buia, , A. Limamb, , W.S.A. Nanaa, , B. Arrietac, T. Rourec
a
University of Lyon, INSA Lyon, GEOMAS, France
b
University of Lyon, France
c
EDF-SEPTEN, France

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

Keywords: In nuclear power plants, equipment is usually anchored in concrete. Earthquake verifications consist not only in
Cast-in-place checking the structure as a whole, but also in maintaining equipment at the anchorages. This study analyzes the
Headed anchors shear behavior of cast-in-place headed anchors in reinforced concrete structures for nuclear power plant
Shear applications. A series of 16 tests of anchors fastened in large-scale reinforced concrete blocks
Experiments
(2250 m × 1850 m × 600 mm) was performed. The experimental program was conducted with quadruple fas-
Reinforced concrete
tenings (250 × 250 × 25 mm). The anchors were fastened in reinforced concrete blocks (C40/50 concrete
grade) by four-headed anchor rods made up of steel smooth bars (S235). The tests were performed by applying a
shear force on the anchor plate until failure, occurring in the anchor rods or the reinforced concrete block.
Experimental tests were conducted to quantify the shear strength of the anchor and the associated failure modes.
We addressed the influence of several variables, such as cracks in the concrete, edge effect, effective embedment
depth, and the influence of cyclic shear loading, which has rarely been addressed in the literature. Then the
experiments were used to evaluate the pertinence of the numerical model for simulating the cast-in-place headed
anchors under shear. The experimental database provided can be completed and improved by other researchers
so as to assess and extend the design models.

1. Introduction place anchorages, which are installed in structures before casting, and
post-installed anchorages, which are placed in hardened concrete [7,8].
Modern fastening technologies have become increasingly important In EDF (Electricité de France) nuclear power plants, cast-in-place an-
in the field of civil and structural engineering. Anchorage in concrete chorages with headed rods are usually used in the connections and are
constructions is used in many structures and has been extensively stu- among the anchor types investigated in this study. Nuclear power plant
died in Klingner et al. [1], Klingner and Mendonca [2], Fuchs et al. [3], equipment is usually anchored in concrete. Earthquake verification
Klingner [4], Eligehausen et al. [5] and FIB Bulletin 58 [6]. Concrete consists not only in checking the structure as a whole, but also in
anchors and headed studs are needed to connect structural steel maintaining equipment at the anchorages. Following the 2011 Fu-
members and the concrete member. Concrete anchor connections are a kushima accident, the accidental design loads were reassessed. The real
critical component of load transfer between steel and concrete members capacity of the anchors under static and dynamic loading must be
affecting structural performance. Failures of anchored connections can known in order to optimize the design and identify safety margins, as
occur under tension, shear, or combined tension-shear loading. The should the real behavior of the anchorages in different configurations in
anchorage systems' load-transfer mechanisms are typically identified as order to validate numerical models. Different observations of damage in
mechanical interlock, friction or bond, and the tensile and shear ca- previous earthquakes had already raised concerns about the seismic
pacities of these fastening systems are based on various geometrical and performance of the anchor connections. We can mention among others
material factors such as concrete strength, steel strength, number of the 1995 earthquake in Kobe, Japan, causing the Osaka Gas Company
anchor rods, effective embedment depth of the rods, rod diameter, and to lose seven transformers, which collapsed due to anchorage failure
influence of an edge effect. A detailed discussion can be found in Eli- [9]; then the earthquake of Northridge, California, in 1994, caused the
gehausen et al. [5]. loss of power from three buildings of a veterans hospital in Los Angeles
In construction practice, there are two types of fastenings: cast-in- due to a damaged transformer caused by an unexpected failure of the

⁎
Corresponding authors.
E-mail addresses: [email protected] (T.T. Bui), [email protected] (A. Limam), [email protected] (W.S.A. Nana).

https://doi.org/10.1016/j.istruc.2018.03.008
Received 1 February 2018; Received in revised form 19 March 2018; Accepted 19 March 2018
Available online 21 March 2018
2352-0124/ © 2018 Institution of Structural Engineers. Published by Elsevier Ltd. All rights reserved.
T.T. Bui et al. Structures 14 (2018) 178–196

V
anchorage [10,11]. Note that the use of the anchors is only optimal V
when the design considers not only the load direction as tension, shear,
Concrete Steel
or a combination of tension and shear, but also the failure modes [5]. fracture
breakout
In the present study, a quasi-static and dynamic (rather cyclic so-
licitation) test campaign is proposed to study the different parameters
of these anchor plates under shear loading. The first experimental
a) b)
campaign of this research concerning cast-in-place headed anchors
under tension load was reported recently in Delhomme et al. [12]. As Fig. 2. a) Concrete pry-out failure modes for shear loading; b) steel failure mode for shear
mentioned, in practice the anchors can be subjected to both tensile loading.
forces and shear; furthermore, in some cases, shear may be the main
action [13]. This study presents the second part of the experimental
campaign of cast-in-place headed anchors but under shear loading.
characterized on unit elements inserted into an unreinforced or weakly
The shear strength of an anchor or an anchor group (anchors) is
reinforced concrete support. The increase in the size and number of
dependent on their failure mode. Possible failure modes associated with
rods on a platen causes operational problems making it difficult to
anchors subjected to shear loading are concrete edge failure [14],
perform tests, related to the increase of resistance capabilities requiring
concrete pry-out failure [15,16] and steel failure mode. A summary of
a larger test bench and heavily reinforced concrete supports so as not to
shear-transfer actions of anchorage in concrete construction can be
damage the concrete support itself. All these reasons explain why nu-
found in Eligehausen et al. [5]. Firstly, anchors loaded in shear toward
merical analysis can be an alternative to experimentally testing an-
a proximate free edge may fail because of the development of a semi-
chors, providing reasonably accurate results at a lower cost and in less
conical fracture surface in the concrete originating at the point of
time. Moreover, more anchor configurations can be evaluated in the
bearing and radiating to the free surface (Fig. 1-b1). This kind of failure
numerical analysis.
is known as concrete edge failure. It is the same for groups of anchors
In the present study, the experiments were also used to evaluate the
loaded in shear and proximate to an edge, which may develop a
pertinence of a numerical model for simulating the cast-in-place headed
common conical fracture surface (Fig. 1-b2), and the development of
anchors' behavior under the shear loading. Note that the numerical
the fracture surface can be interrupted by the presence of a corner
analyses were also applied to anchors under tension loading based on
(Fig. 1-b3) or by the limited depth of the member (Fig. 1-b4).
the previous research data reported in Delhomme et al. [12]. However,
Secondly, anchors and shear studs have limited embedment and
the complexity of the nonlinear finite element model in our study is
when loaded in shear can exhibit sufficient rotation to produce a pry-
primarily due to the fact that the numerical model must be able to
out fracture whereby the primary fracture surface develops “behind”
mimic all surface interactions between the anchor plate, the steel an-
the point of load application (Fig. 2a).
chor rods, and the concrete support material. In addition, the concrete
Finally, steel failure is a failure mode in which the steel rods in
material, i.e., the anchor plates' support material, whose behavior in-
anchor groups are in rupture (Fig. 2b). Anchors loaded in shear gen-
fluences the capacity of the anchors, is a very complex material. There
erally exhibit steel failure when the edge distance and the embedment
are many nonlinear concrete models based on different theories
depth are sufficiently large.
[21–25] and the results of numerical simulation strongly depend on the
Most studies reported in the literature concentrated on a single-
choice of the material model. Moreover, numerical analyses carried out
headed anchor more than on anchor groups [17]. For cast-in-place
to date to study the behavior of anchoring systems under tension or
fastenings, the experimental studies are usually conducted within a
shear loadings are relatively rare in the literature.
short embedment depth. In this study, the anchors are in groups con-
In this study, the concrete damaged plasticity (CDP) model in the
nected with four rods and fastened at a deep embedment depth. The
Abaqus finite element software was used. This concrete model was used
influence of the short rods on the anchor behavior was also addressed.
in Segle [19] to successfully simulate the effect of reinforcement on the
Then edge distance effects, the initial crack in concrete and cyclic
concrete breakout capacity of cast-in-headed bolts in concrete struc-
loading are carefully gauged. In particular, in this study, the concrete
tures based on the tests performed by Eligehausen et al. [26], Gross
slabs were heavily reinforced and are the same as the structures used in
et al. [27] and Nilforoush et al. [28]. Both single anchors and anchor
EDF nuclear power plants.
groups are studied for tension and shear loads. However, in their study,
The anchorage justification approach proposed in building codes
the steel failure mode was not investigated. Numerical analyses were
and design standards is directly inherited from the justification prin-
also performed by Nilforoush et al. [28] where the model was cali-
ciples gradually put in place over the past 3 decades through an ex-
brated against the test results of Nilsson et al. [29] for verification. The
perimental approach. This is based on an empirical representation of
study investigated how the member thickness, size of the anchor head
the behavior of the anchorage describing all the possible failure modes.
and surface reinforcement had an impact on the anchorage capacity for
Each failure mode associated with anchoring technology was tested and
single anchors under tension load. More recently Tsavdaridis et al. [30]
used to calibrate the empirical laws proposed in the design codes and
also used the CDP model to simulate single-headed anchors under
guide. However, the proposed analytical formulations may be in-
tension, leading to the derivation of a simple equation that describes
appropriate and this follows from their development for idealized and
the stiffness of the headed anchor rod embedded in concrete. In Ožbolt
limited anchoring configurations [14,18,19]. The anchor configurations
et al. [31], a 3D finite element (FE) analysis based on the microplane
concerned for example by FIB Bulletin 58 [6] and EOTA [20] are re-
model was performed to study the behavior of single-headed stud an-
latively small and the number of tests conducted on the different con-
chors with large embedment depths loaded in tension. Fornůsek and
figurations studied is rather limited. Most anchor behavior has been
Konvalinka [32] carried out a numerical analysis of headed studs with
different head sizes under tension and the results were compared to the
design results of anchoring in Eurocode 2. The fracture-plastic model
Cementititous2, which is included in the ATENA software, was chosen.
V V V In the present study, the numerical results show that the numerical
V
model can successfully simulate the experimental shear and tension
responses of the anchors in terms of shear and tension strengths and the
b1 b2 b3 b4
associated failure modes.
Fig. 1. Concrete edge failure modes for shear loading.

179
T.T. Bui et al. Structures 14 (2018) 178–196

Anchor rod
Anchor
Anchor plate
washer 250
20 20 25

Ø
20
100
20

250

250
100
60

25
hef 100 100 25

Cross-sectional view Plan view

Fig. 3. Anchor plate with four-headed smooth rods. Units: [mm].

2. Experimental program reinforcement used in the blocks is representative of EDF's nuclear

structures and a configuration of minimal reinforcement. It consists of
2.1. Test specimens one layer of ribbed bar (20 and 25 mm in diameter) in both directions
on the upper and lower parts of the block (Fig. 5). Stirrups (12 mm in
The experimental program was conducted with actual-size anchors diameter) are also installed only to maintain the two layers of re-
(Fig. 3). The anchor plate dimensions were 250 × 250 × 25 mm3. The inforcement. A concrete cover is located 50 mm from the longitudinal
plate was fastened by four anchor rods made up of steel smooth bar reinforcement. A concrete cover is also 50 mm in the direction of the
(S235) 20 mm in diameter with at their end an anchor washer (60 mm hook and perpendicular to the plane of the hook.
in diameter and 20 mm thick). The test program is summarized in Table 1. We addressed the in-
The tests were performed under pure shear loading (without a lever fluence of several variables, such as cracking in concrete, edge effect,
arm). As illustrated in Fig. 4, the shear load was applied with a hy- effective embedment depth and influence of cyclic loading. The ex-
draulic jack connected to a high-strength plate (rigid). This high- perimental program consisted of 16 tests of anchors fastened in the RC
strength plate was welded directly to the anchor plate (fixture). The blocks: two static tests (C1, C2) in concrete with no cracking, two static
anchor plate is embedded in the concrete; it is flush with the free sur- tests in concrete with an initial 1-mm crack opening (C11, C12), four
face of the concrete as described by the figure below. Therefore, the static tests with different effective embedments (C4N, C4S, C3N, C3S),
shear load acting on the anchors can be assumed to act without a lever four static tests with different edge distances (C4, C5, C6, C7), two
arm as described in “Annex C: Design methods for anchorages” in cyclic tests in concrete with no cracking (C13, C14), one cyclic test in
section 4.2.2.3 of EOTA [33]. concrete with cracking (C15), and one static test after the cyclic test
The anchors were cast-in-place in a reinforced concrete (RC) block (C15S).
measuring 2250 × 1850 × 600 mm3 (length × width × height), which
is large enough to avoid splitting failure and edge effects when they are
2.2. Test setup
positioned in the center. There is a 2-cm gap in front of the base plate
and the concrete in the shear direction to ensure that the concrete will
This bench was specifically designed for the project's shear tests [34,35].
not take up any of the applied shear load from the anchor. The
The loading was applied by a servo actuator with a maximum load capacity

Fig. 4. Shear load mechanism without lever arm used in the shear tests.

180
T.T. Bui et al. Structures 14 (2018) 178–196

1850
925 600
Ø25 / 400 Ø25 / 400

Ø20 / 400 Ø20 / 400

2250

1525
725

Ø25 / 400
50

Ø20 / 400
Stirrup Ø12
600

50

Ø20 / 400
Ø25 / 400

Fig. 5. Layout reinforcement. Units: [mm].

of 1000 kN and a displacement rate ± 100 mm. For the monotone case, length to avoid compression on the theoretical zone of the cracked
the loading was displacement-imposed with the speed of 1 mm/min. The concrete cone.
loading was transferred to the anchor plate by two high-strength bars For the reference tests (C1 and C2), only one test was conducted per
measuring 1850 × 100 × 30 mm (length × height × thickness). The con- block. The final rupture obtained for these reference tests was due to
nection between the high-strength bar and the actuator was left free in the rupture of the anchor rods but not the concrete, which remained
rotation. The reaction wall (a reaction frame) consisted of a set of metal intact in the block. Therefore, for the remaining tests, we suggest doing
HEB beams, anchored on the laboratory test slab by prestressed steel bars two tests per block, so two anchors were installed on the same block
(Fig. 6). (Fig. 8). Since there was no damage in the concrete after the anchor
For the boundary conditions of the monotone case, the horizontal rods were broken, the two tests on the same block were considered to
displacements of the RC slab are blocked by a metal bar positioned be independent. In other words, the first test did not cause any damage
along the length in the lower part of the block. This bar is blocked to the concrete block, which made it possible to carry out the second
horizontally by two supports anchored in the laboratory strong floor. To test under similar initial conditions (no damage). This reduces the
avoid rotation, the block is flanged vertically on the opposite side using number of blocks and more tests can be done. For all other tests with
a prestressed system at two points over the width with a force of 400 kN two anchors per block, the same boundary condition principle was
per point (Fig. 7). These two points are positioned over a short 20-cm applied.

Table 1
Experimental results (fcm, fctm measured on testing day).

Test Effective embedment Edge distanced Compressive strength Tensile strength Initial crack Loading Failure mode Ultimate load
hef (mm) (mm) fcm (MPa) fctm (MPa) opening w (mm) Vu,exp (kN)

C1 310 825 60.1 5.02 0 Static Steel rod failure 516

C2 310 825 60.1 5.02 0 Static Steel rod failure 549
C11 310 825 52.4 4.53 1 Static Steel rod failure 520
C12 310 825 52.4 4.53 1 Static Steel rod failure 495
C4N 50 825 58.9 4.61 0 Static Concrete pry-out 377
failure
C4S 50 825 58.9 4.61 0 Static Concrete pry-out 351
failure
C3N 100 825 59.6 4.64 0 Static Steel rod failure 469
C3S 100 825 59.6 4.64 0 Static Steel rod failure 489
C4 310 77 45.3 4.07 0 Static Concrete edge 210
failure
C5 310 135 45.3 4.07 0 Static Concrete edge 245
failure
C6 310 155 45.3 4.07 0 Static Concrete edge 285
failure
C7 310 232 45.3 4.07 0 Static Concrete edge 485
failure
C13 310 825 57.7 4.60 0 Cyclic Steel rod failure 374
C14 310 825 57.7 4.60 0 Cyclic Steel rod failure 371
C15 310 825 57.7 4.60 1 Cyclic Without failure 320
C15S 310 825 57.7 4.60 1 Static after Steel rod failure 431
cyclic

181
T.T. Bui et al. Structures 14 (2018) 178–196

Fig. 6. Test setup of static test: cross-sectional view. Units: [mm].

In the case of cyclic loading, the support system consisted of four prevent filling with concrete.
points of the prestressed system as well as two metal bars positioned ○ The reinforced concrete block is poured and left to dry for at least
along the length in the lower part of the slab on both sides, in order to 28 days.
block horizontal movements and the rotation of the block in two soli- ○ The metallic wedges (length, 550 mm; diameter straight length,
citation directions (Fig. 8). 55 mm; minimum diameter of the cone, 51 mm), after being
greased, are positioned in the steel pipes. These wedges are ham-
2.3. Crack in RC block mered with a hydraulic breaker (Fig. 10). The wedges, if possible,
are pressed over their entire height in three successive passes from
The objective was to create a crack over the entire thickness of the left to right. The wedges are loose and left in the massive concrete.
block that passes through two anchor rods. The method used is based on The crack opening created is approximately 1 mm, over the entire
the reference provided by Eligehausen and Balogh [36] and Elige- height of the block, along the defined crack line (Fig. 10).
hausen et al. [37].
The crack of the RC block was made using loose wedges mechani- 2.4. Instrumentation
cally hammered into the block, following the different steps of the
procedure: The applied force was measured using a sensor attached to the jack.
Its capacity was 1000 kN and the measurement error was 0.3% at full
○ Before casting, the 600-mm-long steel pipes (inner diameter, 53 mm; scale. The following displacements were measured (Fig. 11):
outer diameter, 60 mm), corresponding to the height of the block,
are positioned every 200 mm following the desired crack line ○ The horizontal movement of the jack with a displacement sensor
(Fig. 9). Both ends of the pipes are plugged with scotch tape to (linear variable differential transformer, LVDT);

125 200 200 125

Stirrup
515
125

High strength
bar
200

2250

A A
Section
1125
515
125

925
1850

(b)
Fig. 7. Test setup of the static test: a) plane view; b) picture of test setup. Units: [mm].

182
T.T. Bui et al. Structures 14 (2018) 178–196

1850
925

115
200 200
Prestressed
system

220
1220

2250
220
1525
725
Fig. 8. Test setup: cyclic test. Units: [mm].

○ The horizontal displacement of the anchor plate with two LVDTs The deformations of the headed anchor rods were measured using a
positioned at the end of the plate, on the side opposite the force, in strain gauge positioned in the middle of the four rods in order to obtain
front of the anchor rods (sensors H1 and H2); the anchor's shearing force (strain gauges J1, J2, J3, J4 for four rods)
○ The vertical displacement of the anchor plate using four LVDTs (Figs. 11, 14c). The strain gauges are KYOWA, 120 Ω nominal re-
positioned at four corners of the anchor plate (sensors V1–V4); sistance and 100 mm in length. A protection of the noncorrosive sili-
○ The relative horizontal displacement between the block and the cone rubber type was executed on each gauge. To catch strain dis-
laboratory slab with an LVDT positioned on the block (on the jack tribution along the height of the four rods, it is necessary to bond
side) (B2sensor); several strain gauges per rod, but this can affect rod adherence with the
○ The relative vertical movement between the block and the slab with surrounding concrete and finally can result in a decrease of the shear
an LVDT positioned on the block opposite the jack (B1sensor); bearing capacity of the anchor groups. Consequently, only one strain
○ The rotation angle of the anchor plate (relative to the block) was gauge was used for each anchor rod, glued in the middle of the rod to
measured with an inclinometer. ensure that deformation is always measured at this position. If the

Steel pipes

1850
925
600

200 200 Steel slotted

pipe
2250

Cracking
line
1525

Ø6
725

Fig. 9. Position of the steel slotted pipes before casting. Units: [mm].

183
T.T. Bui et al. Structures 14 (2018) 178–196

Fig. 10. a) The wedges are hammered by a hydraulic breaker; b) 1-mm crack opening.

1850
125 200 1850
Stirrup

232
Inclinometer
Strain gauge C7
125

LVDT
LVDT Inclinometer
Strain gauge
LVDT
LVDT
C6
200

V1 V4
V1 V4 H1
2250

H1
2250

B2 B1
B1

1918
H2
H2 C5 V2 V3
V2 V3
135 155

1125
1125

C4
125

77

Inclinometer
Inclinometer V1
V1 H1 V4
H1 V4
162,5

J1 J4
600
310

162,5
J1 J4
600

J0 B2
300

B1
B1
1850 1850

(a) (b)
Fig. 11. a) Reference tests (C1; C2); b) edge effect tests (C4; C5; C6; C7). Units: [mm].

strain gauge was bonded at the end of the rod, the deformation might Table 2. The maximum aggregate size was 22 mm for all specimens. The
be too small. Otherwise, if the strain gauge was bonded at the top of the concrete characterization tests used cylindrical specimens measuring
rod (near the shear load position), the strain gauge risks becoming 11 × 22 cm and were performed on the block testing day. To determine
unstuck due to the large deformation expected at this position. the concrete's mechanical characteristics, the specimens were poured at
the same time as the block was poured, with the same concrete. The
2.5. Materials specimens were placed in a humid chamber regulated in moisture and
temperature. The concrete block was covered with a plastic film for
The concrete used to design the blocks was a ready-mixed normal- 48 h and then demolded and covered again with a plastic film. The
strength class C40/50 concrete; the mixture proportions are given in room's temperature was equal to that of the room where the specimens

Table 2
Concrete mixture proportions.

CEM I Water (kg) Water reducer Sand 0/4 Gravel 4/ Gravel 11/22
52.5 N (kg) (% of cement) (kg) 11 (kg) (kg)

C40/50 350 147 0.4 872 441 591

184
T.T. Bui et al. Structures 14 (2018) 178–196

Table 3 The analysis of the experimental tests is presented in the following

Properties of steel rod S235. sections.
Type Yield Elastic Ultimate Elastic Ultimate
strength modulus strength strain strain (%) 3.1. Reference tests under static loading
(Mpa) (GPa) (MPa) (%)

S235 Rod 365 198 510 0.18 20 First, two reference tests (C1, C2) with the large embedment depth
(hef = 310 mm), anchored in the center of the RC block (far from the
edge, so the edge distance has no effect) under static loading were
S235 conducted. In particular, for these two reference tests, nine strain
600
gauges were bonded to the longitudinal bars (top reinforcement layer)
inside the block following the theoretical rupture line of the concrete
500
cone (Fig. 13). The objective was to measure the deformations of the
400 concrete at the level of the theoretical rupture cone. At the ultimate
Stress (MPa)

load, Fig. 13 shows very small deformations (maximum elongation,

300 60 με) for the gauges closest to the anchor plate. Since the maximum
deformation was less than the ultimate tensile strain of the concrete
200 (about 150 με), no cracking cone appeared in the concrete block.
The failure obtained from these reference tests was caused by the
100 failure of the anchor rods (Fig. 14) but not the concrete, which re-
mained intact in the block. Fig. 15 shows that bending in the rods at a
0
depth of 162 mm is relatively low, with a maximum compression de-
formation of 200 με, or about −42 MPa. From an applied load of
Strain (m/m)
400 kN, the four rods were solicited in tension. The maximum de-
Fig. 12. Stress/strain curve of the steel smooth bar S235. formation measured for the ultimate load of 516 kN for the C1 test
(549 kN for C2) was 1570 με and 1630 με, respectively, approximately
330 MPa and 342 MPa, respectively. The failure of the two rear anchor
are stored. Therefore, the concrete strength of the concrete cylinders
rods (in which two strain gauges J1 and J2 were bonded), where failure
and concrete blocks remained fairly close. The specimens were tested in
at first corresponded to the ultimate load obtained in the test, was by
compression and in tension to determine the mean values of the com-
shear at the head of the rod near the anchor plate. Two rods in front (in
pressive strength and the tensile strength (with a splitting test) of the
which two strain gauges J3 and J4 were bonded) failed later (also by
concrete. An overview of the concrete properties of the ten slabs tested
shear at the head of the rod near the anchor plate), but the shear effort
is given in Table 1.
was not increased.
The properties of the steel smooth bar S235 used in the anchor rods
The force/displacement curves obtained from two reference tests
are reported in Table 3 and Fig. 12. The high-adherence (HA) steel bars
are presented in Fig. 16. A similar behavior was observed for two C1
used to reinforce the slabs were Fe 500 B with guaranteed 500-MPa
and C2 reference tests, which confirms the relevance of the results
yield strength and greater than 540-MPa resistance at failure.
obtained. The behavior of headed anchor groups consisted of three
phases. The first phase was the elastic linear behavior with the small
3. Experimental results horizontal displacement observed. The rods were still in the elastic
phase. The two rods in front were in compression and the two rear rods
The failure mechanisms obtained from the experimental tests vary were in tension (Fig. 15). The second phase corresponds to the plastic
depending on the nature of the load applied (static or cyclic) and sev- behavior of the rods from the approximately 400-kN load up to the
eral factors such as specimen dimensions (effective embedment depth), ultimate load of 516 kN for the C1 test (549 kN for the C2 test). Four
fastening position (edge distance effect), and concrete characteristics rods were in tension. The third phase was after the post-peak in which
(with or without cracking). Generally, one of two types of failure can be the two rear rods were in failure (corresponding to points A and B in
observed: either ductile failure of the rod (steel rod failure) or brittle Fig. 16). Finally, the two rods in front were in failure, corresponding to
concrete breaking (concrete pry-out failure or concrete edge failure). point C of the force/displacement in Fig. 16. A small increase in the

Y 1125
Strain gauge
800 Test C1
Shear load direction

70
600 G5
)

G1 G9 60
Strain measured at Fmax (

400 G8
G2
200 G7 50
G3
G4
G6 40
1850

G4
G5 G6
30

20 G3
G1 G2 G7 G8 G9
10

X 0
0 250 500 750 1000 1250 1500 1750 2000 2250
2250 Coordinate X (mm)

Fig. 13. Strain measured at the ultimate load Fmax.

185
T.T. Bui et al. Structures 14 (2018) 178–196

J2 J3

162,5
J1 J4
J1 J4
Fig. 14. a) b) Failure mode of anchor rods for C1 test; c) strain gauges on four rods.

Fig. 15. Strain measured at rods: a) C1 test; b) C2 test.

Fig. 16. Force/displacement curve of static tests C1 and C2 in concrete with no cracking.
Fig. 17. Force/displacement curve of static tests in cracked concrete.

effort (at point C) was observed, stemming from the two rods in front
reaching failure. However, this effort was very small. concrete occurred sooner, at the force value of 350 kN instead of
400 kN. The ultimate shear strengths obtained in cracked concrete were
close to the shear strength in concrete with no cracks, with the values of
3.2. Static tests in cracked concrete
520 kN and 495 kN, respectively. The failure of the anchor rods oc-
curred at the ultimate strength (point A and B in Fig. 17).
The C11 and C12 tests were conducted in cracked concrete to
The technique used in Section 3.3 is to create a crack throughout the
evaluate the influence of the crack in the concrete on the shear behavior
thickness of the block, which passes through two anchor rods. The crack
of the headed-anchor groups. The force/displacement curves of two
opening created (outside the anchor plate) was approximately 1 mm,
tests in cracked concrete compared to two reference tests in concrete
over the entire height of the block, along the defined crack line.
without cracking (C1, C2) are illustrated in Fig. 17. Similar behavior
However, after failure when the anchor plate was taken off, we saw that
consisting of three phases was observed: the elastic phase, the plastic
the crack, which was smaller, passed through two anchor rods under
phase and the third phase after post-peak. In comparison with the tests
the anchor plate, with the opening crack measuring about 0.2 mm
with no cracks in the concrete, the same stiffness in the elastic phase
(Fig. 18).
was found. However, the beginning of the plastic phase in the cracked

186
T.T. Bui et al. Structures 14 (2018) 178–196

Fig. 18. Crack opening w = 0.2 mm observed under the anchor

plate after failure.

Fig. 19. Failure modes with different edge distances: a) 77 mm; b) 135 mm; c) 155 mm; d) 232 mm.

Fig. 21. The ultimate load versus the edge distance.

Fig. 20. Load/displacement curve of tests with different edge distances.

the anchor groups, four tests with four different edge distances (77 mm,
135 mm, 155 mm, 232 mm) were conducted. All four tests gave failure
Therefore, the results show that the presence of an initial crack in
modes by concrete edge failure with a common conical fracture surface
concrete did not influence the shear strength of the anchor groups. It
created (Fig. 19). The concrete conical fracture was developed from two
seems that the crack width for such geometry cannot have the influence
steel rods in front of the anchor groups. If the loading was continued,
on the peak load since the failure is due to steel.
two rear rods of the anchor groups ruptured. The edge effect creates a
clear influence on both the failure mode and the shear strength.
The load/displacement curves of the tests with different edge dis-
3.3. Static tests with different edge distances d
tances are presented in Fig. 20. There are two types of failure mode
observed in the experiments. The failure mode by concrete edge failure
To evaluate the influence of the edge effect on the shear behavior of

187
T.T. Bui et al. Structures 14 (2018) 178–196

Fig. 22. Concrete pry-out failure of C3N test with hef = 50 mm: a) concrete failure; b) deformation of anchor rods.

Fig. 23. Influence of embedment depth on: a) load/displacement curve; b) load/rotation angle.

Table 4
Loading cycles.

Vapplied/Vmax Number of cycles Vapplied (kN)

Uncracked C13, C14 Cracked C15

0.2 25 106 64
0.3 15 159 96
0.4 5 212 128
0.5 5 265 160
0.6 5 318 192
0.7 5 371 224
0.8 5 424 256
0.9 5 477 288
1.0 5 530 320
Total 75 100% 100%
Fig. 24. Ultimate load versus embedment depth.

Fig. 25. Schematic test procedure according to ETAG01.

188
T.T. Bui et al. Structures 14 (2018) 178–196

700 700
C1-Static C1-Static
C2-Static C2-Static
C14-Cyclic 500 C13-Cyclic 500

300 300
Force (kN)

Force (kN)
100 100

-8 -6 -4 -2 -100 0 2 4 6 8 10 12 -8 -6 -4 -2 -100 0 2 4 6 8 10 12

-300 -300

Displacement (mm) Displacement (mm)

Fig. 26. Cyclic tests – force/displacement curve in concrete without cracking: a) C14; b) C13.

Cyclic - C15 Cyclic - C15

400 400
300 300
200 200
Force (kN)

100
100
Force (kN)
0
0
-100 -3 -2 -1 -100 0 1 2 3 4
-200
-200
-300
-400 -300
0 50 100 150 200 250 -400

Fig. 27. Cyclic test (C15) in cracked concrete: a) loading history; b) force/displacement curves.

Vmax = 2.5d − 99.7 (d is edge distance in mm, Vmax in kN). Note that
this equation is valid only for the tested configuration. Furthermore, a
comparison with current design approaches and predictions, missing in
this study, should be conducted in a future investigation in which the
contribution of the reinforcement to the anchorage capacity can be
assessed by adding experimental tests on a concrete block without
rebar.

3.4. Static tests with the different embedment depths hef

In addition to two failure modes – steel rod and concrete edge shear
– presented in the previous sections, a third failure mode can occur:
concrete pry-out failure. Given that the anchor rods have limited em-
bedment and are loaded in shear, they can exhibit sufficient rotation to
Fig. 28. Effect of a cyclic test followed by a static test for cracked concrete. produce a pry-out rupture whereby the primary fracture surface de-
velops behind the point of load application. The tests presented in the
previous sections have a considerable 310-mm embedment depth.
was generated first and corresponds to point A in the load/displacement However, in this section four tests with two different embedment
curves (Fig. 20). The second failure mode was reached later by the depths were conducted: two tests with hef = 100 mm (C3N, C3S) and
rupture of two rear rods and corresponds to point B in the load/dis- two tests with hef = 50 mm (C4N, C4S). The concrete pry-out failure
placement curves. Generally, the ultimate loads obtained from the tests was obtained with a short 50-mm embedment depth (Fig. 22a). After
correspond to the failure mode by concrete edge failure and therefore failure, substantial plastic strain was observed in two rods in front of
correspond to point A in the load/displacement curves. Except for the the anchor groups, but they were not in rupture (Fig. 22b). For the
sole test C4 with the smallest edge distance of 77 mm, the ultimate load second embedment depth, 100 mm, the failure mechanism was steel
obtained corresponds to the failure mode of two rear rods. However, for rod failure.
this C4 test, the shear force corresponding to the concrete edge failure The load/displacement curves of these tests are illustrated in
(point A) was used for comparison with the design load. Fig. 23. The same rigidity was found for all tests in the elastic phase.
The ultimate load versus the edge distance is reported in Fig. 21. The test with the short 100-mm embedment depth gives a load/dis-
The tests show decreases in resistance as a function of the edge dis- placement curve very close to the large 310-mm embedment depth in
tance. A linear relation between them was determined for the tests with the plastic phase. Its influence can be clearly observed when the em-
concrete edge failure (except for the C4 test) by the following equation bedment depth was decreased, increasing the rotation angle of the

189
T.T. Bui et al. Structures 14 (2018) 178–196

2250
220 220
1125
RC
block

Anchor
925

plate
1850

Clamping
beam

Prestressed
bar
220 Tension load Anchor
rod
Prestressed
600

bar

Strong
2250 floor

Fig. 29. Test setup of tension tests. Units [mm].

Table 5
CDP model parameters and concrete fracture energies used for the numerical analyses.

Dilation angle Ψ Parameter Kc Eccentricity ∈ Ratio σb0/σc0 Energy Gf [N/mm] Energy Gc [N/mm] Mesh size [mm]

37° (calibrated) 0.6667 (per default) 0.1 (per default) 1.16 (per default) According to fib-Model code 1990 100Gf 9 mm (conv. study)

700
Test T2-310
600

500
Load (kN)

400

300

200

100 Experiment
FE model
0
0 10 20 30 40 50 60 70 80
Displacement (mm)

Fig. 30. Effect of a cyclic test followed by a static test for cracked concrete. Specimen T2-310 (a) experimental and numerical load/displacement responses; (b) failure mode.

anchor plate measured by the inclinometer transducer, as in Fig. 23b. damage progresses without interruption.
The tests show decreases in resistance as a function of the embedment Normally, the maximum shear load Vmax to be applied in the al-
depth (Fig. 24b). ternating shear load is calculated from the mean shear capacity of the
reference static tests (Vmean-NC = 530 kN and Vmean-WC = 507 kN for the
static tests in concrete without a crack and with a crack, respectively).
3.5. Tests under cyclic loading According to ETAG01, the maximum shear load Vmax can be calculated
by Vmax = 0.85 ∙ Vmean ∙ α with the reduction factor α = 0.85 [38].
The cyclic shear loading used comes from ETAG01 “Metal anchor Therefore, the maximum shear load obtained from two cases (un-
for use in concrete – Annex E: Assessment of metal anchors under cracked and cracked concrete) is Vmax-NC = 383 kN and Vmax-
seismic actions – Test C2.1” [33]. The sinusoidal loading cycles were
NC = 366 kN, respectively.
applied as in Fig. 25. The cycling frequency was set at 0.4 Hz. The data However, for the tests in concrete without cracking, to evaluate the
acquisition rate was 50 Hz. The actuator is controlled in force, so that

190
T.T. Bui et al. Structures 14 (2018) 178–196

Fig. 31. Specimen T3-80: boundary conditions.

decreased the anchor's shear capacity by approximately 20% in com-

parison with the static test without a cyclic effect.

4. Numerical results

4.1. Concrete and steel model

The numerical investigations were conducted using the FEM soft-

ware Abaqus. The concrete damaged plasticity (CDP) model in Abaqus
was used to model the behavior of concrete and the elastic-plastic
model was applied to reinforcement steel. A description of the basics of
the CDP model (the damage evolution, yield criterion, hardening law
and the flow rule), the uniaxial tensile and compressive behavior of
concrete coupled with the model can be found in Bui et al. [39] and
Nana et al. [40].
In the present study, the concrete compressive cylinder strength fcm
and the tensile strength fctm obtained from the experiments were used
in the numerical model while and the elastic stiffness Ecm was calcu-
lated using the empirical equation given by Eurocode 2 [41] as below:

Ecm = 9.5(f c + 8)1/3 (1)

For the concrete uniaxial behavior in tension, the post-peak ex-

Fig. 32. Specimen T3-80: experimental and numerical load/displacement responses.
ponential stress–crack opening displacement curve proposed by
Cornelissen et al. [42] was used and can be estimated using the Eq. (2).
The concrete behavior up to the ultimate tensile strength is elastic. The
real capacity of shear anchor groups under cyclic loading, the max- softening portion of the curve uses the fracture mechanics concept
imum shear load Vmax was chosen with a Vmean equal to 530 kN instead based on fracture energy Gf to avoid the mesh sensitivity of the analysis
of the reduced value. Only for the tests in concrete with cracking was [43].
the maximum shear load Vmax chosen based on the reduction value
3
proposed by ETAG01. The objective was to avoid severe damage during σ ⎛ w ⎞ ⎤ −c2 wwc − w (1 + c3)e−c2
=⎡
⎢1 + c1 wc ⎥ e
⎜ ⎟
1
the cycling so that the residual resistance of anchor groups could be f ctm ⎝ ⎠⎦ wc (2)
⎣
evaluated in the static test after the cyclic test. For the tests in concrete
with cracking, to ensure that the anchor rods did not fail during the c1 = 3; c2 = 6.93
cyclic test, the maximum shear load Vmax was taken at a value lower
than the reduction value calculated by ETAG01, so Vmax-NC = 320 kN Gf
wc = 5.14
(instead of 366 kN). The loading cycles of two configurations in cracked f ctm
and uncracked concrete are reported in Table 4.
For the tests in concrete without cracking using Vmax = Vmean, the wc is the crack opening (mm) for σ = 0, according to the Model Code
steel rods failed at the shear force equal to 70% Vmax (371 kN), corre- 1990 design guide [44] the tensile fracture energy Gf (N mm/mm2) is
sponding to 60 cycles applied. The tests were reproducible with the calculated as:
similar force/displacement curve observed in Fig. 26. f 0.7
For the tests in cracked concrete using a reduction of Vmax, as ex- Gf = (0.00005dmax 2 − 0.0005dmax + 0.026) × ⎛ cm ⎞
⎝ 10 ⎠ (3)
pected no anchoring failure occurred during the cyclic test (Fig. 27).
Therefore, a static loading was then conducted until the anchor failed. where dmax is the maximum aggregate size equal to 22 mm.
After the cyclic test in cracked concrete, the static test was conducted For the concrete uniaxial behavior in compression, the stress–strain
on the same anchor groups but by imposed displacement instead of law suggested by Feenstra and de Borst [45] was implemented. Its
controlled force as in cyclic loading. The effect of conducting a cyclic formulation is given by Eq. (4), where εu is the ultimate equivalent
test followed by a static test in cracked concrete is illustrated in Fig. 28. strain, εe an equivalent strain for σ = fcm, h the element size and Gc the
The static test after the cyclic test reduced the initial stiffness and also concrete compressive fracture energy taken equal to 100 Gf.

191
T.T. Bui et al. Structures 14 (2018) 178–196

Fig. 33. Specimen T3-80: experimental and numerical failure modes.

introducing scalar damage variables: respectively, dt and dc, which can

take values from 0, representing the undamaged material, to 1, which
represents a totally damaged material associated with the total loss of
strength.
The stress–strain relationship of reinforcement steel was modeled by
a bilinear curve considering a simple elastic-plastic model. The anchor
rods' connector material is of great importance in the steel failure mode
test simulations. The uniaxial stress–strain relation of anchor rods was
modeled as elastic with Young's modulus equal to 198 GPa. Plastic
behavior was defined in a tabular form, including yield stress and
corresponding plastic strain, ultimate stress and corresponding ultimate
strain obtained from uniaxial tensile tests.

4.2. Using the FE model for anchors loaded in tension

Fig. 34. Specimen C1-310: boundary conditions.

Before studying shear loading, the FE model was used to model
tension loading with the experimental tests presented in the literature
[12].

σ=
⎪ 3 (
⎧ fcm 1 + 4 ε − 2 ε
εe ε2e
2
si ε < ε e) For anchors loaded in tension, specimens T3-80 (hef = 80 mm) and
T2-310 (hef = 310 mm) were chosen to calibrate the numerical model
⎨ ⎡ ε − εe 2

⎩ ⎣
( ⎦
)
⎪ f cm 1 − εu − εe ⎤ si ε e < ε < ε u
(4)
and evaluate its relevance, respectively, for the concrete cone failure
mode (T3-80) and the steel failure mode (T2-310).
In the tension tests, the dimensions of the block and the anchorage
4 fc
εe = system are the same as the dimensions of the shear tests (Fig. 29). The
3 Ec (5) anchor plate dimensions are 250 × 250 × 25 mm3. The anchors are
Gc 11 cast-in-place in a reinforced concrete block (2250 × 1850 × 600 mm3),
ε u = 1.5 − εe which is large enough to avoid splitting failure and edge effects
hf c 48 (6)
when they are positioned in the center. The concrete grade used is C40/
The CDP model in Abaqus is based on the models proposed by 50 with the measured concrete strengths of fcm = 55.5 MPa,
Lubliner et al. [46] for monotonic loading and was later further de- fctm = 4.57 MPa for the T2-310 specimen and fcm = 55.2 MPa,
veloped by Lee and Fenves Gregory [47] to consider dynamic and cyclic fctm = 4.55 MPa for the T3-80 specimen. The contribution of the re-
loadings. Therefore, it provides the ability to model the behavior of the inforcement of the blocks in the tension tests is exactly the same re-
reinforced concrete element subjected to both static and dynamic loads. inforcement used in the blocks of the shear tests described in Section
The model uses concepts from the isotropic damage approach with 2.2. The quasi-static tensile loading was displacement-controlled with a
isotropic tensile and compressive plasticity to represent the inelastic loading rate of 1 mm/min. The concrete block is fixed by means of two
behavior of concrete. In addition, the stiffness damage of concrete was clamping beams connected to the strong floor with four prestressed
considered in this model for both tension and compression behavior by steel bars (Fig. 29).

192
T.T. Bui et al. Structures 14 (2018) 178–196

Fig. 35. Specimen C1-310: (a) experimental and numerical load/displacement responses; (b) Failure mode.

The concrete block and the anchor rod are meshed by eight-noded user-specified ratio of the equi-biaxial compressive yield stress and the
hexahedral solid elements (C3D8R) with a reduced integration scheme initial uniaxial compressive yield stress. The experimental values of
to avoid the shear locking effect, and a two-noded linear beam element σb0/σc0 are between 1.10 and 1.16 [46]. Its value is set by default to
(B31) is used for modeling the rebars. The rebars were embedded in the 1.16 according to ABAQUS Version 6.12 [50], also based on the reliable
concrete, simulating a perfect bond between these two materials. The results of Kupfer et al. [51]. ε is the eccentricity that defines the rate at
bond between the anchor rod and the concrete is considered as “hard which the potential function approaches the asymptote. Parameter ec-
contact” in the normal direction, meaning that a node on one surface is centricity can be calculated as a ratio of tensile strength to compressive
constrained to not penetrate the other surface. The tangential behavior strength [52]. ABAQUS Version 6.12 [50] recommends the default
of the contact interaction was defined as a frictionless formulation. The value of eccentricity ε = 0.1.
mesh size is 9 mm, chosen through a mesh convergence study. This ψ is the dilation angle measured in the p − q plane as the inclina-
choice of element size was based on a number of simulation tests with tion angle of the plastic potential function. This parameter controls the
different element sizes that showed that using a 9-mm mesh size was amount of plastic volumetric strain developed during plastic shearing
required to sufficiently capture the response of anchors in terms of ul- and has a significant influence on the concrete structural response and
timate loads and visualization of the damage. needs to be calibrated. Best agreements with experimental responses for
The CDP model parameters and concrete fracture energies used in concrete were achieved for a dilatation angle chosen between 30° and
the simulation are shown in Table 5. The parameters Kc, eccentricity∈, 40° [52–56]. In our study a calibrated 37° dilation angle was retained,
and the σb0/σc0 ratio of the CDP model used in our study are the default which gives good agreement with the experimental result.
values recommended while the dilation angle Ψ was calibrated. The According to the fib Model Code 1990 design guide, the fracture
concrete compressive cylinder strengths fcm and the tensile strength fctm energy is calculated as Gf = (0.00005dmax 2 − 0.0005dmax + 0.026) , where
obtained from the experiments were used in the numerical model while dmax is the maximum aggregate size. However, according to the current
and the elastic stiffness Ecm was calculated using the empirical equation fib Model Code 2010, the fracture energy is only dependent on the
given by Ecm = 9.5(fc + 8)1/3 [41]. In the calibration process, the ma- concrete compressive strengths fcm and calculated as Gf = 73(fcm)0.18.
terial parameters Kc, ∈ and σb0/σc0 of the CDP model were taken as the Physically, the aggregate size plays an important role in the fracture
default values given by the model because of their reliability. Kc is the mechanics of concrete [57]. Therefore, the fib Model Code 1990 with
ratio of the tensile strength to the compressive meridian and defines the the maximum aggregate size included was chosen in our model.
shape of the yield surface in the deviatory plane. Its value is about 0.64 For the T2-310 test, the comparison of the load/displacement curves
according to Schickert and Winkler [48] and 0.66 according to Richart between the experimental test and the FE model is shown in Fig. 30. It
et al. [49]. The factor per default in the CDP model [50] is Kc = 2/ is apparent that the FE model adopted correlates successfully with the
3 = 0.667, making the yield criterion approach Rankine's formulation. experimental test during both elastic and plastic stages. The figure
It was also found in the Nana et al. [40] study that using Kc = 1 as in shows the failure mode visualized through the maximum principal
the classic Drucker Prager criterion or Kc = 0.667 as in the CDP model plastic strains. The failure obtained by FE modeling for this test cor-
makes no significant difference in the model's behavior. σb0/σc0 is the responds to the steel failure, which corroborates the experimental

193
T.T. Bui et al. Structures 14 (2018) 178–196

Fig. 36. Specimen C4N-50: (a) experimental and numerical load/displacement responses and illustration of failure mode; (b) illustration of rod deformation after failure.

failure mode. The concrete block remains elastic. This result confirms smaller than the average diameter obtained from the 1-m experiment
that the numerical model proposed is able to reproduce the steel failure (inclined about 22° in relation to the cracks). The good quantitative and
mode of the anchors under tension load (Fig. 30). qualitative correlation obtained between the test and the numerical
The T3-80 test with hef = 80 mm is modeled on the same concrete simulation shows that the FE model is also fairly well validated for
geometry and the same boundary conditions as in the T2-310 test. Only concrete cone failure.
the effective embedment of the anchor rods changed, now with
hef = 80 mm. Due to symmetry, one-fourth of the specimen is modeled 4.3. Using FE model for anchors loaded in shear
and boundary conditions are symmetric in the XZ and YZ plane
(Fig. 31). The lower surface at the base of the concrete block and two For anchors loaded in shear, specimens C1-310 (hef = 310 mm),
sides of the block are restrained in the vertical direction (z-direction). C4N-50 (hef = 50 mm), and C4-edge-77 (d = 77 mm) were chosen to
It can be seen in Fig. 32 that the numerical model gives a load/ calibrate the numerical model and evaluate its relevance for steel rod
displacement curve very similar to the experimental curve up to the failure mode, concrete pry-out failure mode, and concrete edge failure
ultimate load. An ultimate numerical load of 305 kN is obtained, which mode, respectively. As in the case of the FE model for anchors loaded in
is very close to the experimental value of 303 kN. The numerical si- tension, the concrete block and the anchor rod are meshed as solid
mulation also gives a concrete cone failure mode, as in the experimental elements (C3D8R) and the rebars as beam elements (B31). A perfect
study. The four rods remain in the elastic behavior domain with a bond was assumed between the rebars and the concrete. The bond
maximum stress of 320 MPa, while the yield strength of the rod mate- between the anchor rod and the concrete is considered as “hard con-
rial is 365 MPa. tact” in the normal direction and as frictionless in the tangential di-
Fig. 33 presents the tension damage patterns obtained from the FE rection. The element size is taken again as 9 mm, fine enough to ensure
analysis (with a legend DAMAGET). The average diameter of the con- mesh convergence and a good description of the damage. Due to sym-
crete break-out surface obtained from FE analysis is about 0.8 m (in- metry, one-half of the specimen is modeled and boundary conditions
clined approximately 25° in relation to the cracks). This is slightly are symmetric in the YZ plane (Fig. 34). The lower surface at the base of

194
T.T. Bui et al. Structures 14 (2018) 178–196

(a)

(b)
Fig. 37. Specimen C1-310: (a) experimental and numerical load/displacement responses; (b) failure mode.

the concrete block and the downstream side of the block are restrained plants. The following conclusions can be drawn from this study:
in the vertical direction (Uz = 0). The foot edge in front of the block is
restrained in horizontal displacement (Uy = 0) (Fig. 34). • For the experimental part, it first quantifies the bearing capacity in
The comparison of the experimental results and the numerical cast-in-place headed anchor groups under shear in uncracked and
analysis in terms of the load/displacement response, as illustrated in cracked reinforced concrete slabs for static loading conditions
Figs. 35, 36, 37, shows that predictions of ultimate loads by the FE (monotonic and cycling) with variation of edge distance and em-
model are in good agreement with the experimental results. In addition, bedment depth, and highlights the associated modes of rupture. The
all the failure modes obtained numerically are identical to those ob- results show that the presence of an initial crack in concrete did not
tained in the tests. It is therefore apparent that the proposed model is influence the shear strength of the anchor groups.
able to reproduce the anchors' behavior under shear loads in terms of • Concerning the edge effect, the shear resistance decreases as a linear
ultimate loads and failure modes. function of the distance between the steel rod and the free edge
surface. With the small embedment depth used, a high rotation
angle of the anchor plate was observed, which can totally change
5. Conclusions and perspectives
the failure mode obtained.

This study contributes to a database of the experimental results of • For cyclic loading, the test results show that the monotonic quasi-
static test after the cyclic test leads to a 20% decrease of the shear
16 tests on real-scale anchors anchored in large RC blocks subjected to a
strength of the anchor groups in comparison with the quasi-static
shear load, with most particularly the influence of cracking in the
test without a cyclic effect.
concrete, edge effect, effective embedment depth, and the influence of
cyclic shear loading. The concrete slabs were heavily reinforced with • Experimental results under static loading were chosen to be com-
pared with the numerical simulation. It is apparent that the
rebar and correspond to the structures used in EDF nuclear power

195
T.T. Bui et al. Structures 14 (2018) 178–196

proposed model is able to reproduce the anchors' behavior under Business Media; 1986.
shear and tension loads in terms of ultimate loads and failure modes. [25] Simo JC, Ju JW. Strain- and stress-based continuum damage models—I.
Formulation. Int J Solids Struct 1987;23:821–40. http://dx.doi.org/10.1016/0020-
7683(87)90083-7.
However, the numerical model is limited only to monotonic loading. [26] Eligehausen R, Bouska P, Cervenka V, Pukl R. Size effect of the concrete cone failure
Cyclic loading was not experimented and therefore should be studied in load of anchor bolts. 1992.
[27] Gross JH, Klingner RE, Graves III HL. Dynamic behavior of single and double near-
the future with a more appropriate concrete model. The design code edge anchors loaded in shear. Struct J 2001;98:665–76. http://dx.doi.org/10.
predictions are important but were not addressed in this study. The 14359/10620.
main contribution of the paper is the extension of the existing experi- [28] Nilforoush R, Nilsson M, Elfgren L, Ožbolt J, Hofmann J, Eligehausen R. Influence of
surface reinforcement, member thickness and cracked concrete on tensile capacity
mental database. However, the experimental database provided can be of anchor bolts. ACI Struct J 2017;114.
completed and improved by other studies so as to assess and extend the [29] Nilsson M, Ohlsson U, Elfgren L. Effects of surface reinforcement on bearing ca-
design models and gauge numerical simulations. pacity of concrete with anchor bolts. Nord Concr Res 2011;2011:161–74.
[30] Tsavdaridis KD, Shaheen MA, Baniotopoulos C, Salem E. Analytical approach of
anchor rod stiffness and steel base plate calculation under tension. Structure
Acknowledgements 2016;5:207–18. http://dx.doi.org/10.1016/j.istruc.2015.11.001.
[31] Ožbolt J, Eligehausen R, Periškić G, Mayer U. 3D FE analysis of anchor bolts with
This work was funded by contract from EDF Septen (CQN02985) in large embedment depths. Eng Fract Mech Fract Concr Mater Struct
2007;74:168–78. http://dx.doi.org/10.1016/j.engfracmech.2006.01.019.
France. [32] Fornůsek J, Konvalinka P. Numerical investigation of head diameter influence on
tensile capacity of headed studs. 2012 IEEE symposium on business, engineering
References and industrial applications. 2012. p. 737–41. Presented at the 2012 IEEE sympo-
sium on business, engineering and industrial applications https://doi.org/10.1109/
ISBEIA.2012.6422988.
[1] Klingner RE, Mendonca JA, Malik JB. Effect of reinforcing details on the shear re- [33] EOTAeditor. ETAG 001—guideline for European technical approval of metal anchor
sistance of anchor bolts under reversed cyclic loading. J Proc 1982;79:3–12. for use in concrete—annex E: assessment of metal anchors under seismic actions.
[2] Klingner RE, Mendonca JA. Shear capacity of short anchor bolts and welded studs: a EOTA ed.Brussels: EOTA; 2012.
literature review. J Proc 1982;79:339–49. [34] Bui TT, Limam A. INSA-EDF 16001-Mode opératoire-Chargement cyclique en ci-
[3] Fuchs W, Eligehausen R, Breen JE. Concrete Capacity Design (CCD) approach for saillement. Internal report. 2016. p. 1–8.
fastening to concrete. Struct J 1995;92:73–94. http://dx.doi.org/10.14359/1533. [35] Bui TT, Limam A. INSA-EDF 130006-Mode opératoire-Essai statique de cisaillement
[4] Klingner RE. Anchor bolt behavior and strength during earthquakes. The simple sur des platines d'ancrage pré-scellées avec des tiges lisses à tête ou des
Commission; 1998. barres haute adhérence. Internal report. 2016. p. 1–14.
[5] Eligehausen R, Mallée R, Silva JF. Anchorage in concrete construction. Ernst & Sohn [36] Eligehausen R, Balogh T. Behavior of fasteners loaded in tension in cracked re-
Verlag für Architektur und technische Wissenschaften GmbH & Co. KG; 2012. inforced concrete. ACI Struct J 1995;92:365–79.
http://dx.doi.org/10.1002/9783433601358.ch4. [37] Eligehausen R, Mattis L, Wollmershauser R, Hoehler MS. Testing anchors in cracked
[6] FIB Bulletin 58. Design of anchorages in concrete: part I–V. Lausanne Int. Fed. concrete. Concr Int 2004;26:66–71.
Struct. Concr 2011. [38] EOTAeditor. Design of metal anchors for use in concrete under seismic actions.
[7] Mahrenholtz P, Eligehausen R. Post-installed concrete anchors in nuclear power EOTA ed.Brussels: EOTA; 2013.
plants: performance and qualification. Nucl Eng Des 2015;287:48–56. http://dx.doi. [39] Bui T-T, Limam A, Nana W-S-A, Ferrier E, Bost M, Bui Q-B. Evaluation of one-way
org/10.1016/j.nucengdes.2015.03.004. shear behaviour of reinforced concrete slabs: experimental and numerical analysis.
[8] Mahrenholtz P, Eligehausen R, Hutchinson TC, Hoehler MS. Behavior of post-in- Eur J Environ Civ Eng 2017;0:1–27. http://dx.doi.org/10.1080/19648189.2017.
stalled anchors tested by stepwise increasing cyclic load protocols. Struct J 1371646.
2016;113:997–1008. [40] Nana WSA, Bui TT, Limam A, Abouri S. Experimental and numerical modelling of
[9] Asia-Pacific Economic Cooperation. Earthquake disaster management of energy shear behaviour of full-scale RC slabs under concentrated loads. Structures
supply system of APEC member economies. Energy Comm. Minist. Econ. Aff. Taipei 2017;10:96–116. http://dx.doi.org/10.1016/j.istruc.2017.02.004.
China 2002. [41] CEN. Eurocode 2. Design of concrete structures - part 1-1: general rules and rules for
[10] E Grauvilardell J, Lee D, Hajjar J, Dexter RJ. Synthesis of design, testing and buildings. EN 1992-1-1. Bruss. Com. Eur. Norm CEN; 2005.
analysis research on steel column base plate connections in high-seismic zones. [42] Cornelissen HAW, Hordijk DA, Reinhardt HW. Experimental determination of crack
Struct. Eng. Rep. No ST-04-02 Univ. Minn. Minneap. MN 2005. softening characteristics of normal weight and lightweight concrete. Heron
[11] Schiff AJ. Northridge earthquake: lifeline performance and post-earthquake re- 1986;31:45–56.
sponse. U.S. Department of Commerce, Technology Administration, National [43] Hillerborg A. The theoretical basis of a method to determine the fracture energy GF
Institute of Standards and Technology, Building and Fire Research Laboratory; of concrete. Mater Struct 1985;18:291–6.
1997. [44] CEB-FIP Model Code 1990. Design code. Com. Euro-Int. Béton Thomas Telford Lond
[12] Delhomme F, Roure T, Arrieta B, Limam A. Static and cyclic pullout behavior of 1993.
cast-in-place headed and bonded anchors with large embedment depths in cracked [45] Feenstra PH, de Borst R. Constitutive model for reinforced concrete. J Eng Mech
concrete. Nucl Eng Des 2015;287:139–50. http://dx.doi.org/10.1016/j.nucengdes. 1995;121:587–95.
2015.03.012. [46] Lubliner J, Oliver J, Oller S, Oñate E. A plastic-damage model for concrete. Int J
[13] Ueda T, Kitipornchai S, Ling K. Experimental investigation of anchor bolts under Solids Struct 1989;25:299–326. http://dx.doi.org/10.1016/0020-7683(89)
shear. J Struct Eng 1990;116:910–21. http://dx.doi.org/10.1061/(ASCE)0733- 90050-4.
9445(1990)116:4(910). [47] Lee Jeeho, Fenves Gregory L. Plastic-damage model for cyclic loading of concrete
[14] Sharma A, Eligehausen R, Asmus J. Experimental investigation of concrete edge structures. J Eng Mech 1998;124:892–900. http://dx.doi.org/10.1061/(ASCE)
failure of multiple-row anchorages with supplementary reinforcement. Struct Concr 0733-9399(1998)124:8(892).
2017;18:153–63. http://dx.doi.org/10.1002/suco.201600015. [48] Schickert G, Winkler H. Versuchsergebnisse zur Festigkeit und Verformung von
[15] Anderson NS, Meinheit DF. Pryout capacity of cast-in headed stud anchors. PCI J Beton bei mehraxialer Druckbeanspruchung. 1977.
2005;50:90–112. [49] Richart FE, Brandtzaeg A, Brown RL. A study of the failure of concrete under
[16] Jebara K, Ožbolt J, Hofmann J. Pryout failure capacity of single headed stud an- combined compressive stresses. Engineering Experiment Station: University of
chors. Mater Struct 2016;49:1775–92. http://dx.doi.org/10.1617/s11527-015- Illinois at Urbana Champaign, College of Engineering; 1928.
0611-9. [50] ABAQUS version 6.12. 2013. (Documentation).
[17] Anderson NS, Meinheit DF. Design criteria for headed stud groups in shear: part 1 - [51] Kupfer H, Hilsdorf HK, Rusch H. Behavior of concrete under biaxial stresses.
steel capacity and back edge effects. PCI J 2000;45:46–75. Journal proceedings 1969. p. 656–66.
[18] Delhomme F, Roure T, Arrieta B, Limam A. Tensile behaviour of cast-in-place [52] Jankowiak T, Lodygowski T. Identification of parameters of concrete damage
headed anchors with different embedment depths. Eur J Environ Civ Eng plasticity constitutive model. Found Civ Environ Eng 2005;6:53–69.
2015;19:703–16. http://dx.doi.org/10.1080/19648189.2014.965850. [53] Genikomsou AS, Polak MA. Finite element analysis of punching shear of concrete
[19] Segle P. Numerical simulations of headed anchors break in reinforced and non- slabs using damaged plasticity model in ABAQUS. Eng Struct 2015;98:38–48.
reinforced concrete structures. Strålsäkerhetsmyndigheten; 2013. [54] Reissen K, Hegger J. Experimental investigations on the shear-bearing behaviour of
[20] EOTA. Guideline for European technical approval of metal anchors for use in bridge deck cantilever slabs under wheel loads. Beton-Stahlbetonbau
concrete, parts 1–6. Brussels: European Organisation for Technical Approvals 2013;108:315–24.
(EOTA). Eota; 2013. [55] Voyiadjis GZ, Taqieddin ZN. Elastic plastic and damage model for concrete mate-
[21] Chen W-F, Han DJ. Plasticity for structural engineers. J. Ross Publishing; 2007. rials: part I-theoretical formulation. Int J Struct Chang Solids 2009;1:31–59.
[22] Grassl P, Jirásek M. Damage-plastic model for concrete failure. Int J Solids Struct [56] Wu JY, Li J, Faria R. An energy release rate-based plastic-damage model for con-
2006;43:7166–96. http://dx.doi.org/10.1016/j.ijsolstr.2006.06.032. crete. Int J Solids Struct 2006;43:583–612.
[23] Hansen NR, Schreyer HL. A thermodynamically consistent framework for theories [57] Walraven JC. Fracture mechanics of concrete and its role in explaining structural
of elastoplasticity coupled with damage. Int J Solids Struct 1994;31:359–89. http:// behaviour. FRAMCOS. 2007. p. 1265–75.
dx.doi.org/10.1016/0020-7683(94)90112-0.
[24] Kachanov L. Introduction to continuum damage mechanics. Springer Science &

196

View publication stats