HEAT OF HYDRATION OF SELF-COMPACTING CONCRETE

Transcription

1 HEAT OF HYDRATION OF SELF-COMPACTING CONCRETE Anne-Mieke Poppe and Geert De Schutter Magnel Laboratory for Concrete Research, Department of Structural Engineering, Ghent University, Belgium Abstract The two most essential properties of self-compacting concrete (SCC) are a high flowability and a high segregation resistance. The use of admixtures in combination with a high concentration of fine particles makes it possible to combine these apparently incompatible properties. However, the high concentration of cement and fillers, often with puzzolanic properties, can lead to the development of a high heat of hydration. This might induce rather high thermal stresses in the hardening concrete element, possibly causing early age thermal cracking. In literature, the incorporation of nonpuzzolanic filler is sometimes recommended, however only very few data concerning the heat of hydration are given. In an experimental research program adiabatic hydration tests are carried out in order to evaluate the heat of hydration of SCC. The results are compared with data obtained for traditional concrete. The applicability to SCC of an existing hydration model is investigated. Based on the obtained results, the heat generation in hardening SCC is studied in more detail. Recommendations are given for further research. 1. Introduction In literature, only very few data are given concerning the heat development in concrete elements made with self-compacting concrete (SCC). In the proceedings of the First International RILEM Symposium held in Stockholm in 1999 [1], which can be considered as the state-of-the-art of that moment, the temperature rise in SCC is only mentioned occasionally in three papers dealing with field applications. In only one paper, some results of calorimeter tests are given [2]. However, the results were used to investigate the influence of the type and dosage of superplasticizer on the peak time of the exothermic reaction. No further results are given concerning the evolution and the 71

2 amount of heat of hydration of SCC. Due to the reduction in aggregate content in SCC, a higher volume of cement is used and/or a high-volume replacement of filler material like fly ash, limestone filler, blast furnace slag. The use of cement replacement is recommended in [3], in order to limit the heat genereation, as the replacement materials are generally less reactive than cement. Non-reactive materials like stone dust could further reduce the heat of hydration. At the Magnel Laboratory for Concrete Research, Ghent University, an extended research program is going on concerning SCC. One of the research topics within this project deals with the heat of hydration during hardening. Adiabatic hydration tests are carried out on SCC, and the results are compared with results obtained for traditional concrete. The applicability of an existing hydration model is also evaluated. 2. Adiabatic hydration tests The total heat production Q (expressed in Joule per gram cement, J/g) and the heat production rate q = dq/dt (in Joule per gram cement and per hour, J/gh) can be calculated by measuring the temperature rise of a perfectly insulated concrete as a function of time during an adiabatic hydration test. At the Magnel Laboratory an adiabatic test setup was developed, as shown in figure 1 [4]. steering equipment Heating element Concrete cylinder Air Water Insulation Figure 1 : Adiabatic hydration test In the adiabatic test setup, a cylindrical concrete specimen (Ø 280 mm, height 400 mm) is surrounded by a water-ring. By means of an electrical heating element, connected with an automated steering equipment, the water ring is kept at the same temperature as the concrete. In this way the heat loss is kept at a neglectable value, and the concrete is 72

3 kept in adiabatic conditions. For practical reasons, an insulation between concrete and water is created using an air-ring. 3. Experimental program In the experimental research program, the heat of hydration is determined on three different compositions of SCC, as shown in table 1. The composition is similar for the three mixes, having the cement type as the main parameter. SCC-1 and SCC-3 are selfcompacting concretes made with rapid hardening portland cement CEM I 42.5 R and CEM I 52.5 R resp. For SCC-2, a blast furnace slag cement CEM III/A 42.5 LA is used. The limestone filler originated from Marquise, France. Each of the mixes is characterised by means of slump flow, V-funnel, U-test, air content and density. The results are given in table 2. The 28-days compressive strength of the three mixes was about 60 to 65 N/mm², whereas Young's modulus was about N/mm² for SCC-1 and about N/mm² for SCC-2 and SCC-3. Table 1: Concrete composition SCC-1 SCC-2 SCC-3 Cement type CEM I 42.5 R CEM III/A 42.5 LA CEM I 52.5 R Cement content 360 kg/m³ 360 kg/m³ 360 kg/m³ Limestone filler 240 kg/m³ 240 kg/m³ 240 kg/m³ Water 165 kg/m³ 165 kg/m³ 165 kg/m³ Sand 0/5 853 kg/m³ 853 kg/m³ 853 kg/m³ Gravel 4/ kg/m³ 698 kg/m³ 698 kg/m³ Glenium l/m³ 2.5 l/m³ 2.2 l/m³ Table 2: Characterization of the fresh mixes Slump flow V-funnel U-test Air content Density SCC mm 8 sec. self-levelling 1.8 % 2330 kg/m³ SCC mm 7.5 sec. self-levelling 1.8 % 2350 kg/m³ SCC mm 6.5 sec. self-levelling 1.8 % 2310 kg/m³ 73

4 In a former project, adiabatic hydration tests were also performed on traditional concrete (TC) consisting per m³ of 300 kg cement, 150 kg water, 670 kg sand and 1280 kg gravel [5]. These tests were carried out with the same cement types as used for the SCC. The 28-days strength of the TC was about 46 N/mm² for TC-1 (based on CEM I 42.5 R), about 45 N/mm² for TC-2 (based on CEM III/A 42.5 LA) and about 60 N/mm² for TC-3 (based on CEM I 52.5 R). Young's modulus was within the range of to N/mm² at 28 days. 4. Experimental results The experimentally obtained adiabatic hydration curves are shown in figure 2 for the mixes SCC-1, SCC-2 and SCC-3. For each of the mixes, the adiabatic curve of the corresponding TC is also given as a comparison. 80 temperature ( C) D A C E C A E B F SCC - CEM I 42,5 R B TC - CEM I 42,5 R SCC - CEM I 52,5 D TC - CEM I 52,5 SCC - CEM III/A 42,5 LA F TC - CEM III/A 42,5 LA time (hour) Figure 2: Adiabatic hydration curves for SCC and TC From the adiabatic curves it can be concluded that the maximum temperature rise is systematically higher for SCC in comparison with TC. A more detailed comparison however can be obtained when the heat of hydration is expressed per unit weight of cement. To do so, the adiabatic temperature curves can be translated into a heat production curve, according to: ρ (t) = c c ( θ(t) θ ) (1) C Q 0 74

5 with Q = cumulated heat of hydration (in J/g cement), c c = specific heat of concrete (in J/kg C), θ = temperatur ( C), θ 0 = start temperature ( C), ρ = concrete density (kg/m³), C = cement content (kg/m³). For the compositions considered, the specific heat of the concrete can be approximated by 1000 J/kg C. Knowing the total heat production Q(t), the heat production rate q(t) in adiabatic conditions can be calculated. In order to be able to compare the different compositions, q(t) can be further transformed into q 20 (r), with q 20 = the heat production rate at a temperature of 20 C and r = the degree of reaction defined as Q/Q max (with Q max the maximal cumulated heat of hydration at the end of the reaction). The temperature influence is modelled by means of an Arrhenius function (4), with apparent activation energy equal to 33.5 kj/mol for portland cement CEM I 42.5 R and CEM I 52.5 R, and 40 kj/mol for blast furnace slag cement CEM III/A 42.5 LA [5]. More details about this transformation can be found in [4]. In this way the results shown in figures 3 to 5 are obtained, for the different cement types considered. 5. Hydration model For a further evaluation of the hydration process in SCC in comparison with TC, the hydration model developed in [4] can be used. In this model, the heat production rate is calculated by: q max, 20 = q.f (r).g( θ) (2) a [ sin ( rπ) ].exp ( br) f (r) = c. (3) E 1 1 g ( θ) = exp (4) R θ with q max,20 the maximum heat production rate at 20 C, a, b and c parameters, E the apparent activation energy, and R the universal gas constant. By means of a least squares method the different parameters a, b and c occurring in equations (3) can be determined, as given in table 3. The experimentally obtained q max,20 and Q max are also given in table 3. 75

6 q20 (J/gh) experiment SCC-1 model SCC-1 model TC degree of reaction r (-) Figure 3 : q 20 (r) for concrete with CEM I 42.5 R q20 (J/gh) experiment SCC-2 model SCC-2 model TC degree of reaction (-) Figure 4 : q 20 (r) for concrete with CEM III/A 42.5 LA 40 q20 (J/gh) experiment SCC-3 model SCC-3 model TC degree of reaction (-) 76

7 Figure 5 : q 20 (r) for concrete with CEM I 52.5 R Table 3: model parameters a (-) b (-) c (-) q max,20 (J/gh) Q max (J/g) CEM I 42.5 R SCC TC CEM III/A 42.5 LA SCC TC CEM I 52.5 R SCC TC A comparison of the results obtained for SCC with the results for TC indicates that for portland cement the SCC shows a higher q max,20 than for the case of TC. The ratio of both values is about 1.4. For the blast furnace slag cement, the opposite phenomenon is recognized. For all cement types, the total heat production Q max expressed in Joule per gram cement seems to be higher for TC than for SCC. The curves obtained with the hydration model are also shown in figures 3 to 5. It can be noticed that in the case of portland cement a discrepancy exists between experimental results and model. At a degree of reaction of about 0.3 to 0.4, a peak in the experimental curve is obtained, causing a deviation between experiment and simulated results. The hydation model is not predicting this peak, and as a consequence the values for the parameters a, b, and c, as given in table 3 for SCC, differ from the values obtained for TC. For the blast furnace slag cement, a better agreement is noticed, although a better result might be obtained when also considering the slag reaction, as explained in [4]. 6. Discussion Especially for portland cement, the heat of hydration of the cement in SCC seems to be different from the heat of hydration developed in TC. The heat production peak shown in figures 3 and 5, together with the differences found for the values of q max,20 and Q max might indicate that the hydration reaction is significantly influenced by the precense of the limestone powder in SCC. In order to further investigate this hypothesis isothermal hydration tests will be carried out on pure cement as well as on mixtures between cement and limestone powder. 77

8 7. Conclusions Based on adiabatic hydration tests on self-compacting concrete (SCC) and traditional concrete (TC), incorporating different types of cement, the following conclusions are obtained: - The maximum adiabatic temperature rise is higher for SCC in comparison with TC. This is related with the higher cememt content in SCC. - The total heat production Q max, expressed in Joule per gram cement, seems to be lower for SCC than for TC. - In the case of portland cement, the maximum heat production rate at 20 C q max,20, expressed in Joule per gram cement and per hour, seems to be higher for SCC than for TC. In the case of blast furnace slag cement, the opposite phenomenon is observed. - The hydration reactionin the case of SCC seems to be altered in comparison with TC. The reason for this might be the presence of the high amount of limestone powder. Further experiments are needed to evaluate this supposition. - Because of the modification of the hydration reaction in the case of SCC, the hydration model developed for TC shows some discrepancy with experimental results when applied to the case of SCC. 8. References 1. Skarendahl, A. and Petersson, O. (Eds.), 'Self-Compacting Concrete', Proc. of the First Int. RILEM Symposium (RILEM Publications, Cachan, 1999) 2. Umehara, H., Hamada, D., Yamamuro, H. and Oka, S., 'Devolopment and usage of self-compacting concrete in precast concrete field', in 'Self-Compacting Concrete', Proc. of the First Int. RILEM Symposium, Stockholm, 1999 (RILEM Publications, Cachan, 1999) Khayat, K.H., 'Workability, testing, and performance of self-consolidating concrete', ACI Materials Journal, 96 (3) May-June (1999) De Schutter, G. and Taerwe, L., 'General hydration model for portland cement and blast furnace slag cement', Cem. Con. Res., 25 (3) (1995) Neyrinck, B., 'Characterization of Belgian cements concerning heat of hydration', Master's thesis (in Dutch), Ghent University,