Eurocode 2 predictions vs. mechanical properties of Self Compacting Concrete P. Van Itterbeeck1, N. Cauberg1, B. Parmentier1, L. Vandewalle2 and K. Lesage2 1

Belgian Building Research Institute (BBRI), Avenue P. Holoffe 21, 1342 Limelette, Belgium, [email protected] 2 KULeuven, Department of Civil Engineering, Reyntjens Laboratory, Belgium

Abstract. Since Self Compacting Concrete was introduced in the late 1980s most and foremost attention was dedicated to the assessment of the fresh concrete properties. However, for designers and users in general the mechanical properties of the hardened concrete are also of crucial importance. In the scientific world a debate still exists whether SCC should simply be regarded as another placing technique or rather as another material. Designers generally adopt the design guidelines prescribed by Eurocode 2, which are based on experimental results of ordinary vibrated concrete. Therefore the aim of this paper is to examine different key parameters, and this by means of own experimental results, and an extensive database of results gathered from literature. These key parameters being: compressive strength, tensile strength and modulus of elasticity. The experimental data are compared with the predictions of EC2, allowing the assessment of its performance and the necessity of adding specific constraints within the design rules for SCC.

Introduction Since the composition of a Self Compacting Concrete (SCC) can differ considerably from that of a Traditional Vibrated Concrete (TVC) there is still some debate whether it should be considered as a new material or simply as a variant of traditional concrete. In literature already a lot of attention was devoted to the characteristics of SCC in the fresh state. Even though the material is already frequently used, the mechanical performance of the hardened end product has not been studied thoroughly. It is not clear whether the design rules presented in Eurocode 2 (EN 1992, EC2) [1], widely used for the calculation of concrete structures in general, are still applicable and accurate for SCC. Within this paper special attention is devoted to some key mechanical aspects of SCC, being the compressive strength (cylinder/cube conversion factors), the tensile strength and the modulus of elasticity. Within this paper not only new experimental results on a considerable number of mix compositions will be presented, in addition also results will be gathered from literature, allowing as such a

2

P. Van Itterbeeck, N. Cauberg, B. Parmentier, L. Vandewalle and K. Lesage

correct assessment of the overall mechanical performance of SCC and the capacity of EC2 for predicting these.

Experimental Since its first development a lot of research was devoted to the performance of SCC. The focus of many of these studies was mainly directed to the mix design and fresh properties of this revolutionary concrete. Within the extensive literature scattered data with regards to the mechanical performance of SCC can nonetheless be found, with compressive strength and other mechanical data routinely being reported. Although very valuable, numerous of these investigations have only been limited to the study of a small amount of mixes. To be able to evaluate the capacity of EC2 to predict the mechanical behaviour of SCC a large number of data needs to be gathered and critically reviewed. Holschemacher et al [2] and Domone [3] already created a database and made a first interesting evaluation of the predictions made respectively by the Model Code 1990 (MC 90) and EC2. In this paper a database of literature results is used, gathering results of which not only sufficient information concerning the material composition, but also the testing procedure, specimen geometry, etc. is available, since these factors are critical for comparing results from different studies. To this database also original experimental results are added, generated in the scope of this study, on 8 carefully selected SCC mixes. Material and specimens fabrication As mentioned in the previous paragraph, 8 SCC mixes were tested in this study. In order to evaluate EC2 for SCC, the influence of some “key” composition parameters on the mechanical performance of the material were investigated. The SCC mixes are presented in Table I. Table I. Mix composition of the different tested SCC mixes CEM I 52,5 R (kg/m³) CEM III/b 42,5 (kg/m³) Limestone (kg/m³) Fly ash (kg/m³) Sand 0-5 mm (kg/m³) Gravel 4-14mm (kg/m³) Water (kg/m³) Superplasticizer (kg/m³) W/C (wt%) C/P (wt%) Powder content (kg/m³) Slumpflow (mm) V-funnel (s) L-box (%) Fresh air content (%)

M1 335 165 955 785 152 6,0 0,45 0,67 500 640 21 87 2,80

M2 335 165 930 770 152 4,7 0,45 0,67 500 640 17 89 2,65

M3 335 165 888 723 200 2,0 0,60 0,67 500 670 4 85 2,05

M4 335 335 861 712 152 8,4 0,45 0,50 670 710 28 94 1,70

M5 250 250 942 785 152 5,3 0,61 0,50 500 650 24 90 2,50

M6 250 250 922 748 152 4,3 0,61 0,50 500 720 14 76 2,20

M7 300 150 955 795 165 3,9 0,55 0,67 450 770 13 92 1,90

M8 300 150 936 779 165 2,9 0,55 0,67 450 740 11 80 2,45

In general, these 8 mixes can be divided into 2 main groups depending on the type of cement utilised, a Portland based cement CEM I (M1 to M6) or blended cement CEM III/b (M7 and M8). In each of these groups there is one reference mix (M1 and M7) and the other mixes are variations on this reference, studying the influence of: fillertype,

Eurocode 2 predictions vs. mechanical properties of Self Compacting Concrete

3

W/C ratio, powder content and c/p ratio. Only 2 types of fillers were considered in this research, limestone filler and fly ash, since these are the most used in the Belgian concrete industry. Within Table I some fresh concrete properties of the different mixes are also recorded. For all specimens prepared with mixes M1 to M6 the moulds were removed after 24 hours. Since mixes M7 and M8 presented low mechanical performances at an early age, for these mixes the moulds were removed after 3 days. Afterwards all specimens were stored in a climate chamber at 20°C and > 95% RH until testing. Test methods The compressive strength is the mechanical property of concrete which is by far the most routinely specified. In EC2 almost all material characteristics are expressed as a function of this parameter. Two specimen geometries are specified in EC2: cubes with a dimension of 150mm or cylinders with a diameter of 150mm and height of 300mm. A classification is made based on these cylinder/cube compressive strengths. Within this paper both types of specimens were tested, allowing as such the evaluation of the conversion factors used in EC2. The experiments performed in this study were conducted according to the recommendations of the standard EN 12390-3. For the determination of the direct tensile strength, EC2 allows for different test methods to serve as reference: (1) a direct tension test, (2) a tensile splitting test and (3) a bending test. Conversion factors for these indirect testing methods are provided in the text of the standard. Until now there is no European standard for the direct tensile testing of concrete. As a result the Belgian standard NBN B 15-211 was applied. Cylindrical specimens with a diameter of 150mm and height of 300mm were utilised, conform the requirements of this standard. For the bending test the European standard EN 12390-5 was followed, allowing a 3point, as well as a 4-point bending test to be conducted. Since a 4-point bending test allows for a better, more realistic estimation of the tensile strength of the material this test procedure was selected. The tests were performed on prismatic specimens of 150x150x600mm³. Different types of specimens are utilised to establish the splitting tensile strength of concrete (cubic, prismatic or cylindrical specimens). In the European standard EN 12390-6 the following statement can however be found with regards to the impact of specimen geometry on the measured tensile strength: “The apparent tensile strength measured depends upon the shape and size of the test specimen used...” The conversion factor within EC2 does however not account for shape and size of the utilised specimen, this in contrast with the conversion factor for the bending strength. The European standard EN 12390-6 also stipulates that “In cases of dispute, the reference method is the use of cylinders of 150mm diameter and 300mm length”, as a result it was opted to use this type of specimen for this present test program. The modulus of elasticity is also an essential parameter in the design of structures. It is for instance used in the calculation of the deflection, requirements for prestressed elements, etc. Two types of elastic moduli are defined in EC2: the static and the dynamic E-modulus. Since the static E-modulus and more specifically the secant Emodulus is the most referred parameter in structural calculations a focus was put on the evaluation of this parameter. EC2 does not clearly state the test procedure or

P. Van Itterbeeck, N. Cauberg, B. Parmentier, L. Vandewalle and K. Lesage

4

specimen geometry for which the expressions are valid. The only specification regarding the testing procedure which can be found in EC2 is the loading level, which is set to 0.40 of the mean compressive strength at the time of testing (generally 28 days). Because of the absence of a European standard for evaluating the static Emodulus the Belgian standard NBN B15-203 was adopted within this study. The only deviation made with regards to this standard was the adopted loading level, which was set to 40% instead of the recommended 33%. The tested specimens were prismatic with a section of 100x100mm² and a height of 400 mm.

Results and discussion Compressive strength As mentioned in a previous paragraph the compressive strength is by far the concrete material parameter that is most routinely specified. The other mechanical properties are usually expressed as a function of the compressive strength. As a result, EN 206-1 sorts concrete in different classes solely based on their compressive strength at 28 days. The same classification is adopted in EC2. The measured compressive strength depends on the size and shape of the utilised specimen, EN 206-1 (and EC2) specifies the compressive strength with classes based on cylinder (with a diameter of 150mm and height of 300mm) or cube (with sides of 150mm) resistance. For design purpose, the cylinder compressive strength is used. The majority of the equations within EC2 are written as a function of this parameter. However for each strength class a conversion factor is presented, allowing the conversion between cube and cylinder compressive strength values. Some researchers have voiced some concern with regards to these conversion factors (strength classes). Due to the specific mix composition of SCC it is not clear whether these are still applicable. In literature nonetheless little attention is devoted to this topic. This conversion factor (strength classes) forms however the base of all structural calculations made with EC2. This paper presents some experimentally obtained conversion factors for SCC mixtures. 1.1 EC2 SCC: Own results SCC: Literature results

fcyl/fcube (-)

1.0 0.9

Variation according to the first CEB International Recommendations (1964)

0.8 M2

M3

0.7 0.6 20

40

60

80

100

120

fcm,cube (MPa) Figure 1. Conversion factor between cube and cylinder compressive strength.

Eurocode 2 predictions vs. mechanical properties of Self Compacting Concrete

5

First of all, some results were collected from literature. These results can be found in Figure 1, as well as the conversion factors defined by EN 206-1 (in function of the strength class). It needs to be mentioned that these conversion factors were calculated through the mean cylinder and cube compressive strength values since in the original publications no information could be found with regards to the scatter present on the SCC experimental results, while EC2 (and EN 206-1) refers to the characteristic compressive strength values. Many of these results are also the outcome of a limited test series, in general only 3 specimens are tested for each of the geometries. Therefore, for the 8 specific SCC mixes analysed in this study for each mix 12 cylinders and more than 25 cubes were tested. This allows for a better assessment of the characteristic strength values and thus a good estimation of the appropriate conversion factors for each of the considered SCC mixes. The results of these experiments are also depicted in Figure 1 and in contrast with the results from literature expressed as a function of their characteristic values. When comparing the results, originating from own research and literature, with the EC2 conversion factors, a tendency for the development of higher conversion factors can be seen. With the exception of two results (mixes M2 and M3), all experimental conversion factors found are greater than the recommended conversion factors of the EC2, which represent average values found for TVC. The first CEB International Recommendations from 1964 state that experimental conversion factors for TVC are generally situated within the region of 0.70 to 0.90. Even taking this variation into account, +/- 50% of the experimental results for SCC are greater than this upper limit of 0.90. For TVC, an increase in conversion factor value was however observed in [4] with increasing cement content. The increase might thus be linked to the higher powder content of SCC, which generally lies between 400 and 600 kg/m³ [5]. (Experimental results in Figure 1 were obtained on SCC mixes exhibiting a powder content ranging between 450 and 670 kg/m³.) Moreover the mixes with fly ash presented a lower conversion factor than their counterparts with limestone filler. EN 206-1 (and EC2) will thus underestimate the cylinder compressive strength of SCC if the current strength classes are adopted. Since the majority of the mechanical properties are expressed as a function of this cylinder compressive strength, as mentioned previous, adopting the current conversion factors however still provides sufficient safety for structural calculations. Tensile strength The tensile strength of concrete is an important parameter within the calculation of minimum reinforcement and crack width. EC2 allows for the determination of the concrete tensile strength through either direct or indirect testing. Conversion factors are provided to convert splitting (s) and bending (b) strength into direct tensile strength. Since a direct tensile test is difficult to perform correctly, researchers often prefer these indirect testing methods for the determination of the tensile strength of concrete. It is however not clear whether the equations provided in the EC2 for on the one hand the prediction of the tensile strength in function of the compressive strength and on the other hand the conversion factors between indirect and direct tensile strength measurements are still applicable for SCC.

P. Van Itterbeeck, N. Cauberg, B. Parmentier, L. Vandewalle and K. Lesage

6

To assess the potential of EC2 to predict the tensile strength of SCC in function of the compressive strength, in the scope of this research a large set of data was collected from literature. These results are presented in Figure 2 in function of their mean cube compressive strength, since cylinder compressive strengths are almost never determined in these publications. All experimental data resulting from either direct or indirect testing are gathered in one graph and expressed in function of the direct tensile strength using the conversion factors provided in EC2. Within the same graph EC2 prediction of the tensile strength is also presented. Given that the equations in EC2 for the tensile strength are expressed in function of the cylinder compressive strength, for the construction of the predictive curve in Figure 2 an average conversion factor of 0.80 was considered. In addition to these literature results an extensive set of own results was built up. 7 6

fctm (MPa)

5 4 3

EC2 EC2 5%-95% Literature (S) Literature (B) Literature (D) Own results (S) Own results (B) Own results (D)

2 1 0 20

40

60

80

100

120

fcm,cube (MPa) Figure 2. Tensile strength in function of the cube compressive strength. (D= Direct tension test, S= Splitting tensile test and B= Bending test).

Direct tension tests, splitting tensile strength measurements as well as bending experiments were conducted on the 8 mixes presented in Table I (on 6 specimens for each series). The results of these experiments are presented in function of their mean cube compressive strength (fcm,cube) in Figure 2 and mean cylinder compressive strength (fcm) in Figure 3, in the latter with an indication of 5% and 95% confidence levels. Data presented in Figure 2 show that SCC exhibits a higher tensile strength than TVC, since the median of the results is higher than the EC2 prediction. This is an observation which was already made by other researchers ([2], [3], [6], [7]). However, when the results obtained in the scope of this research are analysed in function of their cylinder compressive strength (see Figure 3), this statement cannot be supported. When considering the results obtained for the cylinder/cube conversion factors obtained in the previous section of this paper, this will influence the tensile strength predictions made by EC2 presented in Figure 2, leading generally to higher direct tensile strengths. As a result the previous statement that SCC presents a higher tensile strength may no longer be valid.

Eurocode 2 predictions vs. mechanical properties of Self Compacting Concrete

7

7 EC2 EC2 5%-95% Own results (S) Own results (B) Own results (D) Literature (S) Literature (B)

6

fctm (MPa)

5 4 3 2 1 0 10

20

30

40

50

60

70

80

90

fcm (MPa) Figure 3. Tensile strength in function of the cylinder compressive strength. (D= Direct tension test, S= Splitting tensile test and B= Bending test).

A considerable scatter exists on the results, as is clear from figures 2 (between series) and 3 (within series). As a result the question arises whether the 5 and 95% limits used in EC2 for the tensile strength are still adequate/safe enough. The 5% and 95% characteristic values were calculated for the 8 mixes assuming a log-normal distribution (see Figure 3). While the number of samples tested in this research is limited, the intervals presented in Figure 3 indicate that EC2 still provides a safe prediction of the variation on the experimental tensile strength measurements. 8 Direct tensile strength

Splitting tensile strength

Bending strength

EC2

7

fctm (MPa)

6 5 4 3 2 1 0 M1

M2

M3

M4

M5

M6

M7

M8

SCC Mix # Figure 4. Comparison of direct and indirect tensile strength measurements.

P. Van Itterbeeck, N. Cauberg, B. Parmentier, L. Vandewalle and K. Lesage

8

Figure 4 presents a comparison per mix composition between the different direct tensile strength values obtained through the different test methods and use of the conversion factors made available by the EC2. An analysis of these results, presented in Figure 4, shows that the direct tensile strength obtained after conducting cylinder splitting tests and using the conversion factor given in EC2 is a good evaluation of the SCC direct tensile strength. On the other hand, the conversion factor used on bending experiments often seems to provide an overestimation of the direct tensile strength (with the exception of mixes M3 and M5). Table II. Experimental conversion factors, for converting tensile splitting (s) and bending strength (b) into direct tensile strength. S B

M1

M2

M3

M4

M5

M6

M7

M8

0.87 0.51

-

0.88 0.76

-

0.86 0.68

0.78 0.52

0.80 0.54

0.84 0.55

Table II presents the experimental conversion factors obtained for the tensile splitting (s) and bending (b) strength. s is slightly lower than the 0.90 prescribe by EC2. With the exception of M3 and M4 the experimentally obtained conversion factors for bending are considerably lower than the 0.69 prescribed by EC2 for this specific specimen geometry. EC2 will thus provide a slight but acceptable overestimation of the tensile strength when tensile splitting experiments are used as a measure for the direct tensile strength. An adaptation of the b implemented in EC2 seems however necessary for SCC. More experimental results are however necessary to establish a good conversion factor. E-modulus 50 45

Ecm (GPa)

40 35 30 EC2 SCC: Literature SCC: Own research TVC

25 𝐸𝑐𝑚 = 22.

20

1 𝑓𝑐𝑚 ,𝑐𝑢𝑏𝑒 . 0.8 10

0.3

15 20

30

40

50

60

70

80

90

100

110

fcm,cube (MPa) Figure 5. Secant E-modulus: Experimental values vs. Eurocode 2 prediction in function of the mean cube compressive strength. (SCC = Self-Compacting Concrete and TVC = Traditional Vibrated Concrete)

Eurocode 2 predictions vs. mechanical properties of Self Compacting Concrete

9

Due to the higher paste content SCC might be characterised by a lower E-modulus than TVC. Figure 5 compares literature data and results of this study for the secant modulus of elasticity (Ecm) with EC2 predictions. Once again for EC2 prediction based on the cube compressive strength, the conversion factor between cube and cylinder was set at an average value of 0.80. A first analysis of the results presented in Figure 5 reveals the existence of a large scatter on the results. Domone [3] and Holschemacher et al [2] came to the conclusion in their respective studies – after analysis of data taken from literature – that SCC presents a lower stiffness than TVC, especially at lower compressive strength values [3]. However, this conclusion cannot be supported by the results presented in Figure 5. Even when taking into consideration that the conversion factor between cylinder and cube compressive strength might be greater than 0.80 assumed for the construction of the EC2 prediction, it is difficult to conclude that the E-modulus of SCC is inferior to that of TVC. In Figure 6 some results for SCC and TVC are presented in function of their experimental mean cylinder compressive strength. Even though EC2 seems to provide a slight overestimation of the stiffness, the results presented in Figure 6 indicate a similar tendency for both types of concrete. The CEB-FIB ModelCode of 1978 also reported that experimentally obtained E-modulus values for TVC are generally situated within the range of 0.70 Ecm to 0.90 Ecm, and all SCC results remain well within that range. 50 EC2 SCC: Literature SCC: Own research TVC

Ecm (GPa)

45 40 35 30 25

𝐸𝑐𝑚 = 22.

𝑓𝑐𝑚 10

0.3

20

0

10

20

30

40

50

60

70

80

90

fcm (MPa) Figure 6. Secans E-modulus: Experimental values vs. Eurocode 2 predictions in function of the cylinder compressive strength. (SCC = Self-Compacting Concrete and TVC = Traditional Vibrated Concrete)

Conclusions In this study the mechanical performance of Self-Compacting Concrete was analysed and compared with the Eurocode 2 predictions. To this end an extensive database was constructed gathering experimental results from literature and own research. This

10

P. Van Itterbeeck, N. Cauberg, B. Parmentier, L. Vandewalle and K. Lesage

investigation has yielded following interesting conclusions with regards to some key mechanical parameters of SCC:  The conversion factor between cylinder and cube compressive strength was found to differ from that observed for TVC. The higher conversion factors which were observed might be a result of the higher powder content generally adopted in SCC. Additionally the results obtained in the framework of this study seem to indicate that not only the powder content but also the filler type has an impact on the conversion factor. The conversion factors stipulated in EN 206-1 (and EC2) however still provide a safe (over-)estimation of the cylinder compressive strength.  In contrast to previous observations made in literature the tensile strength was found to resemble that of TVC. Taking into account the higher cylinder/cube compressive strength conversion factor the EC2 was found to provide a good approximation of the tensile strength of SCC. Even when using the current cylinder/cube conversion factors prescribed by EN 206-1 (and EC2) a safe (under-)estimation of the tensile strength is made.  A large scatter was observed on the E-modulus results. EC2 seems to provide a slight overestimation of the E-modulus of SCC. However, when taking into account that the scatter found on the TVC results was of a similar magnitude as that found for the SCC mixes, the EC2 might still provide a good estimation of the E-modulus of SCC. Further research on this topic is needed.

Acknowledgments Funding by the Belgian Standardisation Bureau (NBN) under the contract CC CCN/PN/NBN-513 is gratefully acknowledged.

References [1] EN 1992-1-1, Eurocode 2: Design of concrete structures - Part 1-1: General rules and rules for buildings [2] Holschemacher K. en Klug Y. (2002), Lacer 7, p. 123-134 [3] Domone P. L. (2007), Cement and Concrete Composites, vol. 29, p.1-12 [4] Lambotte H. and Van Nieuwenburg D. (1993), University Ghent [5] De Schutter G., Bartos P.J.M., Domone P. and Gibbs J. (2008), Whittles Publishing, ISBN 978-1904445-30-2 [6] Dinakar P., Babu K.G., Santhanam M. (2008), Structural concrete, vol. 9, no. 2, p.109-116 [7] Felekoglu B., Türkel S. and Baradan B. (2007), Building and Environment, vol. 42, p.1795-1802