Mix Design and Properties of Recycled Aggregate Concretes: Applicability of Eurocode 2

Abstract

This work is devoted to the study of fresh and hardened properties of concrete containing recycled gravel. Four formulations were studied, the concrete of reference and three concretes containing recycled gravel with 30, 65 and 100 % replacement ratios. All materials were formulated on the basis of S4 class of flowability and a target C35 class of compressive strength according to the standard EN 206-1. The paper first presents the mix design method which was based on the optimization of cementitious paste and granular skeleton, then discusses experimental results. The results show that the elastic modulus and the tensile strength decrease while the peak strain in compression increases. Correlation with the water porosity is also established. The validity of analytical expressions proposed by Eurocode 2 is also discussed. The obtained results, together with results from the literature, show that these relationships do not predict adequately the mechanical properties as well as the stress–strain curve of tested materials. New expressions were established to predict the elastic modulus and the peak strain from the compressive strength of natural concrete. It was found that the proposed relationship E–f_c is applicable for any type of concrete while the effect of substitution has to be introduced into the stress–strain (ε_c1–f_c) relationship for recycled aggregate concrete. For the full stress–strain curve, the model of Carreira and Chu seems more adequate.

1 Introduction

Aggregates consumption does not cease to grow in France. According to the UNPG (French national union of aggregates producers) and the UNICEM (French national union of industries of careers and building materials) aggregate production is estimated at 431 million tons in 2008, of which 79 % is used in civil engineering field and 21 % for building industry. In addition, 5 % of this amount is produced by recycling demolition wastes. Although this percentage remains low, recycling helps to limit the environmental impact by limiting the exploitation of natural resources. These socio-economic issues are the driving forces promoting the recycled aggregates in concrete.

The valorization of recycled aggregates in concrete is not recent and many studies have shown that material made with recycled aggregates may have mechanical properties similar to those of a conventional concrete mixed with natural aggregates (Etxeberria et al. 2007; Evangelista and de Brito 2007; Li 2008; McNeil and Kang 2013). However, recycled aggregates are characterized by a high water absorption capacity related to the presence of old mortar attached to the surface of aggregates which hinders their wide use (Gomez-Soberon 2002; de Juan and Gutirrez 2009). The water absorption capacity affects both fresh and hardened states properties. At fresh state, the mix design of concrete with recycled aggregates requires an additional quantity of water to obtain a similar workability as a concrete formulated with natural aggregates (Hansen and Boegh 1986). Such a modification may obviously affect the mechanical characteristics of recycled aggregates concrete. Several studies have investigated the microstructure of recycled aggregates concrete and showed that the porosity is modified and increases with the replacement ratio (Gomez-Soberon 2002). It is also acknowledged that the high porosity of recycled concrete leads a reduction of the mechanical strengths (Gomez-Soberon 2002; Kou et al. 2011). Furthermore, several studies have shown however that mechanical properties of concrete made with recycled aggregates depend on other parameters such as the quality of concrete from which recycled aggregates are obtained (Xiao et al. 2005; Casuccio et al. 2008) as well as the replacement ratio (Belén et al. 2011).

The main goal of this work is to determine the properties of recycled aggregate concretes (RAC) at fresh and hardened states depending on replacement ratio. A concrete made with natural aggregate (NAC), designed for control operations, and three RAC with a S4 class of workability and compressive strength levels near to 35 MPa were formulated and tested. The present study also examines the applicability of relationships of Eurocode 2 (EC2) to concretes made from recycled aggregates. These relationships estimate the modulus of elasticity, the peak strain and stress–strain relationship from the simple knowledge of the compressive strength.

2 Materials

2.1 Cement

CEM I CALCIA 52.5 N CE CP2 NF cement in conformity with the standard EN 197-2 was used in all concrete mixes. The chemical and mineralogical compositions calculated by the method of Bogue are presented in Table 1. The density of this cement is 3.11, its Blaine surface is 395 (m²/kg) and its compressive strength after 2 days is 31.7 (MPa).

Table 1 Chemical and mineralogical compositions of the used cement in %.

2.2 Aggregates

The fine aggregates are 0/4 mm silico-calcareos rolled sand. For all mix designs, two size fractions of coarse natural crushed silico-calcareos aggregates were used. The particles size of the first fraction, called G₁, is comprised between 5 and 10 mm while it is comprised between 10 and 20 mm for the second type G₂. Recycled aggregates were delivered in big bags from a retreatment platform of demolition materials. They were sieved in the laboratory into three fractions GR₁ (4/10 mm), GR₂ (10/20 mm) and sand (0/5 mm). In this study only the coarse recycled aggregates were used. They were dried in an oven at 110 ± 5 °C and then stocked in closed containers until the moment of concrete manufacture. The main properties of the natural and recycled aggregates are presented in Table 2 and the grading curves are plotted in Fig. 1. It is seen that the size grading of the coarse natural and recycled gravel was similar with a larger amount of small particles for GR₁.

Table 2 Physical properties of used aggregates.

Fig. 1

Aggregate gradation curves.

Water absorption has been characterized first, according to the standard NF EN 1097-6 at the atmospheric pressure. Dried aggregates were immersed in water during 24 h then dried again in an oven at a temperature of 110 ± 5 °C. It can be noticed in Table 2 that recycled aggregates have a significant higher water absorption capacity and a lower density than natural ones. In spite of the high water absorption capacity of the used RCA, it remains within the range recommended by the design standards (McNeil and Kang 2013; Kang et al. 2014).

The kinetic of water absorption of recycled aggregates GR₁ and GR₂ was also followed and the water uptake was measured by hydrostatic weighting. Aggregates samples were washed first on the sieve of 4 mm and dried in an oven at a temperature of 110 ± 5 °C until mass stabilization. After drying the tested sample was placed between two stainless sieves (of diameter 15 cm and height 7 cm) for hydro-static weighing. The system was hung to a balance of 0.01 g accuracy with a non-elastic wire and the mass variation is continuously recorded. The room temperature is equal to 22 °C. Figure 2 illustrates the measurement system.

Fig. 2

Water absorption measurement system.

The water absorption was calculated according to the following equation:

$${\mathit{\omega }}_a=\frac{M_{wa}-M_{ws}}{M_s}\times 100,$$

(1)

where M_wa is the mass of saturated aggregates in water at time t, M_ws is the mass of dried aggregates in water at t₀, and M_s is the mass of dried aggregates in air. The t₀ indicates the beginning of the test.

Figure 3 shows the evolution of water absorption versus time. At t = 24 h, water absorption of GR₁ and GR₂ are respectively equal to 8.3 and 6.5 %. These results correspond to the values obtained following NF EN 1097-6 standard (cf. Table 2). Results show also that recycled aggregates are saturated after a “long time” more than 24 h, according to several studies (Tam et al. 2008; Djerbi Tegguer 2012). At t = 24 h, water absorption represents only 82 and 85 % of total degree of saturation for GR₁ and GR₂, respectively. Finally, aggregates reach a water absorption of 10 and 7.8 % for GR₁ and GR₂, respectively. At the opposite, for short time relative to mixing time (5 min), the kinetic of absorption is fast.

Fig. 3

Water absorption of recycled aggregates.

2.3 Superplasticizer

The used Superplasticizer is Cimfluid 3002 produced by Axim Italcementi group with a solid content of 30 %. It is a new generation product based on chains of modified polycarboxylate certified in conformity with the standard EN 934-2 and considered as a water reducing admixture.

3 Mix Method and Concrete Proportions

All mixtures were formulated on the basis of the following specifications:

• only coarse natural aggregates are replaced by recycled ones with three volumetric replacement ratios 30, 65 and 100 %;

• the granular skeleton is constituted of a ternary mixture of sand and two gravels G₁ (4/10 mm) and G₂ (10/20 mm);

• at fresh state, all concretes are of S4 workability class where the target slump with the Abrams’s cone is 18 ± 1 cm. According to the standard NF EN 206-1 the slump for a S4 flowability is comprised between 16 and 21 cm;

• at hardened state a compressive strength comprised between 35 and 43 MPa must be guaranteed at the age of 28 days;

• concretes are designated for XF2 class of environmental exposure according to the standard NF EN 206-1, where water to cement ratio (W/C) is lower than or equal 0.5 and the minimum cement content is higher than 300 kg/m³.

A total of four concretes were then produced, a mix with natural aggregates called (NAC) and three concretes with recycled aggregates named RAC30, RAC65 and RAC100. The numbers indicate the rate of substitution. For NAC the cement content is taken equal to 360 kg/m³ according to the standard NF EN 206-1 while for the other mixes this content was modified as will be explained below.

3.1 Optimization of Water to Cement Ratio

The quantity of water for the cement was determined based on the flowability requirement by means of spread tests with the mini flow cone for which dimensions are 8 cm for lower diameter, 7 cm for upper diameter and 4 cm for the height (Fig. 4).

Fig. 4

Water to cement ratio versus relative slump.

For various water/cement ratios the slump d_m and the relative slump Γ_m are computed using the following equation:

$$d_{\text{m}}=\frac{d_1+d_2}{2};{\mathrm{\Gamma }}_{\text{m}}=\frac{d_{\text{m}}^2-d_0^2}{d_0^2},$$

(2)

where d₀ is the lower diameter of the cone, d_m is the average of two wafer diameters d₁ and d₂. In Fig. 4 the relative slump Γ_m is depicted against the water to cement ratio, where a straight line fits the experimental results with a correlation ratio R² = 0.99. The relative slump Γ_m = 5 yields the water content necessary for a flowable paste (El-Hilali 2009).

3.2 Optimization of Solid Skeleton

Granular skeleton was optimized by the method of compaction using vibration. The study started by measuring the packing density of each component, i.e. sand, natural gravels G₁, G₂ and recycled gravels GR₁, GR₂. Binary mixtures of gravel were then tested to determine optimal dosages which give the densest packing. Finally ternary mixtures were tested to optimize the solid skeleton for all mixes.

Packing tests were conducted according to the test method No. 61 of LCPC (Lédée et al. 2004). To determine the packing density of a given granular fraction, a sample is poured in 16 × 32 cm² cylindrical mold then vibrated while applying a pressure of 10 kPa. The packing density is calculated by the equation:

$$C=\frac{H_f}{H_i},$$

(3)

where H _f is the sample height after vibration and H _i the height before vibration. The results for components are summarized in Table 3 where each value is the average of three measurements. From this table it can be noticed that the packing density of recycled aggregates is lower than that of natural aggregates due to the presence of old cement paste.

Table 3 Packing densities of used aggregates.

Optimal proportions of granular mixtures are reported in Table 4 where an increase in the sand content can be observed for recycled aggregates to assess the highest packing. This increase is due mainly to the lower density and lower packing ability of recycled aggregates.

Table 4 Volumetric optimal proportions of mixes.

Values of Table 4 were obtained by conducting packing tests on each mixture with several proportions. The curve representing the variation of packing density as a function of proportions was then plotted. The chosen optimal dosage is the point which corresponds to the maximum of the curve as shown in Fig. 5 for the ternary mixture of sand with recycled aggregates. For concretes RAC30, RAC65 the solid skeleton was first optimized where 30 and 65 % coarse natural aggregates were replaced by recycled aggregates GR₁ and GR₂ and the packing tests were conducted. The results showed that proportions sand to aggregates remain the same as for the mixture sand with recycled aggregates.

Fig. 5

Packing density of ternary mixture S + GR₁ + GR₂.

In order to corroborate the obtained experimental results, the software RENE LCPC was used (Sedran 1999). The software is able, from packing density and the size distribution curves of aggregates, to predict the packing density of a mixture. The results, plotted on Fig. 4, show that theoretical results are in tune with the experimental results.

3.3 Mix Proportions of all Components

Recycled aggregates have not been pre-saturated and the amount of absorbed water was added to the mixing water. Moreover, since the amount of water is important, an additional quantity of cement was added such that the ratio of total water to cement remains constant. For NAC, the dosage of superplasticizer was gradually increased until the target slump was obtained. This dosage has not been modified for the other formulations because water initially added to mixes had allowed to obtain the slump 18 ± 1 cm.

Table 5, containing the mixes, shows that the adopted approach leads to an increase in paste volume with a slight decrease in density. In this table, the effective water, W_eff, is defined according to the standard EN 206-1 as the total water quantity, W_tot, minus the water absorbed by aggregates $W_{\text{eff}}=W_{\text{tot}}-{\mathit{\omega }}_a\times M_{\text{g}}$ where M_g is the weight of dry aggregates.

Table 5 Mix proportions for 1 m³.

Cylindrical 16 × 32 cm² specimens were prepared to determine the compressive strength, elastic modulus and splitting tensile strength. Furthermore plain and prenotched 10 × 10 × 40 cm³ prismatic specimens were cast to determine the flexural strength of studied concretes. After being removed from the mold, they were cured in a water tank at room temperature for 28 days.

3.4 Test Methods

Uniaxial compression and tensile splitting tests were performed using a servo-hydraulic INSTRON machine with a capacity of 3,500 kN by imposing a stress increment rate of 0.5 MPa/s. Each test was repeated at least three times and results shown below are the averages of obtained values. In addition one cylinder of each material was instrumented with two strain gauges in order to determine the elastic modulus, and test were performed by imposing a strain rate of 1 mm/min. Bending tests were performed using a 250 kN closed loop INSTRON machine with a strain rate of 1 mm/min. Finally, splitting strength was measured using the Brazilian test and dynamic modulus of elasticity was determined using E-Meter MK II device.

Water porosity was determined using the vacuum saturation method. The test includes two stages, the first consists on submitting 10 × 10 × 10 cm specimens, dried at 60 ± 5 °C, to vacuum (about 80 mbar) during 3 h. After this period and in a second stage, specimens are immersed in water during 5 days. The water absorption, called water porosity, is determined as follows:

$$W_A=\frac{W_{\text{sat}}-W_{\text{dry}}}{W_{\text{sat}}-W_{\text{wat}}}\times 100,$$

(4)

with W_sat is the weight of the saturated sample, W_dry is the weight of the dry sample, W_wat is the weight of the saturated sample immersed in water.

4 Test Results

4.1 Properties of the Fresh Concretes

The results of workability tests and air content are given in Table 6 where it can be seen that all mixtures compile with the required workability. It can be concluded that the air content of RAC is higher than concrete made with NA when the replacement ratio exceeds 30 %.

Table 6 Properties of fresh concrete mixes.

The slump loss during 2 h is plotted in Fig. 6 for all studied materials where the slump values are averages of two measurements. During the first 20 min, the loss is not significant for RAC65 and RAC100 and this is explained by the excess of available water in the mix. After 20 min the loss is more pronounced when the replacement ratio is higher. This trend was also observed by Poon et al. (2011).

Fig. 6

Slump loss of concrete mixes with time.

In order to verify if the accentuation is due to the continuous water absorption by recycled aggregates and not to the increase in paste volume, the loss in workability of two cement pastes corresponding to NAC and RAC100 was followed during 2 h. Figure 7 shows the changes of paste slumps with time where it can be seen that the two pastes undergo the same kinetic of loss.

Fig. 7

Slump loss of cement pastes with time.

4.2 Properties of the Hardened Concrete Specimens

4.2.1 Water Porosity

Figure 8 presents the water porosities, measured at atmospheric pressure and under vacuum conditions, for all concrete mixes. It can be pointed out that both porosities increase with replacement ratio. These results are in agreement with the results of the literature where the porosity increases with substitution rate (Gomez-Soberon 2002; Belin et al. 2013). It can be also shown that the ratio between the porosity measured under vacuum and the porosity measured at atmospheric pressure is constant and equal to 1.16.

Fig. 8

Water porosities measured at atmospheric pressure and under vacuum.

The increase of porosity with replacement ratio is mainly due to the high porosity of recycled aggregates, to the increase in the paste volume and to the poor interface paste-aggregates as well as to the increase in air content (cf. Fig. 9).

Fig. 9

Relationship between air content and water porosity.

4.2.2 Compressive Strength

Figure 10 shows compressive strength results at 7, 14, 21 and 28 days for the four concretes produced in this work. As illustrated in this figure, comparable strengths were obtained for all concretes with a decrease of 13 % for the concrete RAC30. This strength loss is due to recycled aggregates and to the increase in the total water quantity without correcting the cement content. For substitution ratio higher than 30 %, two phenomena are in competition: increasing the cement content and the replacement ratio of recycled aggregates which contribute to the increase of strength in the first case and the decrease in the second. However, the materials satisfy correctly the imposed specifications given in part 3 i.e. S4 flowability and 35 MPa compressive strength concretes.

Fig. 10

Compressive strength as a function of age and replacement (means and standard errors).

Compressive strength results were compared to the Féret strength equation with 54 results found in references (Xiao et al. 2006; Etxeberria et al. 2007; Evangelista and de Brito 2007; Gomes and Brito 2009; Belén et al. 2011; Martinez-Lage et al. 2012; Pereira et al. 2012; Manzi et al. 2013). The Féret strength equation is:

$$\frac{f_c^{\prime }}{f_{cm}}=K{\left(\frac{v_c}{v_c+v_w+v_a}\right)}^2,$$

(5)

where K is the Féret coefficient which depends on mix design and age, $f_c^{\prime }$ is the compressive strength of concrete (MPa), f _cm is the normal compressive strength of cement, v _c is cement content in concrete (m³/m³), v _w is the water content in concrete (m³/m³) and v _a is the air content in concrete (m³/m³).

The K value was evaluated based on the both natural and recycled aggregates concretes compressive strength (cf. Figure 11). It is found that a value of 5.27 fits adequately the experimental results with a correlation factor R² = 0.7. The obtained value is close to the theoretical value of K = 5 found in the literature (Julio et al. 2006; Hacene et al. 2009).

Fig. 11

Relationship between relative compressive strength and cement, water and air contents.

Féret’s equation can therefore help to explain the obtained compressive strengths for RAC65 and RAC100. Indeed, at a constant W/C ratio, when the concentration of cement increases in the paste volume (i.e. the reduction of effective water to cement ratio), the compressive strength is maintained constant despite the increase in air content (Table 5).

Moreover, the compressive strength of RAC30 decreased due to the reduction in cement concentration in the cement paste.

4.2.3 Elastic Modulus

The variation of both static and dynamic modulus of elasticity, denoted E, at the age of 28 days is plotted in Fig. 12. It can be seen that the recycled aggregates have a significant effect on the elastic modulus where it decreases with the increase of replacement ratio. These results are in good agreement with the literature results which indicate a decrease in the elastic modulus (Xiao et al. 2006; Casuccio et al. 2008). This reduction is the consequence of the application of recycled aggregates with a higher porosity and a lower elastic modulus than those of the natural coarse aggregates.

Fig. 12

Variation of modulus of elasticity as a function of replacement ratio.

4.2.4 Flexural and Splitting Tensile Strengths

The flexural strength obtained for all concretes is shown in Fig. 13 with the tensile splitting strength. It is possible to conclude that the tensile strength with recycled aggregates is negatively affected when replacement ratio increases. The loss of both flexural and tensile strengths is about 6 % for RAC30, 11 % for RAC65 and reaches 20 % for RAC100.

Fig. 13

Flexural and splitting strength as a function of replacement.

The correlation between tensile strength and the porosity is illustrated on Fig. 14. The results, as might be expected, show a decrease in tensile strength when the porosity increases. Figure 15 shows the normalized static modulus of elasticity versus the normalized tensile strength. The linear correlation indicates that the higher porosity of recycled aggregates affects both characteristics. These results are in agreement with those established by Evangelista and de Brito (2007).

Fig. 14

Splitting tensile strength versus water porosity.

Fig. 15

Correlation between the elastic modulus and tensile strength.

4.2.5 Analysis of Peak Strain and Stress–Strain Relationship Under Compression

Stress–strain curves were obtained by uniaxial compressive tests for all materials developed in the present work. The analysis of these curves shows that the peak strain corresponding to the maximum stress increases when increasing the replacement ratio. The evolution of this strain, normalized by the strain of NAC, is shown in Fig. 16 as a function of the rate of substitution with the results of Belén et al. (2011) and Martínez-Lage et al. (2012). The results show a linear increase of the normalized peak strain with the replacement ratio. However, this increase is more significant in the context of our work and it is probably attributed to the adopted experimental conditions.

Fig. 16

Influence of RCA content on the peak strain.

The influence of the recycled aggregates content on the complete stress–strain curves was also investigated. The results shown on Fig. 17 indicate that the shape of the post-peak curve is more spread when the replacement ratio is important. This observation highlights a more dissipative behavior when recycled aggregates are used, and may be explained by a more diffuse damage related to the nature of recycled aggregates.

Fig. 17

Normalized stress–strain relationship.

5 Prediction of Stress–Strain Relationship and the Applicability of EC2

Eurocode 2 proposes the following expression for predicting elastic modulus E (in MPa) from compressive strength $f_c^{\prime }$ (in MPa).

$$E=22000{\left(\frac{f_c^{\prime }}{10}\right)}^{0.3}.$$

(6)

The validity of the previous expression was verified using more than 230 concretes formulated with natural aggregates (see Table 7 in “Appendix” section). The results are plotted in Fig. 18 and the data analysis shows that the expression of EC2 does not allow a satisfactory prediction of elastic modulus. Equation (6) fits experimental results with correlation factor R² = 0.68 while the proposed Eq. (7) provides a better description of the experimental results.

Table 7 Elastic modulus database.

Fig. 18

Elastic modulus of NAC as a function of compressive strength.

$$E=17553{\left(\frac{f_c^{\prime }}{10}\right)}^{0.42}.$$

(7)

The elastic modulus of RAC developed in this work together with the results found in the literature (Gomez-Soberon 2002; Etxeberria et al. 2007; Evangelista and de Brito 2007; Casuccio et al. 2008; Domingo-Cabo et al. 2009; Belén et al. 2011) are shown in Fig. 19. From this figure it can be seen that Eq. (7) allows a better prediction of elastic modulus than those proposed in EC2 (Eq. (6)) for RAC. It can be pointed out also that the majority of experimental values are within the interval of ±10 % bounded by the two dotted red lines. Finally, it can be concluded that the relationship between E and $f_c^{\prime }$ does not seem affected by the nature of gravels.

Fig. 19

Variation of modulus of elasticity of RAC as a function of compressive strength.

EC2 proposes also a relation between the peak strain, ε_c1, to the compressive strength $f_c^{\prime }$ regardless the loading rate and the specimen size. This relation is given by the following expression:

$${\mathit{\epsilon }}_{\text{c}1}=0.7{\left(f_c^{\prime }\right)}^{0.31}\le 2.80/00.$$

(8)

The validity of this expression was verified for 66 concretes of the literature with natural aggregates (see Table 8 in “Appendix” section). The results presented in Fig. 20 show that the previous expression of EC2 does not predicted well the peak strain. However, it can be shown that the proposed expression given by Eq. 9 is more suitable for the prediction of peak stain.

Table 8 strain at peak stress database.

Fig. 20

Peak-strain as a function of compressive stress.

$${\mathit{\epsilon }}_{\text{c}1}=1.1{\left(f_c^{\prime }\right)}^{0.175}0/00$$

(9)

For the full stress–strain curve, EC2 proposes the simple following equation:

$$\frac{\mathit{\sigma }}{f_c^{\prime }}=\frac{k\mathit{\eta }-{\mathit{\eta }}^2}{1+(k-2)\mathit{\eta }},$$

(10)

with $\mathit{\eta }={\mathit{\epsilon }}_{\text{c}}/{\mathit{\epsilon }}_{\text{c1}}$, $k=1.05\times E_{\text{cm}}({\mathit{\epsilon }}_{\text{c}1}/f_{\text{c}})$ and E_cm the secant modulus of elasticity. The model requires the knowledge of the static modulus of elasticity, the compressive strength and the peak strain ε_c1. The application of this model to recycled aggregates concrete shows that it does not reproduce suitably the post-peak behavior (Fig. 21). Another simple expression was proposed by Carreira and Chu (1985).

Fig. 21

Strain–stress curves.

$$\frac{\mathit{\sigma }}{f_{\text{c}}^{{}^{\prime }}}=\frac{\mathit{\beta }\left(\frac{{\mathit{\epsilon }}_{\text{c}}}{{\mathit{\epsilon }}_{\text{c}1}}\right)}{\mathit{\beta }-1+{\left(\frac{{\mathit{\epsilon }}_{\text{c}}}{{\mathit{\epsilon }}_{\text{c}1}}\right)}^{\mathit{\beta }}},\text{with}\mathit{\beta }=\frac{1}{1-\frac{f_{\text{c}}^{{}^{\prime }}}{E·{\mathit{\epsilon }}_{\text{c}1}}}$$

(11)

Equations (7), (9) and (11) may therefore be used for the modeling of full stress–strain relationship of recycled aggregates concrete with the modification of peak strain with the replacement ratio. Figure 21 presents a comparison between the curves calculated using the modified model of Carreira and Chu (Eq. (9)) and the model of EC2 (Eq. (8)). It can be seen that this modified model is more adequate for the modeling of post-peak behavior as the model of EC2.

6 Conclusion

In this paper, a natural aggregates concrete, NAC, and three recycled concrete aggregates RAC30, RAC65 and RAC100 were prepared on the basis of an imposed constant flowability at fresh state and a target compressive strength of about 35 MPa at 28 days. Moreover only coarse aggregates were replaced by recycled ones with three volumetric replacement ratios being respectively 30, 65 and 100 %. Based on the experimental results the following conclusions can be drawn:

• The use of recycled aggregates up to 30 % does not affect the demand of water of concrete, but generates a reduction of 14 % of the compressive strength. By increasing the replacement ratio, the cement content increases to maintain constant W/C ratio causing an increase in the compressive strength which counterbalances the negative effect of recycled aggregates.

• Recycled aggregate concretes had lower elastic modulus, splitting and flexural tensile strength than normal aggregate one.

• The strain–stress curves under uniaxial compression show that the post-cracking branch is more spread out compared to NAC. In addition, the peak-strain increases by increasing the replacement ratio. These phenomena are explained by the more progressive and diffuse damage of concrete due to the presence of recycled aggregates.

New relationships for prediction of concrete’s elastic modulus, and a peak strain from compressive strength were proposed. The predicted results for RAC were closer to experimental results than values predicted by equations proposed in EC2. For the complete strain–stress curve, a model based on the Carreira and Chu’s model was proposed. The modified model is more adequate for the modeling of post-peak behavior than the model of EC2.

Appendix: Database of Mechanical Characteristics

See Tables 7 and 8.