Steel bers as only reinforcement for at slab construction Experimental

investigation and design

Julien Michels
a,b,
, Danile Waldmann
a
, Stefan Maas
a
, Arno Zrbes
a,c
a
Faculty of Sciences, Technology and Communication, University of Luxembourg, 6, Rue Coudenhove-Kalergi, L-1359 Luxembourg City, Luxembourg
b
Structural Engineering Research Laboratory, Swiss Federal Laboratories for Materials Science and Technology Empa, 129 berlandstrasse, CH-8600 Dbendorf, Switzerland
c
University of Applied Sciences Bingen, Berlinstr. 109, D-55411 Bingen-am-Rhein, Germany
a r t i c l e i n f o
Article history:
Received 26 January 2011
Received in revised form 6 May 2011
Accepted 13 June 2011
Available online 13 July 2011
Keywords:
Steel ber reinforced concrete
Fiber orientation
Flat slabs
Experimental investigation
Slab design
a b s t r a c t
This paper presents the investigation on the bearing behavior of concrete at slabs with steel bers as
only reinforcement. In a rst step, experimental investigation on large-scale plates with symmetrical
loading around the column is presented. The results show an absence of punching shear failure and give
information on ber distribution and orientation in steel ber reinforced concrete (SFRC) elements with
growing thickness. In general, a decreasing ber orientation with an increasing plate height can be
noticed. Furthermore, a size effect with lower tensile strength at cracked state for large scale elements
in comparison to small beam specimens is shown. As bending failure occurred for all large-scale plate
specimens, yield line theory is applied for both results discussion and the development of a design con-
cept. Design abaci eventually show possible spans and slab thickness for satisfactory structural behavior
at ultimate load state.
2011 Elsevier Ltd. All rights reserved.
1. Introduction
Adding steel bers transforms the quasi-brittle concrete into a
ductile material. Whereas in this case the maximal tensile strength
is hardly enhanced (see [1]), the ber-knitting through the crack al-
lows stress transfer even at large crack openings. Amplitudes of the
residual strength enhancements mainly depend on the bers
geometry and dosage. Detailed information of the effect of the ber
presence in the cement matrix on residual strengthin exural bend-
ing tests on small beam specimens can be veried among others in
Barros and Sena Cruz [2] and Barros et al. [3]. It is shown that for one
specic ber type, higher ber dosage involves higher residual ex-
ural strength. Similar observations with specimens loaded in direct
tension are presented by Rossi [4]. Altun et al. [5] present compres-
siontests oncylinders withtwo different concrete mixtures andtwo
ber dosages (20 and 40 kg/m
3
). The presence of a ber reinforce-
ment involved a lower ultimate compressive strength compared
to the unreinforced specimens. However, ultimate strain increased
with growing ber content. In general, workability of the conglom-
erate represents an upper limit for ber dosages. Correlations be-
tween splitting tensile and exural strength and compressive
strength of ber reinforced concrete is found by Xu and Shi [6].
Over the last decades, steel bers in an amount of 2060 kg/m
3
were found highly useful as an alternative reinforcement in slabs
on ground-construction (refer to [7,8]). These research works pres-
ent experimental and numerical studies on such plate elements.
The results indicate a moderately higher bearing capacity but
clearly improved ductility of ber reinforced specimens compared
to plain concrete. Due to the benecial behavior of the material
against crack extension, the bers can act as a partial or even com-
plete replacement of the conventional reinforcement. Subse-
quently the same philosophy can be applied to at slabs, which
represent an easy technique in building construction. The simple
platecolumn connection and the resulting simple formwork
geometry allows a fast and cost-reducing plate erection. A com-
plete substitution of steel bar reinforcement with steel bers could
lead to further cost reduction due to lower necessary manpower.
Even if the material price of SFRC with high dosages is not neces-
sarily competitive against conventional reinforced concrete, the
necessary working time for the placement of reinforcing bars could
be partly omitted.
High ber dosages can lead to high bearing loads with simulta-
neously good rotational capacities. In 2004, a full-scale slab with
20 cm thickness and nine elds (3 3) with 6 m column-to-col-
umn span in both horizontal directions were carried out in Bissen
(Luxembourg). The used concrete mixture with a ber dosage of
100 kg/m
3
was identical to the one used for the actual project.
0950-0618/$ - see front matter 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.conbuildmat.2011.06.004

Corresponding author at: Structural Engineering Research Laboratory, Swiss

Federal Laboratories for Materials Science and Technology Empa, 129, ber-
landstrasse, CH-8600 Dbendorf, Switzerland. Tel.: +41 587654339; fax: +41
587654455.
E-mail address: [email protected] (J. Michels).
Construction and Building Materials 26 (2012) 145155
Contents lists available at ScienceDirect
Construction and Building Materials
j our nal homepage: www. el sevi er . com/ l ocat e/ conbui l dmat
Gossla [9] reports about satisfactory plate behavior at service load
state with no crack appearance at a central eld loading with 6 kN/
m
2
. At a load of 3.5 kN/m
2
in the edge eld small cracks appeared.
However, opening was inferior to 0.2 mm. A ductile material
behavior with yield line creation was observed when approaching
the ultimate load state. Point loading at midspan of the central
eld revealed a total bearing force of 462 kN, equivalent to a dis-
tributed load of 24 kN/m
2
including self-weight. In 2007, the same
material was used for another large-scale test in Tallinn (EST) with
a plate height of 18 cm and spans of 5 m (Destre [10]). Practical
applications of the specic steel ber concrete type, mainly for
industrial and housing construction, are shown in Pepin [11].
The aimof the research presented in this article is the analysis of
the bearing capacity of at slabs with only steel bers as reinforce-
ment under symmetrical loading around the column as well as the
effect different of slabthicknesses onthe bearingbehavior. First, fail-
ure mode and bearing capacity are assessed by performing loading
tests on plate cut-outs in the negative bending area around the sup-
port zone for inner columns. Simultaneously, information on the -
ber dispersion and orientation in large-scale steel ber concrete
elements withgrowing element height is collected. The obtainedre-
sults are subsequentlyanalyzedbymeans of yield-line theory. Final-
ly, a design suggestion for practical application of such steel ber
reinforcement is presented and the suitability of the presented rein-
forcement for larger practical applications is commented.
2. Experimental investigation
2.1. Materials
For all experimental investigations the same concrete composition was used.
Table 1 summarizes the proportions of each component. The undulated steel bers
had a length l
f
= 50 mm and a diameter d
f
= 1.3 mm (ber slenderness k = 50/
1.3 = 39). A photo of the bers is given in Fig. 1. All test specimens were casted with
the same ber dosage of 100 kg/m
3
, which is equivalent to 1.3% of the total volume.
Producers preliminary analysis on the suggested material revealed that a further
increase in vol.% did not lead to any signicant strength enhancement in relation
to the increased material costs and workability of the concrete. General impacts
of bers geometry and quantity can be found in Yazici et al. [12] and Altun et al.
[5]. Uniaxial tensile strength of the ber is 900 MPa, further information on charac-
teristics can be found in the technical data sheet presented by the producer [13].
Average concrete compressive strength on cylinder f
c,cyl
at the testing date is gi-
ven in Table 2. Preliminary 4-point bending tests according to the German Concrete
Association [14] revealed an average uniaxial tensile strength f
ct
of 2.5 MPa. Maxi-
mal exural tensile strength f
ct,
of 5.5 MPa (at maximal force value) and residual
exural tensile strength f
,res
of 3.7 MPa (at a tensile strain of 25) was measured.
2.2. Test campaign
2.2.1. Slab geometry and test setup
Octogonal specimens were casted for experimental analysis carried out in the
structural engineering laboratory at University of Luxembourg. Concrete was pro-
duced in the plant and later delivered to the structural testing hall. The steel bers
were manually added at the end of the mixing process. To avoid disturbing a homo-
geneous ber dispersion due to any kind compaction technique as presented in Get-
tu et al. [15], no vibration after or during casting was applied to the concrete.
Under symmetrical surface load around the column the test plate diameter for
representing the negative bending moment area can be estimated with 44% of the
total span. In case of punctual load symmetrically positioned around the support,
the diameter can be lowered to 39% (refer to [16]). A total span of 6 m was chosen
for the rst four plates. This results in a test specimen diameter of 2.34 m. The
square column side length is kept constant at 35 cm. Plate thickness is gradually in-
creased for the rst four plates from 20, over 25 to 30 cm. Furthermore, one spec-
imen is weakened by a square opening (35 35 cm) next to the column side. In
order to provoke higher shear loads in the column area, the last two specimens
present a slab thickness of 40 cm conjointly with a reduced diameter of 1.9 m (total
span of 5 m). Similarly, one specimen offers one square opening (25 25 cm) next
to the column. All geometrical parameters are summarized in Table 2. It is noted
that in at slab construction, span length is usually 7 m or higher. As in the present
investigation the failure mode is also of importance, span length was deliberately
reduced in order to enhance the probability of a punching shear failure.
The test setup is presented in Fig. 2. Eight hydraulic jacks symmetrically dis-
posed around the column support apply constant vertical displacement at the plate
border with a velocity of 0.2 mm/min. With the present test setup, no lateral dis-
placement prohibition as well as no moment and shear load redistribution is taken
into account. Additionally, only symmetrical loads are tested.
Nomenclature
a, b, c exponential relation parameters
d column side length
d
f
ber cross-sectional diameter
f vertical deection at the plate border
f
1
, f
2
concrete post-cracking uniaxial tensile strength at 0.1
and 25
f
c,cyl
concrete compressive strength on cylinder
f
ct
concrete uniaxial tensile strength
f
ct,
concrete exural tensile strength
f
res,
concrete residual exural tensile strength at 25
h plate height
l
f
ber length
l
pl
plastic length
m bending moment in the yield line
m
R
bearing moment in the yield line
m
Rd
design bearing moment in the yield lines
m
d
design bending moment in the yield lines
n
f
number of bers per unit area
r column radius
s offset from the column edge at the upper plate surface
w crack opening
x compression zone height
A
f
ber cross-section
D test plate diameter
F
max
maximal force
L span
P reaction force at the column
R plate radius
R

length of yield line

V
f
ber volume fraction
W work capacity/energy absorption
a
c
creep factor
e
1
, e
2
characteristic tensile strains (0.1 and 25)
e
cu
maximal concrete compressive strain
k ber slenderness
g
//
ber orientation factor in horizontal direction
c
R
safety factor for residual SFRC tensile strength
w rotation angle
v
1
, v
2
geometry factors related to the post-cracking strengths
f
1
and f
2
h hinge rotation
Table 1
Concrete mixture.
Sand 700 kg/m
3
Aggregate 4/8 500 kg/m
3
Aggregate 8/16 550 kg/m
3
Cement CEM I 42.5 R 350 kg/m
3
Fly ash 120 kg/m
3
Water/cement 0.5
Plasticizer 1.26%
Fiber content 100 kg/m
3
Air content 3.50%
146 J. Michels et al. / Construction and Building Materials 26 (2012) 145155
2.2.2. Forcerotation behavior and failure mode Results and discussion
Fig. 3 presents the plates total force evolution plotted against the angular rota-
tion, which is obtained by dividing the measured border deection by the plate ra-
dius (see Fig. 2). The evolution of the energy absorption W, equivalent to the area
under the force-deection curve, is also presented. Fig. 4 shows the crack pattern
at test end on the upper surface (negative bending stresses) for each test specimen.
In both gures, one can observe a ductile material behavior with the creation of
yield lines on the upper plate surface. Absence of a punching failure cone is noticed
for all test specimens. In addition to the previous observation, evolution of radial
compressive strain gives information about the failure mode. Three LVDTs A, B
and C installed at the bottom side measured horizontal displacement of the con-
crete at three locations in radial direction from the column to the plate border
(Fig. 5). Dividing the displacement value by the base length of the transducer, a
strain value is obtained. For compressive strains A (Fig. 5) a continuous increase
in main bearing direction with growing plate rotation can be observed. Further-
more, two LVDTs perpendicular to A and B record tangential displacement subse-
quently transformed into tangential strain. In case of a punching failure, a
decrease in the radial compressive strain next to the column at around 80% of the
total bearing load occurs, presented among others by Guandalini [17] and Beutel
[18]. Plate failure in bending is in accordance with weakly reinforced concrete
plates, where a punching failure is only possible with a sufciently high upper ten-
sile reinforcement. An increase in the linear elastic limit with growing plate thick-
ness is noted, from 120 kN for the 20 cm plate height up to 420 kN for the 40 cm
thick specimen. Furthermore, a higher bearing capacity is directly correlated to
the plate height the thicker the plate the higher the total bearing load. Total force
of 230 kN was observed for a slab height of 20 cm, whereas the thickest specimen
had a bearing capacity of 650 kN. A direct comparison between the full slab and the
weakened slab with an opening next to the column (PL1 and PL4 or PL5 and PL6,
respectively) shows a lower bearing capacity for the second specimen type.
The slender plates present, contrary to the more compact specimens, a less steep
softening behavior. For the same rotation angle, the crack opening for a thicker ele-
ment is higher and thus residual tensile strength lower than for a more slender spec-
imen. Furthermore, possible lower strength values due to a less pronounced ber
orientation for a thick plate specimen might affect the bearing capacity drop, too.
Hence, the decrease in force is faster for the more massive test specimens. The energy
absorptionrotationcurves also reect the giveninformationabout the differences in
the post-peak behavior. An almost linear evolution is observed for the rst two spec-
imens PL1-25-0 and PL-20-0, whereas growing plate height involves a less pro-
nounced curve angle after having passed the peak force.
2.2.3. Fiber dispersion and orientation Results and discussion
After the end of the test, two to three yield lines were analyzed more in detail
for each specimen with regard to the ber dispersion and orientation. The number
of bers N
f
were manually counted for the two surfaces of each yield line and sub-
sequently transformed into the number of bers per unit area n
f,//
and the equiva-
lent ber orientation factor in horizontal direction g
//
. The law proposed by
Krenchel [19], including the ber volume fraction V
f
and the ber cross-section
A
f
, has been applied:
g
==
n
f ==

A
f
V
f
1
Four layers over the specimen height and three rows in horizontal directions
(see Fig. 6) were xed for each yield line in order to assess the bers dispersion.
Table 3 and Fig. 7 give an overview on the obtained orientation factors.
In general, a relatively small variation of the horizontal orientation factor g
//
within one layer can be observed for all specimens and each layer. However, the
variation of the dispersion and thus of the orientation factor is much stronger over
the specimen height. Especially the rst two specimens PL1-25-0 and PL2-20-0
present high differences between different layers. Higher ber presence in the low-
er layers indicate a ber segregation while and after casting. For PL3-30-0 and PL-
25-1, the phenomenon is diminished but still observable. The last two specimens
PL5-40-0 and PL6-40-1, with a higher concrete consistency (F5 instead of F3 accord-
ing to the European design code for construction materials properties [20] at fresh
state) offer a stiffer matrix with lower superplasticizer content and thus prevent -
bers from sinking towards the formwork bottom due to gravity forces. As it can be
observed in Table 3 and Fig. 7, the differences between the orientation factors from
the rst to the fourth layer are much smaller for the last two plates.
By summarizing the different orientation factors for the different subdivisions
into one average value for each specimen height, one can derive a relation between
plate thickness and the obtained orientation factor. A decreasing ber orientation in
horizontal direction can be pointed out. Results from specimen PL3-30-0 seem to be
the outliers from the general tendency. Erdem [21] and Lin [22] suggest an expo-
nentially decreasing relation of the type g(h) = a e
bh
+ c in order to connect ber
orientation in horizontal direction to specimen geometry (height). Both dene an
asymptotic value for a complete three-dimensional orientation, subsequently
non-linear exponential regression is performed on the experimental results. In gen-
eral, a direct comparison between the new curve and the relations of Erdem [21]
and Lin [22] is not possible, as both researchers use different geometrical elements
to the ones in this case. Erdem [21] builds his results on tests carried out on small
plate elements with a maximal height of 150 mm (see also Soroushian and Lee
[23,24]), whereas Lin [22] investigates beam elements with different width and
thickness. Further experimental data on small beams specimens is available in
Rosenbusch [25], Dupont and Vandewalle [26] and the RILEM round-robin analysis
[27].
An exponentially decreasing relation (see Eq. (2)) between orientation factor
and specimen thickness can be obtained as a result of the own tests. The curve is
presented in Fig. 8.
g
zp==
0:699 e
0:0024h
2
In this equation, the orientation factor trends to 0 with h ?1. No precise
experimental data is available in the present case, therefore it is proposed that
the presented relation should be limited to specimens with a maximum thickness
of 400 mm. It is emphasized that during casting ber balling was encountered for
the last two specimens with 400 mm thickness. These ber balls (taken out of
the concrete mixture) with high concentration of bers might implicate a lower -
ber presence in the total concrete volume. Hence, the indicated values for ber
presence and orientation have to be considered with care, as a better mixture might
offer higher values. Furthermore, yield lines open at the cracks with the lowest ber
density, thus it is possible that higher ber concentration could be observed in adja-
cent parts to the yield lines. More complete information regarding the inuence of
the method of fabrication on strength properties is elaborated in Swamy and Stav-
rides [28].
3. Calculation model and slab design
3.1. Yield line theory
As all test specimens failed in bending with a ductile material
behavior, yield line theory (see [29]) can be used to obtain further
information on the materials behavior. In order to evaluate the
Fig. 1. Undulated steel ber TABIX 1.3/50.
Table 2
Summary of the test specimen geometry.
Plate Diameter D (m) Thickness h (mm) Column dimensions (mm mm) Openings (mm mm) Compressive strength (f
c,cyl
) (MPa)
PL1-25-0 2.34 250 350 350 54
PL2-20-0 2.34 200 350 350 53
PL3-30-0 2.34 300 350 350 42
PL4-25-1 2.34 250 350 350 1 (350 350) 46
PL5-40-0 1.9 400 250 250 40
PL6-40-1 1.9 400 250 250 1 (250 250) 42
J. Michels et al. / Construction and Building Materials 26 (2012) 145155 147
bearing capacity of a structure by means of yield line theory, an
equilibrium between outer loads and inner resistance forces is
established. Simultaneously, the admitted failure pattern needs
to be kinematically admissible. In general, the principle of virtual
works is applied for resolution: the outer works provoked by the
outer loads are compared to the inner works due to material resis-
tance in the yield lines.
3.2. Cross-section analysis
In order to obtain the structural resistance of the steel ber con-
crete cross-section, equilibrium between compression and tension
forces is set. As shown in Fig. 9a), a parabolic stressstrain relation
according to the DIN 1045-1 [30] is applied in the compressive
zone, whereas a simplied bilinear stressstrain relation dened
by the two points (e
1
,f
1
) and (e
2
,f
2
) in allusion to the bulletin on -
ber concrete of the German Concrete Association (DBV) [14] is used
in tension. The rst strain value e
1
is adopted with 0.1 as pre-
sented in the previous reference. Maximal concrete strain is
adopted with 3.5 (e
cu
) and 25 (e
2
) in compression and ten-
sion, respectively.
It is noted that the presented calculation and design procedure
represents a simplication of the stressstrain behavior of SFRC in
tension. As 4-point bending tests (see Section 2.1) according to [14]
are used as elements for small and large scale comparison, this
simplied approach is applied. More recent design suggestions re-
quest the inclusion of the uniaxial tensile strength in addition to
several predened post-cracking tensile strength levels for design.
Detailed information is given in the DAfStb-Richtlinie [31], RILEM
TC-162 [32], Teutsch et al. [33], ACI [34], and di Prisco et al. [35].
3.3. Inverse analysis and discussion
For the presented test setup, the analogy to a round plate sim-
ply supported along its border with central point load can be estab-
lished. For this loading case, the bending moment m in the yield
line is related to the applied concentrated force P (equivalent to
the reaction force on the center column in the present case) by
the following equation:
P
2 p m
1
2r
3R
3
Fig. 2. Cross-sectional and top view of the experimental test setup (dimensions in (mm)).
148 J. Michels et al. / Construction and Building Materials 26 (2012) 145155
With the previously presented cross-section analysis and the
related constitutive material law it is possible to dene the
strength parameters in tension f
1
and f
2
for each plate specimen
by the means of an iterative analysis consisting in a comparison
between the calculated results to the experimental data. For the
present analysis, only the full specimens without openings are
taken into account. Comparison between the experimental and
calculated values of the [14] bearing moment m
R
plotted against
the hinge rotation h is shown in Fig. 10. The hinge rotation (see
Fig. 9b) can be dened with the following Eq. (4). If the length
(l
pl
) of the zone in which plastic deformation occurs is equal to
the height of the tension zone hx on each side of the opening
crack, the hinge rotation h corresponds to the SFRC tensile strain
e
ct
([14,33]). The calculated strength values f
1
and f
2
for each
plate height are presented in Table 4.
H
w
h x

w
l
pl
e
ct
4
It can be observed that on average, the calculation model overes-
timates the experimental results at low rotation levels. This stiffer
behavior is due to the fact that the yield lines are supposed to appear
immediately after the linear elastic domain in the calculation model,
whereas a more distributed cracking with limited width on the top
surface of the plates was clearly visible prior to larger crack openings.
Fig. 3. Forcerotation curves of the experimental plate tests.
Fig. 4. Crack pattern on the top side after test end.
J. Michels et al. / Construction and Building Materials 26 (2012) 145155 149
Fig. 5. (a) Deection and (b) compressive strain on the bottom plate side in main bearing direction (in the columns axis) plotted against the offset distance s from the column
(example of PL1-25-0).
Fig. 6. Area division in one yield line for the evaluation of the ber dispersion and orientation.
Table 3
Fiber orientation factor g
//
in horizontal direction in the different yield lines observed after test end.
Test Layer Column Center Border Average s.d.
a
Column Center Border Average s.d.
a
PL1 Yield line 1 Yield line 2
1 0.20 0.18 0.16 0.18 0.02 0.22 0.24 0.21 0.22 0.01
2 0.31 0.26 0.28 0.28 0.02 0.27 0.38 0.27 0.31 0.05
3 0.29 0.29 0.25 0.28 0.02 0.33 0.34 0.37 0.35 0.02
4 0.81 0.70 0.71 0.74 0.05 0.51 0.61 0.45 0.52 0.07
PL2 Yield line 1 Yield line 2
1 0.42 0.20 0.23 0.29 0.10 0.37 0.22 0.23 0.28 0.07
2 0.33 0.30 0.27 0.30 0.02 0.37 0.21 0.28 0.29 0.06
3 0.39 0.44 0.43 0.42 0.02 0.37 0.34 0.43 0.38 0.04
4 0.65 0.80 1.07 0.84 0.17 0.51 0.33 0.74 0.53 0.17
PL3 Yield line 1 Yield line 2
1 0.32 0.25 0.27 0.28 0.03 0.54 0.34 0.32 0.40 0.10
2 0.41 0.34 0.16 0.31 0.10 0.43 0.55 0.43 0.47 0.06
3 0.59 0.47 0.21 0.42 0.16 0.48 0.47 0.50 0.48 0.01
4 0.90 0.41 0.44 0.58 0.23 0.44 0.52 0.79 0.59 0.15
PL4 Yield line 1 Yield line 2
1 0.31 0.34 0.23 0.29 0.05 0.16 0.11 0.25 0.17 0.06
2 0.32 0.26 0.17 0.25 0.06 0.22 0.34 0.29 0.28 0.05
3 0.31 0.34 0.35 0.33 0.01 0.36 0.32 0.50 0.39 0.07
4 0.52 0.56 0.71 0.60 0.08 0.37 0.38 0.47 0.41 0.05
Yield line 3
150 J. Michels et al. / Construction and Building Materials 26 (2012) 145155
Table 3 (continued)
Test Layer Column Center Border Average s.d.
a
Column Center Border Average s.d.
a
1 0.22 0.24 0.17 0.21 0.03
2 0.38 0.29 0.29 0.32 0.04
3 0.35 0.28 0.34 0.32 0.03
4 0.35 0.61 0.46 0.47 0.11
PL5 Yield line 1 Yield line 2
1 0.23 0.20 0.16 0.20 0.03 0.13 0.28 0.19 0.20 0.06
2 0.27 0.26 0.22 0.25 0.02 0.18 0.27 0.23 0.23 0.04
3 0.28 0.25 0.19 0.24 0.04 0.22 0.26 0.22 0.23 0.02
4 0.34 0.30 0.24 0.29 0.04 0.31 0.31 0.23 0.28 0.04
Yield line 3
1 0.16 0.26 0.24 0.22 0.04
2 0.23 0.19 0.29 0.24 0.04
3 0.27 0.25 0.33 0.28 0.03
4 0.28 0.27 0.29 0.28 0.01
PL6 Yield line 1 Yield line 2
1 0.22 0.27 0.12 0.20 0.06 0.35 0.18 0.14 0.22 0.09
2 0.21 0.16 0.16 0.18 0.02 0.28 0.13 0.18 0.20 0.06
3 0.20 0.17 0.27 0.21 0.04 0.26 0.25 0.16 0.22 0.05
4 0.28 0.30 0.24 0.27 0.02 0.30 0.21 0.31 0.27 0.04
Yield line 3
1 0.14 0.19 0.19 0.17 0.02
2 0.15 0.11 0.19 0.15 0.03
3 0.21 0.13 0.28 0.21 0.06
4 0.18 0.15 0.31 0.22 0.07
a
s.d. = standard deviation.
Fig. 7. Horizontal orientation factors g
//
in the different yield lines for each plate specimen.
J. Michels et al. / Construction and Building Materials 26 (2012) 145155 151
Fig. 11a shows the residual tensile strength f
1
and f
2
plotted
against the ber orientation factor dened in Nomenclature. Out
of the presented tendency, it can be concluded that the residual
tensile strengths are dependent on the ber orientation, itself a
function of the geometrical dimensions. Hence, a size effect regard-
ing post-cracking tensile strength is present for large scale SFRC
elements. Rosenbusch [25] describes identical results on differ-
ences in residual tensile strength due to different ber orientation,
whereas Soroushian and Lee [23] present similar observations with
regard to the maximum uniaxial tensile strength. The following
linear relations connect the presented residual tensile strengths
f
1
and f
2
to the horizontal orientation factor g:
f
1
3:08 g
==
0:62 5
f
2
4:34 g
==
0:68 6
Fig. 8. Fiber orientation factor dependent on the specimen height.
Fig. 10. Moment hinge rotation curves for h = 20, 25, 30 and 40 cm experimental results against calculations.
Fig. 9. (a) Constitutive material laws in compression and tension as admitted to the cross-section analysis (e
1
= 0.1, e
2
= 25) and (b) geometrical denitions of the crack
opening.
152 J. Michels et al. / Construction and Building Materials 26 (2012) 145155
Preliminary 4-point bending tests performed by the authors on
small beams (700 mm (length) 150 mm (width) 150 mm
(height), span of 600 mm) with the same ber concrete material
showed mean residual tensile strength of 2.2 and 1.6 MPa for f
1
and f
2
, respectively. The average ber orientation factor was 0.5.
Hence, the geometry factors v
1
and v
2
for each specimen height
can be derived by relating the residual tensile strength values f
1,i
and f
2,i
for a dened specimen thickness to its corresponding value
of the 4-point bending beam f
1,150
and f
2,150
:
f
1;i
f
1;150
v
1

3:08 g
i
0:62
3:08 g
150
0:62
7
f
2;i
f
2;150
v
2

4:34 g
i
0:68
4:34 g
150
0:68
8
The index i refers to the respective specimen height. The
descending curves for v
1
and v
2
with growing plate height are
shown in Fig. 11b. These factors will be subsequently used for
deriving the design strengths of large-scale SFRC plates.
3.4. Slab design
3.4.1. Possible failure pattern
Failure pattern [36] as presented in Fig. 12a is considered for de-
sign. Parallel yield lines develop in the column rows as well as at
mid-span. Preliminary analysis [37,38] on different possible failure
pattern revealed that this type for an edge panel (plate clamping
only over two borders) was found to be the decisive case for equal
span lengths in both horizontal directions. The presented crack
pattern develops if the equally distributed load is applied in all slab
Table 4
Residual tensile strengths f
1
and f
2
for different speci-
men heights obtained by iterative analysis.
h (mm) f
1
(MPa) f
2
(MPa)
200 1.9 1.3
250 1.7 1
300 1.9 0.8
400 1.3 0.2
Fig. 11. (a) Tensile strength f
1
and f
2
dependent on the horizontal orientation factor and (b) geometry factors v
1
and v
2
for different element thickness.
Fig. 12. (a) Failure pattern for edge slab panel under constant uniformly distributed loading and (b) design diagram for different slab thickness.
J. Michels et al. / Construction and Building Materials 26 (2012) 145155 153
elds. It is assumed that bending failure occurs for all loading sce-
narios, punching failure is excluded.
3.4.2. Design procedure and safety concept
For practical application, the structural designer needs both
compression as well as tensile and residual tensile strength values
of a determined type of SFRC in order to be able to determine the
bearing capacity of a cross-section. Usually, these values are given
by the concrete producer, who obtains the results by performing
compression tests on cubes and cylinders and exural tests on
small beams as foreseen by most of the design and model codes.
However, from the previous results, it was shown that the tensile
strengths on small-sized beam specimens are not representative
for large-scale constructions. Hence, the suggested geometry fac-
tors are introduced in order to correctly assess the tensile strengths
in case of plate elements with different thickness. Furthermore, de-
sign codes always lower the effective experimental strengths of the
material throughout different safety factors. In the present study,
compression strength was taken into account according to DIN
1045-1 [30]. For the material resistance in tension, semi-probabi-
listic safety design (presented in Michels [37] and Michels et al.
[38]) on residual tensile strengths observed in 4-point bending
tests according to the DBV-bulletin [14], reveals a safety factor c
R
of 1.8 for mean strength values. The used statistical input data
(strength variation) is in accordance to values presented in Hemmy
[39]. A creep factor a
c
of 0.85, as suggested in [14], is included, too.
Regarding the value of 0.85, the authors believe that this proposi-
tion is too optimistic regarding the clearly lower experimental val-
ues obtained by Gossla [40]. Creep of SFRC at cracked state is an
essential research topic and further research data is necessary in
order to correctly assess the materials long-term behavior.
To summarize, tensile and compression strength values are
lowered by their respective safety factor. Subsequently, in order
to respect the observed size effects, the obtained tensile strengths
are reduced by the aforementioned geometry factor. Finally, a
creep factor is included. The same cross-section analysis as pre-
sented in Section 3.2 is applied to obtain the design bearing capac-
ity of the cross-section.
Eventually, in combination with the failure pattern admitted in
Section 3.4.1, the bearing moment m
Rd
in the cross-section has to
be compared to the bending moment m
d
in the yield line. Eq. (9)
presents the control.
m
Rd
P
?
m
d
)Structural safety assured! 9
A summarizing ow-chart is given in Fig. 13.
3.4.3. Design diagram
The abaci shown in Fig. 12b give design values for different span
lengths (equal in both directions) and slab heights by considering
the failure pattern presented in Section 3.4.1. The bearing capacity
of the cross-section was evaluated at a tensile strain of 25, con-
sidered by the majority of guidelines as the maximal allowed strain
for moment redistribution.
The results shown in Fig. 12b take into account the dead loads
(including a oor screed of about 5 cm (1.2 kN/m
2
)) of the struc-
ture with an increase factor of 1.35 according to DIN 1055 [41]
and a trafc load of 3 kN/m
2
majored with a factor of 1.5. The load
value is typical for the normal use of an ofce building. With an
increasing span length, the bending moments m
d
increase, whereas
the bearing moment m
Rd
in the cross-sections remains constant. If
the related line of the bearing moments (dotted line) stays above
Fig. 13. Flow-chart of the design procedure.
154 J. Michels et al. / Construction and Building Materials 26 (2012) 145155
the increasing bending moment line (full line), structural safety at
ultimate load state is assured.
It arises from the diagrams that possible span lengths with a
sufcient structural safety range between 3.5 and 4.5 m, clearly be-
low common values of about 7 m for conventional reinforced con-
crete. Due to the increasing dead loads, thicker plates do not offer
any considerable advantage. Hence, practical application should be
limited to special cases in which reduced span lengths are
mandatory.
4. Conclusions
The presented results allow to draw a certain number of conclu-
sions. Even by using a high ber dosage of 1.3% of the total volume,
no punching failure is observed. All test specimen exhibit a ductile
failure in bending. For other similar ber types used as only rein-
forcement, a punching shear failure is in general also not likely
to occur, as the upper tensile zone is unable to guarantee a suf-
cient tensile resistance in order to allow the diagonal crack devel-
opment. Supplementary analysis on the ber presence in the yield
lines reveals a high dependence of the ber dispersion on the con-
crete composition. A cement matrix, which is not stiff enough,
might implicate a ber segregation towards the formwork bottom.
Hence, high attention has to be paid to an adequate concrete mix-
ture at an early stage. Furthermore, it could be demonstrated that
ber orientation and residual tensile strength at cracked state de-
crease with growing slab thickness. Two geometry factors imple-
mented into a simplied bilinear stressstrain law in tension
incorporate this size effect. By following a standard design proce-
dure taking into account safety factors on the resistance side as
well as increase factors for the present loads, possible spans for
the proposed material composition and ber dosage/type are de-
rived. Compared to conventional RC at slabs, one notices shorter
span lengths for the same slab thickness. Currently, considerable
lack of knowledge persists regarding the exact creep behavior of
the used material and ber dosage/type. In general, a correct de-
sign with any ber type should always include a correct reection
of the long-term performance, which should be provided by the
producer. The presented experimental results, especially on ber
orientation for large-scale SFRC elements, should be veried by
further experimental analysis.
Acknowledgements
The authors want to express their gratitude to their industrial
partner ArcelorMittal Wire Solutions for the nancial support of
the project as well as Dipl.-Bauing. ETH Ren Pepin for his profes-
sional assistance. Furthermore, the help of the structural labora-
tory staff at University of Luxembourg is highly appreciated.
References
[1] Reinhardt H-W. Beton. In: Betonkalender; 2005.
[2] Barros JAO, Sena Cruz J. Fracture energy of steel ber-reinforced concrete.
Mech Compos Mater Struct 2001;8(1):2945.
[3] Barros JAO, Cunha VMCF, Ribeiro AF, Antunes JAB. Post-cracking behaviour of
steel bre reinforced concrete. Mater Struct/Mat Constr 2005;38:4756.
[4] Rossi P. Mechanical behaviour of metal-bre reinforced concretes. Cem Concr
Compos 1992;14:316.
[5] Altun F, Haktanir T, Ari K. Effects of steel ber addition on mechanical
properties of concrete and RC beams. Constr Build Mater 2007;21(3):65461.
[6] Xu BW, Shi HS. Correlations among mechanical properties of steel ber
reinforced concrete. Constr Build Mater 2009;23(12):346874.
[7] Sorelli L, Meda A, Plizzari G. Steel ber concrete slabs on ground: a structural
matter. ACI Struct J 1997;103(4):5518.
[8] Meda A, Plizzari G. New design approach for steel ber-reinfocred concrete
slabs-on-ground based on fracture mechanics. ACI Struct J
2004;101(3):298303.
[9] Gossla U. Flachdecken aus Stahlfaserbeton. Beton- und Stahlbetonbau
2006;101(2):94102.
[10] Destre X. Free suspended elevated at slabs of steel bre reinforced concrete:
full scale tests and design. In: 7th international RILEM-symposium on bre
reinforced concrete, Chennai; 2008. p. 94150.
[11] Pepin R. Structural applications for SFRC. Central European Congress on
Concrete Engineering; 2009.
[12] Yazici S, Inan G, Tabak V. Effect of aspect ratio and volume fraction of steel
ber on the mechanical properties of SFRC. Constr Build Mater
2007;21(6):12503.
[13] ArcelorMittal Wire Solutions. Technical Data Sheet Undulated bres Tabix
1.3/50; 2010.
[14] DeutscherBeton-Verein EV. (DBV). Merkblatt Stahlfaserbeton; 2001.
[15] Gettu R, Gardner DR, Saldivar H, Barragan BE. Study of the distribution and
orientation of bers in SFRC specimens. Mater Struct/Mat Constr
2004;38:317.
[16] Schfers U. Konstruktion, Bemessung und Sicherheit gegen Durchstanzen von
balkenlosen Stahlbetondecken im Bereich von Innensttzen. Heft Nr. 357 der
Schriftenreihe des DAfStb; 1984.
[17] Guandalini S. Poinonnement symtrique des dalles en bton arm. Ph.D.
thesis, cole polytechnique Fdrale de Lausanne; 2005.
[18] Beutel R. Durchstanzen schubbewehrter Flachdecken im Bereich von
Innensttzen. Ph.D. thesis, Rheinisch-Westflisch Technische Hochschule
Aachen; 2002.
[19] Krenchel H. Fibre reinforcement: theoretical and practical investigations of the
elasticity and strength of bre-reinforced materials. Akademisk Forlag; 1964.
[20] DIN EN 206-1 Beton: Festlegung, Eigenschaften, Herstellung und
Konformitt; 2001.
[21] Erdem E. Probabilistisch basierte Auslegung stahlfasermodizierter
Betonbauteile auf experimenteller Grundlage. Bochum. Ph.D. thesis, Ruhr-
Universitt Bochum; 2002.
[22] Lin Y-Z. Tragverhalten von Stahlfaserbeton. Ph.D. thesis, Universitt Karlsruhe;
1996.
[23] Soroushian P, Lee C-D. Tensile strength of steel ber reinforced concrete.
Correlation with some measures of ber spacing. ACI Mater J
1990;87(6):5416.
[24] Soroushian P, Lee C-D. Distribution and orientation of bers in steel ber
reinforced concrete. ACI Mater J 1990;87(5):4339.
[25] Rosenbusch J. Einuss der Faserorientierung auf die Beanspruchbarkeit von
Bauteilen aus Stahlfaserbeton. Beton-und Stahlbetonbau 2004;99(5):3727.
[26] Dupont D, Vandewalle L. Distribution of steel bres in rectangular sections.
Cem Concr Compos 2005;27(3):3918.
[27] RILEM TC162-TDF. Round-robin analysis of the RILEM TC 162-TDF beam
bending test: Part 3 - bre distribution. Mater Struct/Mat Constr
2003;36(263):6315.
[28] Swamy RN, Stavrides H. Inuence of the method of fabrication on strength
properties of steel bre concrete. Mat Constr 1976;9(4):24353.
[29] Johansen KW. Yield-line theory: Cement and Concrete Association. 1st ed.;
1962.
[30] DIN 1045-1: Tragwerke aus Beton, Stahlbeton und Spannbeton, Teil 1
Bemessung und Konstruktion; 2001.
[31] Deutscher Ausschuss fr Stahlbeton. Richtlinie Stahlfaserbeton. Berlin: Beuth
Verlag; 2010.
[32] Vandewalle L, Nemegeer D, Balazs L, Barr B, Bartos P, Banthia N, et al.
Recommendations of RILEM TC 162-TDF: re-design method. Mater Struct/
Mat Constr 2000;33(226):7581.
[33] Teutsch M, Wiens U, Alfes C. Stahlfaserbeton nach DAfStb-Richtlinie
Stahlfaserbeton. Beton-und Stahlbetonbau 2010;105(8):53951.
[34] ACI Committee 544. Design consideration for steel ber reinforced concrete.
ACI Struct J 1988;85(5):56380.
[35] Di Prisco M, Plizzari G, Vandewalle L. Fibre reinforced concrete: new design
perspectives. Mater Struct/Mat Constr 2009;42(9):126181.
[36] Favre R, Jaccoud J-P, Burdet O, Charif H. Dimensionnement des structures en
bton arm Aptitude au service et lments de structures: presses
polytechniques et universitaires romandes; 1997.
[37] Michels J. Bearing capacity of steel ber reinforced concrete at slabs. Ph.D.
thesis, University of Luxembourg; 2009.
[38] Michels J, Maas S, Zrbes A, Waldmann D. Tragverhalten von Flachdecken aus
Stahlfaserbeton im negativen Momentenbereich und Bemessungsmodell fr
das Gesamtsystem. Beton-und Stahlbetonbau 2010;105(8):496508.
[39] Hemmy O. Zum Gebrauchs-und Tragverhalten von Tunnelschalen aus
Stahlfaserbeton und stahlfaserverstrktem Stahlbeton: Heft Nr. 549 der
Schriftenreihe des DAfStb; 2004.
[40] GosslaU. TragverhaltenundSicherheit betonstahlbewehrter Stahlfaserbetonteile.
Ph.D. thesis, Technische Universitt Braunschweig; 1999.
[41] DIN 1055-1 and DIN 1055-3. Einwirkungen auf Tragwerke; 2002.
J. Michels et al. / Construction and Building Materials 26 (2012) 145155 155