Materials and Design 54 (2014) 3242

Contents lists available at ScienceDirect

Materials and Design

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

Mechanical behavior of hybrid steel-ber self-consolidating concrete: Materials and structural aspects
Dimas Alan Strauss Rambo, Flvio de Andrade Silva , Romildo Dias Toledo Filho
Civil Engineering Department, COPPE, Universidade Federal do Rio de Janeiro, P.O. Box 68506, CEP 21941-972, Rio de Janeiro, RJ, Brazil

a r t i c l e

i n f o

a b s t r a c t
This work presents the preliminary results of an experimental investigation on the mechanical behavior of self-consolidating concrete reinforced with hybrid steel bers in the material and structural scale. Straight and hooked end steel bers with different lengths and diameters were used as reinforcement in ber volume fractions of 1.0 and 1.5%. In the fresh state the concrete was characterized using the slump ow, L-box and V-funnel tests. To determine the effect of the hybrid reinforcement on the plastic viscosity and shear yield stress a parallel plate rheometer was used. Following, the mechanical response was measured under tension and bending tests. In the exural test, the movement of the neutral axis was experimentally determined by strain-gages attached to compression and tensile surfaces. Furthermore, the mechanical response of the material under bi-axial bending was addressed using the round panel test. During the test the crack opening was measured using three linear variable differential transformers (LVDTs). The cracking mechanisms were discussed and compared to that obtained under four point bending and direct tension. The obtained results indicated that the ber hybridization improved the behavior of the composites for low strain and displacement levels increasing the serviceability limit state of the same through the control of the crack width. For large displacement levels the use of the longer bers led to a higher toughness but with an expressive crack opening. Due to its structural redundancy the round panel test allowed the formation of a multiple cracking pattern which was not observed in the four point beam tests. Finally, the obtained materials properties were used in a nonlinear nite element model to simulate the round panel test. The simulation reasonably agreed with the experimental test data. 2013 Elsevier Ltd. All rights reserved.

Article history: Received 24 April 2013 Accepted 5 August 2013 Available online 14 August 2013 Keywords: Hybrid reinforcement Round panel test Crack formation Steel ber Self-consolidating concrete

1. Introduction The rst works on ber reinforced concrete (FRC) were realized in the 1950s and 60s decades of the last century with the aim of understanding the mechanical behavior of steel ber reinforced concrete [1,2]. Since that period, other bers have been evaluated as reinforcement in concrete elements, but steel is still the most used ber. Its popularity is associated with the fact that steel presents a good afnity with concrete, the ease of use, the high toughness and resistance to static and dynamic loads [3]. Several categories of ber reinforced concrete have been developed over the past three decades presenting different mechanical properties. Conventional FRC presents an increase in the ductility when compared with the plain matrix showing a strain softening behavior after the appearance of the rst crack. On the other hand the high performance ber reinforced cementitious composites (HPFRCC) exhibit a deection hardening behavior and

Corresponding author. Tel.: +55 (21) 2562 8493x48; fax: +55 (21) 2562 8484.
E-mail address: fsilva@coc.ufrj.br (Flvio de Andrade Silva). 0261-3069/$ - see front matter 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.matdes.2013.08.014

can also present a strain-hardening type of response accompanied by multiple cracking in tension which leads to an improvement in strength and toughness compared to the non-reinforced matrix [46]. In the last few decades signicant improvements in the development of cement based materials have been achieved resulting in high performance concrete that can present uniaxial compressive strength ranging from 150 to 400 MPa [5,7,8]. These improvements were only possible due to developing techniques of cement paste microstructure densication using efcient superplasticizing chemical additives and ultra-ne particles. The production of hybrid ber reinforced self-consolidating concretes aims to combine the mechanical properties of two or more different bers to the rheological characteristics of self-consolidating matrices. Hybrid reinforcement systems can be used in order to take advantage of each individual ber properties [9], which depend on the shape, type, size and the volume fraction of the used bers [1]. These composite systems can improve not only exural and tensile strength, but can also lead to a change in the cracking mechanisms. The manufacturing, the ber dispersion and the ber orientation are very important to improve the post-cracking

D.A.S. Rambo et al. / Materials and Design 54 (2014) 3242

33

response of the ber reinforced concrete [10]. Thus, the rheological properties of the matrices should be suitable for the ber addition [11]. It is worth noticing that most of the research performed on FRC systems found in the literature uses in the mix design matrices that do not contain coarse aggregates, but instead, ne aggregates [1214]. Kim et al. [14] studied the exural performance of hybrid ultra high performance FRC using one micro (L = 13 mm) and four types of macro (L = 30 and 62 mm) high strength steel bers. The results showed an improvement in deection and toughness for hybrid systems in comparison to systems reinforced by micro bers only. Akcay and Tasdemir [15], produced four different HSFRSCCs (hybrid steel ber reinforced self-compacting concrete) and reported that it is possible to add a volume fraction of bers up to 1.5% without affecting its workability. The mechanical behavior showed that the ber hybridization increased the concrete fracture energy and ductility. A multi-scale reinforced cement composite was developed by s (LCPC) Rossi et al. at the Laboratoire Central des Ponts et Chausse [16]. Two types of materials were developed by using the multiscale concept. A cement based compositewhich was reinforced by 7% of two metal bers of different geometries, and the CEM TECmultiscale that was reinforced by 11% of three classes of steel bers [17,18]. Both materials present a tension hardening behavior but the MSCC can achieve up to 15 MPa under direct tension while the CEMTEC up to 20 MPa. The mechanical behavior of ber reinforced concrete is usually evaluated using bending tests, mostly performed in small prisms. This type of test does not represent the real structural behavior because it results in a different cracking mechanism and normally leads to a higher dispersion on experimental results [10,1921]. Structural or quasi-full scale tests as ASTM: C-1550, however, have greater representation in relation to the concrete volume, failure mechanisms and toughness. As reported by Bernard [20] the mechanical tests need to reect the material variations and not variations associated to the test method. Round panel tests were performed in steel ber reinforced concretes with different dimensions. Results indicate that the main advantage of this test is to allow the detection of a multiple crack pattern which is not observed in tests with small beams. Bernard [22] investigated the inuence of support conditions on exural and shear behavior of steel ber reinforced concrete slabs. According to the author, bending tests performed on panels supported on three points show a consistent failure mode and allows a more reliable measure of the concrete performance when compared to alternative methods of support. Minelli and Plizzari [19] performed a comparison between round panel and exural beam tests. Results reported that the geometry and fracture area involved in round panel tests leads to a lower dispersion resulting in a better representation of the real structural behavior. The effects of the steel ber hybridization on the rheological and mechanical properties of self-consolidating FRC are addressed in the present work. Two different hybrid FRC systems were produced, using straight and hooked end steel bers with different lengths, in ber volume fractions of 1.0% and 1.5%. The self consolidating concrete matrix was designed and produced based on the compressible packing model. A parallel plate rheometer was used to determine the inuence of ber hybridization on the plastic viscosity and shear yield stress. Furthermore, empirical rheological tests were performed. Mechanical tests were carried out in the structural and materials scale and the changes in the cracking mechanisms were investigated. Direct tension and four point bending tests were performed in the materials scale while the round panel tests for the structural testing. A non-linear nite element model was used to simulate the mechanical behavior of the studied FRC system in the structural scale.

2. The compressible packing model The compressible packing model (CPM) was developed by de Larrard and his collaborators and used in this research to design the matrix of the self-consolidating ber reinforced concrete [23,24]. Composite materials like concrete are made up of grains embedded in a matrix. The aim of the design is to use the least possible amount of binder by combining these grains in order to minimize the concrete porosity [23]. The equation representing the virtual packing density of a granular mix containing n classes of grains, ordered in such a way that its diameters are d1 ! d2 ! . . . di ! di1 ! . . . ! dn , when the class i is dominant, is expressed by the following equation:

ci

bi i 1  n   X X 1 i 1 yj 1 bi bi;j bi 1 b yj 1 ai;j b bj j
j1 ji1
(i)

where c is the virtual packing density when the ith class is dominant; yi is the volumetric fraction of the ith class; bi is the virtual packing density of the ith class; it represents the volume of grains contained in an unitary volume, compacted with an ideal compaction energy that would correspond to a maximum virtual packing; ai,j and bi,j represent the loosening effect and the wall effect exerted by the grains, respectively; they can be determined either experimentally or by the following formulas:

ai;j

q 1:02 1 1 dj =di
1:50

bi;j 1 1 di =dj

The virtual compactness of the mix can be found by using the formula:

c inf ci

where inf indicates the least value. The actual compactness depends on three main parameters: the size of the grains, the shape of the grains, and the method of processing the packing. The compressible packing model allows making the transition from virtual compactness, which cannot be obtained in practice, to the actual compactness of the mix, which depends on the energy being applied at the time of placing. A scalar K called compaction index enables connecting the virtual compactness (c) with the actual compactness (/). This scalar is strictly dependent on the protocol implemented for the particular mix. As K tends to innity, the compactness / tends to the virtual compactness c. The general shape of the compaction index equation, for n classes of grains, is as follows:

n X i 1

yi =bi 1=/ 1=ci

where / is the actual compactness of the granular mix. The values of index K are calculated from the binary mixes for each placing processes. K assumes a value of 4.5 when the compaction process is the simple pouring, 6.7 for water demand and 9 when the placing process is vibration plus 10 kPa compression [23]. If the actual compactness for a single granular class i (/i) is experimentally determined, by means of a compaction process having compaction index K, it is possible to use Eq. (5), derived from Eq. (4), to determine the virtual compactness of the granular class i.

bi

/i 1 K K

34

D.A.S. Rambo et al. / Materials and Design 54 (2014) 3242 Table 1 Mix composition of concretes. Constituent Mixtures Matrix Coarse aggregate (G) (kg/m3) Sand (S1) (kg/m3) * Sand (S2) (kg/m3) Silica mesh 325 (kg/m3) Cement (kg/m3) Fly ash (kg/m3) Silica fume (kg/m3) Superplasticizer (kg/m3) Viscosity modier (kg/m3) Water (kg/m3) Straight ber (SF1) (kg/m3) Hooked end ber (SF2) (kg/ m3) Superplasticizer (%) Water/binder ratio
*

Eq. (4) is an implicit equation in / and allows the determination of the actual compactness since the other variables are all known. To use the model it was determined the virtual compactness, size grading distributions and specic gravity of the constituents as well as the cement contribution to compressive strength and the saturation dosage of the chemical additive.

C1.0%H 468 830 100 70 360 168 45 45.1 0.36 150 39 39 4 0.50

C1.0% 468 830 100 70 360 168 45 45.1 0.36 150 0 78 4 0.50

C1.5%H 454 830 100 70 360 168 45 45.1 0.36 150 39 78 4 0.50

C1.5% 454 830 100 70 360 168 45 45.1 0.36 150 0 117 4 0.50

3. Materials and processing The matrix was designed following the CPM routine described in Section 2. For the ber addition the rheological behavior of the matrix was adjusted through the use of superplasticizer and a viscosity modier. The materials used in the self-consolidating concrete composition were a Brazilian slag cement type CPIII 40 with a 28 days compressive strength of 40 MPa, river sand with two classes of particle size: one ranging from 0.15 mm to 4.8 mm and the other from 0.15 mm to 0.85 mm, coarse aggregates with maximum diameter of 9.5 mm, y ash, silica our (ground quartz), silica fume, and a polycarboxilate superplasticizer with solid contents of 31.2%. The water/cementitious material ratio of the selfcompacting composite was 0.32. Fig. 1 shows the grain size distribution of the used binder materials and aggregates. Two steel ber types, one straight (SF1) and one with hooked ends (SF2), were used as reinforcement. The straight ber presented a tensile strength of 1100 MPa, elastic modulus of 200 GPa and a density of 7.85 g/cm3, and the ber with hooked ends (SF2) a tensile strength of 1150 MPa, elastic modulus of 200 GPa and a density of 7.85 g/cm3. The length and aspect ratio of the SF1 ber were 12 mm and 67 and those presented by SF2 were 35 mm and 65, respectively. The SF1 ber was produced with a brass coating providing the ber a relatively smooth surface. Five concrete mixtures, with the proportions presented in Table 1 were produced. One control mixture without steel bers, two mixtures with a ber volume fraction of 1.0% (78 kg/m3) named as C1.0% (1.0% of SF2) and C1.0%H (0.5% of SF1 + 0.5% of SF2) and two mixtures with a ber volume fraction of 1.5% (117 kg/m3) named as C1.5% (1.5% of SF2) and C1.5%H (0.5% of SF1 + 1.0% of SF2). The concretes were produced in a room with controlled temperature of 21 C 1 C using a planetary mixer (previously moistured) of 100 l capacity. The cementitious materials were homogenized by dry mixing for 60 s prior to the addition of sand. The dry ingredients were mixed for an additional 60 s prior to addition of superplasticizer and water. The mixture was blended for 8 min.

494 830 100 70 360 168 45 45.1 0.36 150 0 0 4 0.50

Sand (S2): sand (S1) with diameter less than 0.85 mm.

For the production of the samples, the concrete mixtures were placed in the steel molds in one layer. The round panel forms were made using a thin steel sheet and a wood base with thickness of 20 mm. After consolidation, the round panel surfaces were leveled with a steel plate. The samples did not suffer any kind of vibration. The specimens were covered in their molds for 48 h prior to moist curing for 28 days in a cure chamber with 100% RH and 23 1 C. 4. Rheological tests The effect of the hybrid reinforcement on the plastic viscosity and shear yield stress was measured with a parallel plate rheometer developed at the LCPC [25]. Ten rotation speeds ranging from 0.2 rev/s to 0.8 rev/s were used to perform the tests. These speeds correspond to strain rates of 0.25 s1 and 6.0 s1. The tests with V funnel were performed in an apparatus with rectangular top and bottom sections of 515 mm 75 mm and 65 mm 75 mm (length and width), respectively [26]. Through this test the elapsed time (in seconds) between the opening of the bottom and the time when all the concrete ows through the lower section can be determined in order to measure the concrete ow ability. The L box test was used to measure the lling and passing ability of the studied concretes [27]. Three different congurations of bars (12 mm of diameter) were used. The space between bars was 94 mm (one bar), 58.66 mm (two bars) and 41 mm (three bars).

Fig. 1. Particle size distribution of (a) the used cementitious materials and (b) aggregates.

D.A.S. Rambo et al. / Materials and Design 54 (2014) 3242

35

Finally, the slump ow test was used to describe the ow ability of the fresh mixtures in unconned condition [28]. The determination of the concrete spread was performed by the arithmetic average of two measurements of the perpendicular concrete mass diameter. 5. Mechanical testing 5.1. Direct tensile test Direct tensile tests were performed in a Shimadzu universal testing machine model UH-F1000kN with a computer-controlled hydraulic servo system. The tests were controlled by the actuator displacement at a rate of 0.1 mm/min (strain rate of 1.6 105 s1). Four dumbbell shaped specimens measuring 400 mm 100 mm 48.5 mm (length width thickness) were tested using a gage length of 100 mm with xed-hinged boundary conditions for each concrete mix produced. Fig. 2 shows the testing arrangement for the tensile test. Steel plates were glued on both ends of the specimen at the gripping regions. A set of mechanical grips were used. The tensile load and actuator displacement were recorded. 5.2. Four point bending test The bending tests were performed in the same Shimadzu universal testing machine used in direct tensile tests. Three specimens were produced for each concrete mix produced, with cross sectional dimensions of 100 mm 100 mm (width thickness) and

length of 400 mm. The span used between supports was 300 mm long and between the loading knives 100 mm. The tests were controlled by the actuator displacement at a rate of 0.5 mm/min. The deections of the beams at the mid-span were measured using one LVDT. The neutral axis movement was experimentally determined by strain-gages (51.1 mm of length) attached to compression and tensile surfaces. The crack opening in the bottom sample face of the beams was measured using one LVDT positioned on the mid-span and aligned with the length. Furthermore, the cracking development was also recorded during the loading cycle of the bending test at regular time intervals for posterior image analysis. A Nikon D90 digital camera with an AF Micro Nikkor 60 mm lens (f/2.8D) and frame grabber captured images of 4288 2848 pixels in resolution at 60 s intervals. The images were used to measure the crack width during bending tests. 6. Round panel test Structural tests were performed through round panel tests based on the ASTM: C-1550. The nominal diameter and the thickness of all tested specimens were 750 mm and 75 mm, respectively. Three symmetrically arranged pivots (40 mm of diameter) were used to support the samples (see Fig. 3). Three round panel specimens were tested to failure in a Shimadzu servo-controlled testing machine for each concrete mix produced. The load was applied using a rigid steel cylinder (50 mm of diameter) onto the upper surface of the specimen at an actuator displacement rate of 0.5 mm/min. The deection response (until 40 mm) was measured using a LVDT positioned at the central part of the bottom surface of the specimens. During the tests the crack opening, in the bottom surface, was measured using a triangle shaped system of three LVDTs with three equal sides of 190 mm. 7. Discussion and analyses 7.1. Rheological properties Fig. 4 shows the inuence of steel bers on the self-compacting concrete ow measured with the parallel plate rheometer used in this study. It can be seen that while the unreinforced concrete behaves linearly following the Bingham model the FRC curves are non-linear. The rheological parameters of the ber reinforced concretes were then calculated using the HerschelBulkley model where, s0o a and b are characteristics material parameters that describe the fresh behavior of concrete.

_b s s0o ac

To correlate the material parameters s00 and a of the Herschel Bulkley model with the numerical parameters Co, A and b the expressions derived by De Larrard and co-workers [29], shown in Eqs. (7) and (8), were used. In those expressions R1 and R2 are the internal and external sample radii of the rheometer (20 mm and 120 mm, respectively), and h is the height of the sample (100 mm).

s0o

3
3 2pR3 2 R1

Co
h
b

a 0; 9

b 3 2p
b 1 3 Rb 2

b3 R1

Fig. 2. Direct tensile test set-up.

The plastic viscosity was determined using the equivalent parameters a and b using Eq. (9), which correlates the Herschel _ max represents the maximum Bulkley and Bingham models, where c _ max 6 s1 . strain rate used during the test c

36

D.A.S. Rambo et al. / Materials and Design 54 (2014) 3242

Fig. 3. Round panel test set-up: (a) details of the supports and LVDTs and (b) detail of the crack opening measurement system.

Fig. 4. Inuence of steel bers on the self-compacting concrete ow of (a) concrete reinforced with 1.0% of steels bers and (b) with 1.5%. The experimental points of the FRC were adjusted by the HerschelBulkley model.

 l

3a b1 _ c b 2 max

From Fig. 4 and Table 2 it can be seen that, although a non-linear (HerschelBulkley) behavior was noticed, the non-hybrid systems presented a shear yield stress and viscosity lower than that observed in the unreinforced matrix. On the other hand the hybridization of bers (i.e. the addition of the 12 mm straight steel ber) increased the viscosity of the mix. This behavior was more pronounced for the reinforcement ratio of 1.5%. Comparing the hybrid system reinforced with 1.0% (Fig. 4a) to the one with 1.5% (Fig. 4b) it can be seen that the shape of the curves are different and that for speeds above 0.5 rev/s the non-linearity increases in a different rate. This behavior indicates that the shear yield strength increases when increasing the ber volume fraction of the hybrid systems. This was conrmed when computing the shear yield stress which was found to be 111 Pa for the mix reinforced with 1.0% and 220 Pa for that reinforced with 1.0% of steel ber. As showed in Table 2, even with the incorporation of steel bers, all concretes had adequate viscosity and deformability under
Table 2 Results of rheological tests. Mixtures Spread (mm) V-funnel (s) L-box (H2/H1) I Matrix C1.0%H C1.0% C1.5%H C1.5% 620 705 720 620 700 18.15 18.39 19.28 BL 19.56 0.95 0.95 1.00 0.53 0.86

their own weight. All studied concretes can be classied as selfcompacting as they presented values of slump ow consistent with the lower limit given by EFNARC 2005 [30] (P550 mm). The slump ow diameter values ranged from 620 to 700 mm. The unreinforced matrix and the concretes produced with 1.0% of bers (C1.0% and C1.0%H) also meet the EFNARC specication for viscosity (V funnel) and passing ability (L-box). The concrete reinforced with 1.5%H was blocked in the V funnel and in the L-box with 2 bars. All the ber reinforced concrete mixes were blocked in the L-box test with three bars. This occurs due to the small spacing between the bars that is not common in structures where such ber reinforcement dosages (1.0% and 1.5%) and aspect ratios are used. 7.2. Mechanical behavior Fig. 5 presents one representative curve obtained in the direct tensile test for each material investigated in the present work. The results of evaluation of all curves are given in Table 3. A small branch of hardening was only observed for the mixtures C1.5% and

Viscosity (Pa s) II 0.90 0.90 0.86 BL 0.50 III 0.85 BL BL BL BL 187 230 157 341 156

Shear yield stress (Pa)

183 111 178 220 131

BL = congestion of ow section. I, II, III = number of steel bars.

D.A.S. Rambo et al. / Materials and Design 54 (2014) 3242

37

Fig. 5. Inuence of the steel ber reinforcement ratio in the direct tensile behavior of the self-consolidating concretes.

C1.5%H. The composites reinforced with 1.0% of steel ber presented a strain softening behavior. The tensile curves were depicted in 3 distinct phases. Phase I corresponds to the elasticlinear range where both matrix and the ber behave linearly. Due to low volume fraction of bers (61.5%) the stiffness of the composite is dominated by matrix properties and this zone is limited to strain measures of up to 0.016%. This was conrmed by the elastic modulus value of approximately 30 MPa computed for all concretes. Phase II is marked by the bend over point (BOP) which corresponds to the formation of matrix cracking. Phase III was characterized by a strain softening behavior for composites reinforced with 1.0% and strain hardening for 1.5% of reinforcement. The addition of the 12 mm bers did not affect the ultimate tensile strength (UTS) and rBOP. The maximum UTS of 4.46 MPa was ob-

served for the C1.5%. The strain hardening was marked by the formation of only one visible cracking. This crack was able to widen from 0.1% to 0.3% of deformation where a strain softening behavior started. Fig. 6a presents one representative curve obtained in the fourpoint bending tests for each reinforcement ratio and hybridization investigated in the present work. The results of evaluation of all curves are given in Table 4. All ber reinforced concretes presented a deection hardening behavior with a single cracking formation surrounded by ramications. The curves can be divided in four main phases. The phase I corresponds to the linear elastic region where both matrix and ber behave linearly. In this region both compression and tension strains are compatible. The limit of proportionality (LOP) delimit the phase II. The LOP increased with the addition of ber reinforcement from 7.58 MPa (unreinforced matrix) to 10.70 MPa (C1.0%H) (see Fig. 6b). This can be explained by the action of the 12 mm bers that bridges the micro-cracks. When further increasing the reinforcement ratio to 1.5% it was noticed a tendency of decreasing in LOP probably due to an increase in the matrix porosity. The post LOP range (phase III) is characterized by a deection hardening behavior but with the formation of one single crack. Phase IV is characterized by the strain softening response due to the widening of the crack. The highest exural behavior was observed for the composite C1.5% which presented strength of 17.66 MPa. This represents an increase of 2.3 times in comparison to the unreinforced matrix. It seems that the ber hybridization did not inuence the exural strength. The toughness computed as the total area under the load displacement curve showed to be sensitive to the hybridization of bers. At low levels of displacement (i.e. 0.2 mm) the hybrid composites (C1.0%H and C1.5%H) showed a higher efciency on energy absorption capacity in comparison with non-hybrid systems (C1.0% and C1.5%). For example, the concrete C1.0%H presented a toughness of 6.72 J while the C1.0% 5.61 J. This can be attributed to the ability of the

Table 3 Average results of the direct tensile test. Standard deviation values are presented in parentheses. Mixtures Tensile strength, deformation and modulus Tensile toughness

rBO (MPa)
Matrix C1.0% C1.0%H C1.5% C1.5%H 2.19 3.75 3.74 3.89 3.09 (0.57) (0.49) (1.31) (0.59) (0.25)

eBOP (%)
0.0097 0.0163 0.0162 0.0113 0.0142 (0.0022) (0.0087) (0.0048) (0.0016) (0.0024)

Et (GPa) 28.93 30.40 30.53 32.07 31.11 (1.56) (1.33) (2.43) (3.28) (3.53)

rUTS (MPa)
3.89 3.85 4.46 3.74 (0.59) (1.31) (0.36) (0.22)

eUTS (%)
0.0888 (0.12) 0.0650 (0.071) 0.174 (0.092) 0.257 (0.040)

T0.5mm (J) 5.7 5.8 6.3 5.7 (1.1) (2.1) (1.4) (0.5)

T1.0mm (J) 10.4 (2.9) 9.9 (4.1) 11.1 (2.8) 12.3 (1.1)

T2.0mm (J) 16.9 15.0 16.8 19.2 (5.7) (5.3) (4.7) (1.9)

T3.0mm (J) 20.5 18.5 21.1 23.8 (7.5) (6.4) (6.1) (2.2)

T4.0mm (J) 22.5 20.3 25.0 26.1 (8.1) (7.5) (9.2) (4.5)

Fig. 6. Inuence of the steel ber reinforcement in the mechanical behavior of the self-consolidating concrete in the materials scale: (a) stress vs. displacement curves and (b) zoom in (a) up to 0.4 mm of displacement.

38

D.A.S. Rambo et al. / Materials and Design 54 (2014) 3242

small bers (SF1) to delay micro-cracks in the concrete matrix. However, for high displacement levels, the opposite occurs, and the hybrid reinforced mixtures showed a more pronounced softening branch and lower toughness values than that of the non-hybrid reinforced mixtures. Kim et al. [14] performed a similar work and their results for bending tests of FRCs reinforced with a mix of hooked end and micro bers resulted in an increase of bending strength and toughness. It should be noted that they have used a higher ber dosage and a ne mortar matrix which resulted in higher efciency of bermatrix stress transfer. The data acquisition of strains from the compression and tension faces during the bending test allowed the computation of the neutral axis, NA (normalized with specimen depth, h) as seen in Fig. 7a and b. During the bending test the section remains constant at the centroidal location (0.5 h) until the rst crack formation which happens at a deection of about 0.050.06 mm. A sudden change of the NA to 0.3 h happens for the non-hybrid system and is observed after the formation of the rst crack. This NA continues to reduce until the displacement of 0.1 mm to a value of approximately 0.230.25 h. This value indicates that assuming plane sections remaining plane, 7678% of the sample cross section is in tension while the rest is in compression. For the hybrid systems (in special the C1.5%H mixture) a gradual decrease in NA is observed as a result of stress redistribution due to crack formation. This value reaches 0.3 h at a displacement of 0.1 mm indicating that the ber hybridization was effective in mitigating the microcrack development as previously indicated by the toughness results calculated at low displacements. The bending and tensile responses are compared in Fig. 8. The inset plot shows the relationship between LOP vs. BOP and MOR (modulus of rupture) vs. UTS. It can be seen that under bending, loads associated with the formation of the rst crack occur at stress levels 2.5 times higher as those observed for the direct tension tests. At these stress levels a similar response was observed for all composites. Although a greater variability was noticed at ultimate stress states, it can be concluded that values reported for MOR is approximately four times greater than that of the UTS. The localized strain measured from the electrical resistance gage positioned at the tensile surface is also shown in Fig. 8. The results show that strains corresponding to rst crack formation (from 0.01% to 0.016%) are in the same range for bending and tensile loads. It can also be seen that the strain response agrees well with s results until the crack appearance (LOP). the LVDT Fig. 9a shows one representative curve obtained in the round panel test for each studied material. Table 5 presents the results and calculations related to the tests. In the same way as that observed for the exural beam tests the round panel test results can be divided in four phases. Phase I corresponds to the linear elastic region where the compressive and tensile strains are compatible. The formation and propagation of the rst crack delimits the second phase. During phase III all FRCs exhibited a strain deection behavior with the formation of three major axial cracks. Several multiple ne cracks (see the inset in Fig. 9a) were also noticed. The localization and widening of one major crack leads the FRC to failure which is followed by a strain softening behavior. While the unreinforced concrete presented the formation of only three major radial cracks all steel ber reinforced concretes also presented a formation of multiple micro-cracks. A similar behavior was also reported by Bernard when testing FRC round panel specimens [20,22]. In a different manner than that observed for the exural beam tests the ber hybridization in the panels were benecial in increasing the rst crack and ultimate loads for both studied reinforcement ratios. For example, for the reinforcement ratio of 1% the rst crack and ultimate load increased from 29.76 to 41.16 MPa and from 39.66 to 45.50 MPa, respectively. When looking at Fig. 9b it can be seen that for low displacements (i.e.

T4.0mm (J) T3.0mm (J) T2.0mm (J) T1.0mm (J) T0.2mm (J) dMOR (mm) MOR (MPa) Pu (kN) dLOP (mm) Flexural strength and displacement Table 4 Average results of the exural beam test. Standard deviation values are presented in parentheses. Flexural Toughness

rLOP (MPa)
PLOP (kN)

Mixtures

Matrix C1.0% C1.0%H C1.5% C1.5%H

25.27 29.60 36.02 34.95 31.42

(2.51) (2.78) (3.31) (2.82) (3.29)

7.58 8.88 10.70 10.48 9.42

(0.75) (0.84) (1.01) (0.84) (0.98)

0.049 0.059 0.053 0.056 0.049

(0.007) (0.002) (0.007) (0.007) (0.007)

43.08 44.36 58.89 51.46

(11.06) (2.97) (4.49) (13.35)

12.93 13.30 17.66 15.43

(3.32) (0.89) (1.34) (4.00)

0.83 0.53 0.72 0.67

(0.16) (0.22) (0.07) (0.05)

5.61 6.72 7.16 7.33

(0.72) (0.56) (0.15) (0.26)

38.57 40.82 49.87 45.24

(2.56) (3.73) (2.11) (12.05)

71.25 71.92 97.67 87.53

(4.53) (6.56) (5.41) (24.13)

95.24 (7.20) 91.37 (8.90) 133.60 (6.51) 117.77 (34.65)

112.9 (10.23) 105.28 (11.02) 163.21 (6.95) 140.68 (43.71)

D.A.S. Rambo et al. / Materials and Design 54 (2014) 3242

39

Fig. 7. Neutral axis computation: (a) non-hybrid and (b) hybrid systems.

Fig. 8. Comparative of exural and tensile response. The curve in red shows the local strain measurement of the electrical gage positioned in the tensile surface of the exural specimen.

<2.5 mm) the ber reinforcement plays a major role in the mechanical behavior of the FRCs. For displacements at phase I (in the linear elastic region) it was observed that the hybridization in the C1.5%H increased the load bearing capacity showing the

capacity of the short 12 mm bers of bridging the micro-cracks and increasing the serviceability limit of the structure. The toughness values computed at 5 and 10 mm also conrmed this mechanism. At these displacements the energy absorption capacity was superior for hybrid concretes than that observed by the concretes reinforced only with the 35 mm hooked end bers (refer to Table 5). The crack opening was measured during the round panel and beam exural test for the concrete specimens reinforced with 1.5% of the hooked end bers. Fig. 10a shows one representative curve along with the crack opening measurements of the three major cracks for the round panel and of the single crack for the exural beam. Fig. 10b presents a detail plot of the crack opening until a central displacement of 6 mm. The crack opening development for the round panel can be divided in three phases. The rst phase corresponds to the linear elastic range. It can be seen that the three used LVDTs only measured the matrix deformation which was linear elastic up to a deformation of 1.2 mm. At this point the several radial cracks start to form. At phase II, from the central displacement of 1.2 mm (0.73Pu) until 2.0 mm (0.89Pu), the originated cracks propagated across the radius of the round panel specimen. At the end of phase II (0.89Pu) a crack opening of 0.2 mm was measured. It should be noticed that codes and specications for design of most reinforced concrete structures establish that the maximum allowable crack widths should range from 0.15 to 0.38 mm. Phase III initiates at the displacement of 2 mm. From this point and above the three cracks start a widening process

Fig. 9. Inuence of steel ber reinforcement on the round panel test: (a) load vs. displacement results and typical crack pattern at 40 mm for C1.5% and (b) a zoom in the results up to 2.5 mm of displacement.

40

D.A.S. Rambo et al. / Materials and Design 54 (2014) 3242

Table 5 Average results of round panel test. Standard deviation values are presented in parentheses. Mixtures First crack strength P (kN) Matrix C1.0% C1.0%H C1.5% C1.5%H 30.71 29.76 41.16 40.25 46.36 (3.72) (2.66) (0.61) (0.58) (2.62) d (mm) 0.72 0.95 1.46 1.23 1.16 (0.05) (0.08) (0.25) (0.16) (0.51) Flexural strength P (kN) 39.66 45.50 53.82 63.38 (3.86) (1.97) (4.76) (5.96) d (mm) 3.42 3.26 5.30 5.23 (0.26) (0.54) (0.56) (0.65) Toughness T5.0mm (J) 164 189 207 255 (13.1) (13.2) (17.7) (30.8) T10mm (J) 336 368 481 549 (40.3) (15.1) (30.5) (64.8) T20mm (J) 519 547 841 612 (111) (25.9) (87.9) (105) T30mm (J) 781 (150) 627 (98.5) 1021 (144) 1207 (250) T40mm (J) 870 (280) 649 (170) 1077 (320) 1303 (325)

Fig. 10. Crack opening measurements: (a) correlation of the crack opening measurements using three LVDTs and the load vs. displacement curve obtained in the round panel and exural beam test for the C1.5%; (b) Zoom in the crack opening measurements up to 6 mm.

which for this specic specimen followed different rates. It was observed a mean crack opening of 1.5 mm at the displacement corresponding to the maximum load. During the strain softening region the cracks continued to open until the nal measured central displacement of 40 mm. The nal measured cracking opening was 12 mm for LVDT 1 and 27 mm for LVDTs 2 and 3. In the exural beam test the crack initiated at the mid-span displacement of 0.056 mm (0.6Pu). Results obtained from LVDT measurement and image analysis showed similar values. The crack opening was linear up to 2.0 mm (end of the strain softening

phase). Different from the round panel test it was not possible to identify the point at which the crack crossed the specimen. At the modulus of rupture point a crack width of 1.5 mm was measured (same value as reported for the round panel). It was observed that for the same load ratios the crack width measured in the exural beam test was higher. For example the crack width computed at 0.89Pu was found to be 0.36 mm in comparison to 0.2 mm obtained at the same load ratio for the round panel. From the results reported above it can be concluded that the 35 mm hooked end ber is effective in bridging the major and localized cracks leading to a more stable crack propagation whereas the 12 mm bers can limit the initiation and propagation of microcracks. This behavior is only visible in the structural scale testing performed in the round panels. 7.3. Finite element analysis A nonlinear simulation of the mechanical behavior of round panel samples was performed in the nite element software ABAQUS using the concrete damaged plasticity model [31]. The global element was created according with the dimension of the real sample. The round panel sample was discretized using 10404 shell elements with reduced integration (S4R). These type of elements are usually employed in structures in which the thickness is approximately 1/10 the overall size. The three supports were xed in three mesh nodes spaced 120 C radially in accordance with the test geometry. Each support was modeled as hinges, unable to move in the load direction. The used constitutive material relationship is shown in Fig. 11. The experimental results were obtained in the present work and in the authors previous work [32]. In order to determine the loaddisplacement curves, incremental displacements were imposed at the central node and the reaction force was measured. Fig. 12 compares the experimental results with the simulation and shows the stress and displacement

Fig. 11. Uniaxial compressive and tensile stress strain relationship used in the nite element simulation.

D.A.S. Rambo et al. / Materials and Design 54 (2014) 3242

41

Fig. 12. Comparison of nite element analysis and experimental tests for the C1.5% and the stress and displacement elds immediately after the initiation of the rst crack.

elds of the FEA. The simulation results obtained from the nite element analysis showed a good correlation with the experimental results, as the model accurately predicted the load vs. displacement curve. The simulation at rst crack reveals a maximum stress of 3.69 MPa in the central part as well as over the three major cracks. These agree well with the mean rst crack stress of 3.89 MPa found in the direct tension tests. Central displacement and load corresponding to rst crack were smaller than the mean values found in the experiments indicating that the modeled boundary conditions were stiffer than the real tests. The use of round panel tests to validate the nite element model is important for further use in structural design. More research is needed to verify the model with others bers in order to be able to use this knowledge for the material and structural design purposes. The use of the nite element analysis can be a powerful and unique tool to design structural elements reinforced with bers thus saving enormous time in laboratory.

and ultimate loads. A comparison of crack width performed in the materials (four point bending) and structural scale (round panel) showed that the widths related to same loading ratio are smaller in the panel tests. Furthermore, crack widths computed at up to 90% of the ultimate load were in the range allowed in most of the design codes for reinforced concrete. Finite element analysis was performed for the round panel tests. The results showed a good correlation with the experimental tests. Stresses at rst crack were also similar in experimental FE simulation and experiments. It seems that the boundary conditions in the model are somehow stiffer than the ones used in the experiments leading to lower displacements at rst crack.

References
[1] Bentur A, Mindess S. Fibre reinforced cementitious composites. England: Elsevier Applied Science; 1990. [2] Balaguru PN, Shah SP. Fiber-reinforced cement composites. New York: McGraw-Hill; 1992. [3] Holschemacher K, Mueller T, Ribakov Y. Effect of steel bres on mechanical properties of high-strength concrete. Mater Des 2010;31:260415. [4] Naaman AE. High performance ber reinforced cement composites. In: Proceedings of the IABSE symposium on concrete structures for the future. Paris, France; September 1987. p. 3716. [5] Naaman AE, Reinhardt HW. Characterization of high performance ber reinforced cement composites HPFRCC. In: Naaman AE, Reinhardt HW, editors. Proceedings of the 2nd international RILEM workshop on high performance ber reinforced cement composites (HPFRCC2). London: E & FN Spon; 1996. p. 124. [6] Toledo R D, Koenders EAB, Formagini S, Fairbairn EMR. Performance assessment of ultra high performance ber reinforced cementitious composites in view of sustainability. Mater Des 2012;36:8808. [7] Rossi P. Ultra-high performance bre reinforced concretes (UHPFRC): an overview. In: Rossi P, Chanvillard G, editors. 5th International RILEM symposium on bre-reinforced concretes (FRC) BEFIB. France: RILEM Publications; 2000. p. 87100. [8] Buitelaar P. Ultra high performance concrete: developments and applications during 25 years. In: International symposium on UHPC. Germany; 2004. [9] Sukontasukkul P, Mindess P, Banthia N. Penetration resistance of hybrid bre reinforced concrete under low velocity impact loading. In: Annual Conference of the Canadian Society for Civil Engineering. Montreal; 2002. [10] Di Prisco M, Dozio D, Colombo M. On the bearing capacity of FRC structures is the material characteristic value the right choice? In: Toledo Filho RD, Silva FA, Koenders E, Fairbairn EMR, editors. Proceedings of the 2nd international RILEM conference on strain hardening cementitious composites (SHCC2-Rio), Rio de Janeiro; 2011. p. 27987. [11] Kuder KG, Ozyurt N, Mu EB, Shah SP. Rheology of ber-reinforced cementitious materials. Cem Concr Res 2007;37:1919. [12] Chen Y, Qiao P. Crack growth resistance of hybrid ber-reinforced cement matrix composites. J Aeros Eng 2011;24:15461. [13] Kim DJ, Naaman AE, El-Tawil S. Comparative exural behavior of four ber reinforced cementitious composites. Cement Concr Compos 2008;30:91728.

8. Conclusions The following conclusions can be drawn from the present work on mechanical behavior of self-compacting concrete: The results obtained from the parallel plate rheometer showed that while the unreinforced concrete behaves linearly following the Bingham model the FRC curves are non-linear following HerschelBulkley. The hybridization of bers increased the viscosity of the mix. Also for the hybrid system the increase in ber volume fraction lead to an increase in the shear yield strength. The direct tensile tests did not show any improvements for ber hybridization. Nevertheless for concretes reinforced with 1.5% of bers a small strain hardening branch has been observed. The exural beam tests showed that all concretes have a deection hardening behavior but with the formation of one macro crack. The ber hybridization was effective to increase the toughness for central displacements up to 0.2 mm. The computation of the neutral axis by means of electrical strain gages showed a gradual decrease when hybrid bers are used in comparison to a sudden drop at 0.3 h observed for the non-hybrid systems. Therefore, the ber hybridization was effective in mitigating the micro-crack formation during bending tests. All FRCs used in the round panel tests were characterized by a formation of three major cracks and several micro-cracks. The ber hybridization was very effective in limiting the initiation and propagation of micro-crack thus increasing the rst-crack

42

D.A.S. Rambo et al. / Materials and Design 54 (2014) 3242 [22] Bernard ES. Behaviour of round steel bre reinforced concrete panels under point loads. Mater Struct 1999;33:1818. [23] De Larrard F. Concrete mixture proportioning: a scientic approach. London: E&FN SPON; 1999. [24] Sedran T. Rhologie et rhomtrie des btons: application aux btons autonivelants. Doctoral dissertation. Ecole Nationale des Ponts et Chausses; 1999. 484p. [25] de Larrard F, Hu C, Sedran T, Szitkar JC, Joly M, Claux F, et al. A new rheometer for soft-to-uid fresh concrete. ACI 1997:23443. [26] NBR 15823-5. Concreto Autoadensvel. Parte 5: Determinao da viscosidade Mtodo do funil V. Rio de janeiro: Associao brasileira de normas tcnicas; 2010. [27] NBR 15823-4. Concreto Autoadensvel. Parte 4: Determinao da abilidade passante Mtodo da caixa L. Rio de janeiro: Associao brasileira de normas tcnicas; 2010. [28] NBR 15823-2. Concreto Autoadensvel. Parte 2: Determinao do espalhamento e do tempo de escoamento Mtodo do cone de Abrams. Rio de janeiro: Associao brasileira de normas tcnicas; 2010. [29] De Larrard F, Ferraris CF, Sedran T. Fresh concrete: a HerschelBulkley material. Mater Struct 1998;31:4948. [30] EFNARC. The European guidelines for self-compacting concrete. Specication, production and use. In: European Project Group; 2005. 63p. [31] ABAQUS Version 6.11 nite element program. Analysis Users Manual. [32] Rambo DAS. Concretos autoadensveis reforados com bras de ao hbridas: aspectos materiais e estruturais. MSc dissertation. Rio de Janeiro, Brazil: COPPE/UFRJ; 2012. 185p.

[14] Kim DJ, Park SH, Ryu GS, Koh KT. Comparative exural behavior of hybrid ultra high performance ber reinforced concrete with different macro bers. Constr Build Mater 2011;25:414455. [15] Akcay B, Tasdemir MA. Mechanical behaviour and bre dispersion of hybrid steel bre reinforced self-compacting concrete. Constr Build Mater 2012;28:28793. [16] Rossi P, Acker P, Malier Y. Effect of steel bers at two stages: the material and the structure. Mater Struct 1987;20:4369. [17] Boulay C, Rossi P, Tailhan JL. Uniaxial tensile test on a new cement composite having a hardening behavior. In: 6th Rilem symposium on bre reinforced concretes (FRC), BEFIB. Varenna; 2004. p. 618. [18] Parant E. Mcanismes dendommagement et comportements mcaniques dun composite cimentaire br multi-chelles sous sollicitations svres: fatigue, choc, corrosion. Doctoral dissertation. France: Ecole Nationale de Ponts et Chausses; 2003. [19] Minelli F, Plizzari GA. Fiber reinforced concrete characterization: round panel vs. beam tests toward a harmonization. In: Balzs GL, Nehme SG, editors. Proceedings of the 3rd Central European Congress on Concrete Engineering CCC 2007, Visegrd: Publishing Company of Budapest University of Technology and Economics; 2007. p. 21320. [20] Bernard ES. Design performance requirements for bre reinforced shotcrete using ASTM C-1550. In: Proceedings of the second international conference on engineering development in shotcrete. Cairns: Taylor & Francis; 2004. [21] Destree X, Mandl J. Steel bre only reinforced concrete in free suspended elevated slabs: case studies, design assisted by testing route, comparison to the latest SFRC standard documents. In: Walraven, Stoelhorst, editors. Tailor Made Concrete Structures. London: Taylor & Francis, Group; 2008. p. 43743.