The size effect of quasibrittle materials, such as concrete, is an important, and well-investigated area of interest. Since the specimen size of laboratory tests is always much smaller than that of real structures, it is essential to understand and handle the phenomena of size effect in order to gain reliable designing parameters. In the case of plain and reinforced concrete a great deal of research can be found in the literature in connection to the size effect (e.g. [1,2,3]). Fig. 1 shows the most widely used plot to present special aspects of the size effect of quasibrittle materials based on Bazant and Planas [1].
Size effect on the strength of concrete (adapted from Bazant and Planas [1])
The most accepted and simplest size effect law for concrete (Eq. 1) based on nonlinear fracture mechanics is derived by Bazant [2]. Where
Although the use of fibre-reinforced concrete (FRC) is more and more popular, there are exiguous investigations on the size effect of FRC [4,5,6,7,8,9]. In fact, most of the existing model codes and guidelines of FRC constitutive laws (e.g. the
According to the introduction, contradictory statements of the size-effect of FRC can be found in the literature, whether it should be taken into account during the design process, or it can be neglected. Any of these statements are proved satisfactorily. The reason for this is that measuring experimentally the effect of the specimen size on the mechanical behaviour of FRC precisely is not simple, as the dispersion of the experimental results is notoriously huge. For an adequate laboratory series of measurements, more specimens would be required than in the case of plain concrete, which, moreover, differs from the standard formwork. Because of the huge dispersion and the relatively few specimens, the commonly used evaluating methods (comparing the experimental force-crack opening or displacement curves) can lead to different conclusions even in the case of a repeated experiment under the same conditions. That is the reason for the lack in the literature of a common position in the case of the size effect of FRC.
Based on the observations of Vlietstra [8] it is probable that the size effect is the characteristic only of the concrete material, and the effect of the fibres can be handled separately. However, there is no evidence of this assumption until now. The commonly used analytical models handle FRC as a homogeneous material during the evaluation of experimental results and during the designing process. So, the material parameters of these models (e.g., [15,16,17,18]) depend on the combined behaviour of the concrete matrix and the fibres, so all of these parameters are necessarily dependent on the size effect. That is the reason, why these parameters are not applicable for verifying the above-mentioned assumption and for determining the share of the concrete matrix and the fibres in the size effect. A deeper understanding of the material behaviour is needed. In the following, authors examine the size effect of FRC separating the effect of the concrete matrix and the fibre reinforcement with the help of the semi-discrete analytical (SDA) beam model of Tóth and Pluzsik [19]. The SDA beam model examines FRC material at the meso level: the concrete material is handled as a homogeneous continuum, approximating its behaviour on the basis of fracture mechanics, and then the effect of the embedded fibres are built “one by one” into the model, handling them as discrete elements. Consequently, the SDA model contains new material parameters, which handle the effect of the concrete matrix and the fibres in the inhomogeneous cross-section separately. The dispersion of these parameters is smaller than that of the commonly used parameters of analytical models, so with their help, a relevant conclusion can be drawn in case of a standard number of experimental samples of concrete. The aim of this research is to prove that in the case of softening material the effect of the specimen size influences only the behaviour of concrete material, and so the peak strength, while the residual loadbearing capacity (which is mainly affected by the bridging effect of fibres) and the crack propagation are independent of the structural size.
Based on the RILEM recommendation for the experimental procedure [20] twenty one test specimens were prepared. Table 1 contains the comparison of the performed experiment and the recommendation of RILEM (where
Comparison of the performed experiment and the recommendation of RILEM [20]
Proportions | RILEM recommendation | Performed experiment |
---|---|---|
>2.5 | 3.33 | |
0.15< |
0.083 | |
notch width | <0.5 |
0.125 |
>3 |
9.4 |
|
<5 |
4.7 |
|
>10 |
18.75 |
|
≥4 | 4 |
The test specimens were cast in three different sizes according to Table 2 and Fig. 2. Two plain concrete (PC) and five fibre reinforced concrete (FRC) beams were prepared in each size. The beam width was uniformly 150 mm, and all the beams had a bottom notch in the middle cross-section, with a depth of
Specimen dimensions and experimental layout
Dimensions of the test specimens
Specimen ID | Beam depth ( |
Span ( |
Notch depth ( |
---|---|---|---|
FRC_S | 75 | 250 | 6.25 |
FRC_M | 150 | 500 | 12.5 |
FRC_L | 300 | 1000 | 25 |
Proportions of concrete matrix
Component | Dosage [kg/m3] |
---|---|
Cement | 350 |
Water | 192 |
Aggregate 0/4 | 631 |
Aggregate 4/8 | 685 |
Aggregate 8/16 | 487 |
Superplasticizer | 2 |
The concrete beams were examined by a three-point bending test according to the recommendations of RILEM TC 162-TDF [21] (Fig. 2). The measured values were the loading force (
Measurement of the crack opening
To make the results comparable, the above detailed CMOD measuring method was performed in the case of all beam sizes. However, in the case of the mediumsized beams (FRC_M), the CMOD gauge of the testing machine also measured the crack opening, so the two different methods could be compared. For example, in the case of specimen FRC_M_03 (Fig. 4), the CMOD was measured by the CMOD meter of the testing machine and the mechanical dial gauge crack width monitor, too. Since for the manual measurement (green line), the testing machine was stopped at regular intervals, in the experimental
Comparison of manual and machine CMOD measurement in the case of FRC_M_03
Before the first crack occurs and in the descending part of the
After fracturing the number of fibres crossing the cracked cross-section was manually determined, divided the cross-sections into five horizontal zones (the bottom cut is the sixth zone) (Fig. 5). Table 5 contains the fibre distributions of the beams distinguishing between the two different failure modes. The appearance of the broken and pulled-out fibres is significantly different, the end of the pulled-out fibres remains intact, while the broken fibres disinte-grate into bundles of fibres (Fig. 6).
Cracked cross-sections divided into five zones
Appearance of pulled-out and broken fibres
Fig. 7 shows the experimental F-CMOD results of the three different sizes in the case of PC and FRC beams.
Experimental
The evaluation of the experimental results is performed with the help of the semi-discrete analytical (SDA) beam model of Tóth and Pluzsik [19]. The SDA model is not handling FRC as a homogeneous material, but as an inhomogeneous one which consists of fibres, concrete matrix and the bond between them. The effect of the fibres is modelled separately from the concrete matrix and is neglected before the first crack. The
Experimental and theoretical τ-s and F-u curves of the pulled-out fibres (Reprinted from Tóth and Pluzsik [19])
To take rupture failure mode into account a rupture condition is determined. The disruption of the fibres is related to the critical force (
The model calculates if the force of the fibres of a given displacement with a given embedment length, is within the boundary condition or not. If the value of the fibre force is higher than
The use of the two different failure modes in the SDA model was verified based on experimental results: the ratio of broken fibres determined by the model was adjusted to the manually counted experimental results.
All the data defining the fibre (
The softening curve of
Softening curve of plain concrete (Reprinted from Tóth and Pluzsik [23])
The model first calculates the moment (
Stress distribution and distances along the cross-section (Reprinted from Tóth and Pluzsik [19])
A thorough description of the model can be found in Tóth and Pluzsik [19, 24]. As input parameters the model contains the geometrical and material properties of the fibre and the concrete. Besides these given parameters the model has two free parameters
In the following, the size-dependent properties of FRC beams will be examined with the help of the SDA model. The SDA model separates the behaviour of the concrete material (characterized by
The experimental results of the plain concrete beams are shown in Fig. 6. It can be seen that the loadbearing capacity increases with the size of the beam. However, the nominal stress (
Nominal stresses and q values of PC specimens
Specimen ID | Beam depth ( |
Nominal stress ( |
Average of |
Deviation of |
Slope of the softening curve ( |
---|---|---|---|---|---|
PC_S_06 | 75 | 2.73 | 2.9 | −5.9 | 15 |
PC_S_07 | 75 | 3.07 | 5.9 | ||
PC_M_06 | 150 | 2.84 | 2.81 | 1.2 | 15 |
PC_M_07 | 150 | 2.77 | −1.2 | ||
PC_L_06 | 300 | 2.50 | 2.48 | 1 | 15 |
PC_L_07 | 300 | 2.45 | −1 |
The resulting nominal forces were compared to Bazant's size effect law (Eq. 1) (Bazant 1984).
Size effect of Bazant compared to experimental results of PC
After verifying Bazant's size effect law for the PC results, the material parameter of the SDA model was determined for all the PC specimens. The slope of the softening curve (
Experimental and theoretical F-CMOD curves of PC_M (
In the case of the test results of the FRC beams the same phenomenon can be observed as with the plain concrete: the nominal stress (
Nominal stresses and
Specimen ID | Nominal stress ( |
Average of |
Average of |
Deviation of |
Pulled-out fibres [pcs] | Broken fibres [pcs] | Rate of pulled-out fibres [%] | |
---|---|---|---|---|---|---|---|---|
FRC_S_01 | 3.29 | 3.47 | 3.4 | 3.16 | 8% | 47 | 7 | 87% |
FRC_S_02 | 3.47 | 2.4 | −24% | 51 | 10 | 84% | ||
FRC_S_03 | 3.34 | 3.0 | −5% | 52 | 10 | 84% | ||
FRC_S_04 | 4.25 | 3.4 | 8% | 30 | 22 | 58% | ||
FRC_S_05 | 3 | 3.6 | 14% | 39 | 13 | 75% | ||
FRC_M_01 | 3.43 | 3.10 | 3.6 | 3.64 | −1% | 82 | 26 | 76% |
FRC_M_02 | 2.67 | 3.6 | −1% | 87 | 26 | 77% | ||
FRC_M_03 | 2.94 | 3.4 | −7% | 70 | 35 | 67% | ||
FRC_M_04 | 3.18 | 3.6 | −1% | 83 | 42 | 66% | ||
FRC_M_05 | 3.3 | 4.0 | 10% | 108 | 32 | 77% | ||
FRC_L_01 | 2.58 | 2.71 | 2.6 | 3.52 | −26% | 179 | 15 | 92% |
FRC_L_02 | 2.83 | 3.6 | −2% | 179 | 20 | 90% | ||
FRC_L_03 | 2.88 | 4.4 | 25% | 119 | 35 | 77% | ||
FRC_L_04 | 2.47 | 3.8 | 8% | 160 | 36 | 82% | ||
FRC_L_05 | 2.79 | 3.2 | −9% | 159 | 24 | 87% |
Size effect of Bazant compared to experimental results of FRC
The FRC beams were also modelled by the SDA model, using the real fibre distribution of the specimens. Parameter
The calculated and measured F-CMOD curves fitted well in all cases, as shown in Fig. 14, for example for FRC_S_04 and FRC_M_05.
Experimental and theoretical
Fig. 15 shows the residual nominal strength (
Residual nominal strength (
It has to be noted, that in the case of this experiment, the fibre length was smaller than the critical fibre length, so the fibres in the majority were pulled out (Column 8 of Table 5).
It has to be also noted, that in this investigation concrete reinforced with synthetic fibres was used which has softening material behaviour. In the case of hardening materials (e.g. steel fibre reinforced concrete) the peak load is affected by the fibres also, therefore further research is needed to understand the size effect behaviour of them.
Finally, it is worth noting, that all the beams were prepared under laboratory conditions, where the mixing of fibres into the concrete is carefully controlled. However, in the case of larger structures the clumping of the fibres (because of unequal mixing) can cause serious problems. Therefore, during the preparation of the fibre reinforced concrete structures special attention should be paid to the appropriate mixing of fibres, ensuring the even fibre distribution is taken into account in the design process.
The size effect of synthetic fibre-reinforced concrete beams was investigated with the help of the SDA model of Tóth and Pluzsik [19]. The SDA model handles separately the concrete material, the fibres and the bond between them. Investigation of the FRC beams by inverse analysis has shown, that the size effect significantly influences the nominal stress at the peak load due to the softening concrete material, but does not affect the residual loadbearing capacity which is influenced by the connection between the concrete matrix and the fibres (namely the pull-out behaviour of the fibres, represented by
Before the FRC structure cracks the effect of synthetic fibres is negligible (for safety), so the size effect of concrete can be handled as in the case of plain concrete, which is a well-investigated field of interest [e.g. 1–3]. After the first crack occurs, the residual load-bearing capacity of the FRC beam is mainly given by the bridging effect of fibres, and the effect of the concrete can be negligible. Since the pullout behaviour of the fibres is a local phenomenon, it is not affected by the size effect (parameter
According to the above statements, the
It was also proven by the evaluation of experimental results that the two parameters of the SDA model (