SOME PARAMETERS AFFECTING THE HEAT OF HYDRATION OF BLAST-FURNACE SLAG CEMENT

Transcription

1 ENSET Oran (Algeria) - October 12-14, 29 SOME PARAMETERS AFFECTING THE HEAT OF HYDRATION OF BLAST-FURNACE SLAG CEMENT A. Bougara 1, J. Khatib 2, H. Khellafi 3 T. 4. Durability of materials and structures ABSTRACT The effects of varying the slag content, the fineness of the slag, the curing temperature and the water to cement ratio on the early hydration of Algerian blastfurnace slag cement have been investigated by isothermal conduction calorimetry, by following the rate and total heat output for 72 hours at several temperatures. The objective of the work presented is to use the calorimetric behaviour of the Portland cement slag blend providing more information to promote its use in the cement industry. The results showed that addition of slag contributes to a reduction in the heat output which is very beneficial for mass concrete and hot weather concreting. However, at 6 o C the increase in heat output from a cement blend incorporating 5% slag was far greater than that of the cement alone indicating that slag hydration is accelerated by higher temperatures. Small differences in slag fineness had little effect on the early hydration but could be correlated with the heat output. KEYWORDS Slag, isothermal calorimetry, temperature, fineness, water to cement ratio. 1 Laboratory of Material Sciences and Environment, University of Hassiba Benbouali Chlef, Algeria, aekbougara@hotmail.com 2 Civil Engineering Materials, University of Wolverhampton Wulfruna Street WV1 1SB, UK, j.m.khatib@wlv.ac.uk 3 Laboratory Materials and Structures, Sciences and Technology, Oran University, Algeria, khelafi@yahoo.fr 9

2 ENSET Oran (Algeria) - October 12-14, INTRODUCTION Blastfurnace slag is a by-product from the manufacture of iron and consists essentially of silicates and aluminosilicates of calcium and other bases [Bougara 29]. The El-Hadjar iron factory (Algeria) produces about half a million tonnes annually but only 2% of the total slag is used in the roads and construction industries [Benghazi and Zeghichi 28]. Its use is limited because little is known of the material and the reluctance of constructors to use a product which has not been fully characterised. Furthermore, Algerian slag is known to exhibit low reactivity due to its low CaO/SiO 2 ratio [Bougara 29]. In Algeria, the experts in cement are trying to persuade the constructors to use this product with all the required safety factors. Besides this, the insufficiency in the manufacture of the cement in this country is another factor that pushes the constructors to search for mineral admixtures available in this country. It is well known that the major difference between Portland cement and slag cement is the reduced heat of hydration. Under adiabatic conditions, blast-furnace cement hydrates more slowly than a Portland cement (OPC) of the same strength class [Lang 22]. Slag cement, with its high content of slag, is commonly used to reduce the temperature rise in mass concrete, such as in foundations, dams and bridge abutments [Bamforth 198]. The temperature generated in the cores of such structures as a result of heat of hydration can be minimized by using slag cement and as a consequence, minimize the risk of crack formation. Wu et al. [1983] examined the early hydration of slag cement containing 4 to 65% of slag at different temperatures. They found that the effect of elevated temperature on slag cement hydration is much greater than for Portland cement, illustrating the benefit that slag cement realizes from thermal activation. In a comparative study of the methods used to measure the heat of hydration, Milestone and Rogers [1981] observed that small differences in fineness produced small differences in heat of hydration when using isothermal calorimetry. However, they reported that large differences in particle size distribution do cause changes in heats of hydration. The work carried out by Hwang and Shen [1991] on the effect of blast-furnace slag and fly ash on the hydration of Portland cement, showed that the height of the second peak in the calorimetric curve is inversely proportional to the water-solid ratio at fixed slag content and that the time of its appearance was prolonged. They explained such phenomena by the fact that increments in the slag content may reduce the heat of hydration generated by C 3 S and C 3 A. Ma et al. [1994] investigated a slag cement with 65% replacement in the temperature range 1-55 o C. They suggested that temperatures above 4 o C are high enough to activate slag hydration but below this temperature, the presence of slag retards cement hydration. Recently, Escalante-Garcia and Sharp [2] analysed two neat cement pastes and three blended cement pastes including granulated ground blast-furnace slag by means of isothermal conduction calorimetry at five temperatures in the range 1-6 o C. Their conclusion was that the total heat evolved from the slag blended cement was far greater than that of the cement at 6 o C, indicating that the slag reactivity was strongly favoured by the higher temperature. The objective of the present work is to monitor the calorimetric behaviour of the Algerian slag hydrating with Portland cement at different temperatures, using different replacement level of the slag, water cement ratios of the mix and finenesses of the slag. The results gathered provide very helpful information for a better utilization of the Algerian slag in the manufacture of the blended cement. 1

3 ENSET Oran (Algeria) - October 12-14, EXPERIMENTAL 2.1 Description of the technique An isothermal conduction calorimeter measures the heat output from a sample at an isothermal temperature. The heat produced by the sample is conducted away through a heat flow sensor to a heat sink. The conductor forming the thermal path is a thermopile which produces an electrical output proportional to the heat flow. The temperature of the sample during its hydration is essentially kept constant. The equipment used is a JAF Wexham development calorimeter with a silicone oil bath. A schematic of the JAF calorimeter is given in Figure Materials Figure 1. Shematic of the JAF calorimeter with two cells The materials used in this work were granulated blast-furnace slag provided by El-Hadjar iron factory and a UK CEM I cement (PC), (Lafarge CEM I 52.5N) with specific surface of 42 m 2 /kg. The slag was ground in a ball mill to four specific surface areas; 25, 31, 41 and 5 m 2 /kg. The oxide composition of the materials is given in Table 1. The particle size distribution of the PC cement and the four ground slags was determined using a laser diffraction particle size analyser (Coulter LS13) and is presented in Fig.2 and Fig.3. Table 1: Chemical composition of the materials used Chemical composition: % CEM I Slag SiO 2 Al 2 O 3 Fe 2 O 3 Mn 3 O 4 MgO CaO Na 2 O K 2 O TiO 2 BaO LOI SO 3 Moisture CaO/SiO 2 Glass content % 11

4 ENSET Oran (Algeria) - October 12-14, CEM I 52.5N 8 Cumulative (%) Cumulative (%) Particule size (µm) Figure 2 Particle size distribution of the OPC cement m²/kg 41 m²/kg 2 31 m²/kg 1 25 m²/kg Particule size (µm) Figure 3 Particle size distribution of the four ground slags 2.3 Sample preparation and testing methods In this investigation, two cement paste systems were investigated; PC (as the control), and blended cement slag with different replacement levels of slag. Samples were mixed externally. Powders and water warmed to the temperature of the bath before mixing. 15 g of powder was mixed by hand with distilled water at a specific water/cement ratio in a small plastic bag. The bags were then sealed with cellotape. The sealed sample was wrapped around the calibration heater and placed in an aluminium can containing transformer oil, used to effectively transfer heat between the sample and the aluminium can. The aluminium can was placed inside the calorimeter cell in contact with the thermopiles. When insulation was put in place, the cells were sealed and lowered into the oil bath. Readings were taken every 6 minutes for the first 24 hours and every hour thereafter for up to 72 hours. Further details are described by Cembureau [1977]. 3. RESULTS AND DISCUSSION 3.1 Effect of the replacement level of slag To assess the effect of replacing PC with slag at 2 o C, four systems were investigated; pure cement and slag cement containing 3, 5 and 7% slag by weight. The fineness of the slag was 5 m 2 /kg. Fig 4 shows the rate of heat output of the four systems hydrated at 2 o C using a water to binder ratio of.4. The main diagram shows the plot of heat evolution against time, while the inset shows the total heat of hydration. PC without slag showed a prominent peak II, attributed to the hydration of the cement component [Wu et al. 1983] whereas in the slag cement curves, as the slag content was increased, the shape of the curve altered and a shoulder to peak II was observed, becoming more distinct with increasing slag 12

5 . SBEIDCO 1 st International Conference on Sustainable Built Environment Infrastructures in Developing Countries ENSET Oran (Algeria) - October 12-14, 29 content. This is thought to be peak S attributed to the hydration of the slag [Wu et al. 1983]. As the slag content was increased, the rate of heat output for the peak II decreased from 2.9 W/kg for the Portland cement to 1 W/kg for the slag cement containing 7% slag. Also, the total heat evolved after 72 hours decreased from 282 kj/kg for the PC cement to 95 kj/kg for the blended slag with 7% slag (see the inset in Fig.4). From this, it is noticeable that the presence of the slag contributes clearly to a reduced heat output (about 66% reduction) which is a benefit in mass concrete. It is to be noted that the time to reach the peak II decreased as the slag content increased, suggesting that cement hydration was accelerated in the slag blend due to the increased water to cement ratio, which occurs with increasing slag content. The curves obtained at 2 o C were very similar to those previously reported by Utton [26], using the same calorimeter. At 2 o C, the heat output was low for all systems, indicating that the reactions are slow at low hydration temperatures. After 7 hours of hydration, the total amount of heat evolved continued to increase, suggesting that the hydration continues slowly after 72 hours with the relatively low curing temperature. Rate of heat output (W/kg) Peak II % slag 1% CEM I 3% slag 7% CEM I 5% slag 5% CEM I 7% slag 3% CEM I Peak S Totale heat output (kj/kg) Time (hour) Time (hour) Figur 4: Calorimetric curves for PC and slag cement with 3, 5 and 7% replacement level, cured at 2 o C 3.2 Effect of the temperature To examine the effect of increasing hydration temperature on PC and the slag cement with 5% slag, three temperatures were selected; 2, 4 and 6 o C. The fineness of the slag was kept at 5 m 2 /kg and the water to binder ratio at.4. The results for the slag cement with different replacements are compared with PC in Fig 5. The curves recorded are only plotted for 24 hours to show the shapes of the curves better. It is clear that as the temperature increases, the time to reach the maximum peak is reduced, indicating that an increase in temperature accelerates the hydration reaction of both cement and slag. The peak S observed on the slag cement curves tends to merge in one peak as the temperature increases. Similar observations have been reported by Escalante-Garcia and Sharp [2]. The gain in heat output at 4 o and 6 o C was 57% and 84% for PC and 67% and 87% for the slag cement incorporating 5% slag respectively. This means that the slag cements were more thermally activated by the higher temperatures than PC. Wu et al. [1983] reported similar findings using hydration temperatures of 15 o, 27 o, 38 o and 6 o C and deduced that if the hydration temperature decreased markedly, the hydration rate of slag cements will diminish more than that of PC. 13

6 . SBEIDCO 1 st International Conference on Sustainable Built Environment Infrastructures in Developing Countries ENSET Oran (Algeria) - October 12-14, Rate of heat output (W/kg) C 4 C 2 C Total heat output (kj/kg) C 4 C 2 C Slag fineness 5 m²/kg 6 C 4 C 2 C 5% CEM I / 5% slag 1% CEM I / % slag Figure 5 Isothermal conduction calorimetry curves for neat cement and slag cement with 5% slag at various temperatures In this investigation, the slag cement hydration was clearly accelerated at 4 o C but slowed down above that. Ma et al. [1994] suggested that temperatures above 4 o C were high enough to activate slag hydration. 3.3 Effect of slag fineness The effect of varying the slag fineness was also investigated using four finenesses; 25, 31, 41 and 5 m 2 /kg at a water binder ratio of.4 and at 6 o C. Rate of heat output (W/kg) CEM I 52.5 N Finenesse 25 m²/kg Finenesse 31 m²/kg Finenesse 41 m²/kg Finenesse 5 m²/kg 5% Slag T=6 C, W/C=.4 Total heat output (kj/kg) Times (hours) Figure 6 Isothermal conduction calorimetry curves for the slag cement with different finenesses The results for PC and slag blends with different finenesses are shown in Fig 6. The rate of heat output of PC is much greater than that of all the slag cements, even that with the finest slag. After about 6 hours, the rate of heat evolution for all the slag cements has almost the same shape, illustrating that the hydration was accelerated by the temperature rise. During the first 24 hours, the 14

7 ENSET Oran (Algeria) - October 12-14, 29 total heat output reached 158, 163, 178 and 184 kj/kg for the slag with finenesses of 25, 31, 41 and 5 m 2 /kg respectively. Although the difference between the various slag cements is not large, the heat evolved seems to be proportional to the specific surface of the slag cements. Douglas and Zerbino. [1986] thought the heat evolved by the slag cements they analysed was proportional to their fineness, as it was noticeable that the heat evolved increased as the fineness of the slag increased. The finest slag gained over 14% in heat output with respect to the coarsest one. This finding is consistent with that of ACI committee 27 [1973] who reported that the variation in heat of hydration determined by isothermal calorimetry caused by different finenesses were small and observed only at early ages. It may be seen from the calorimetric curves that in the present sample pastes there was little evolution of heat after about 48 hours indicating that the hydration reaction was very slow at the end of the curing hydration period. 3.4 Effect of the water to cement ratio The water to cement ratio is another parameter that may influence the evolution of the heat of hydration of the slag cement blends. Paste samples were made using three different water to binder ratios;.35,.4 and.45. These were hydrated at 6 o C for 72 hours. The fineness of the slag used was 31 m 2 /kg and the replacement level was 5%. 12 Rate of heat output (W/kg) T 6 C, 5% slag W/C=.35 W/C=.4 W/C=.45 Total heat output (kj/kg) Figure 7 Isothermal conduction calorimetry curves for the slag cement with different water to cement ratio Fig 7 illustrates the effect of varying the water to binder ratio on the calorimetric curve of the slag cement pastes. During the first few hours of hydration, the calorimetric curves of the three samples were identical with only one peak, indicating the hydration of the cement components. The height of the peak was 11 W/kg at 1.7 hours of hydration. After about 8 hours, the total heat evolved from the samples started to differentiate into different values which suggests the beginning of the hydration of the slag component at the different water to cement ratios. The hydration of the sample with a water to binder ratio of.4 is accelerated and the heat released increased considerably; to about twice that of the other samples (.35 and.45). The total heat released by the samples after 72 hours is inversely proportional to the water to cement ratio except for w/b =.35. This result seems to be consistent with the finding obtained by Hwang and Shen [1991] but with small differences. They found that with a 15

8 ENSET Oran (Algeria) - October 12-14, 29 slag content of 2%, the rate of the heat output was inversely proportional to the water-binder ratio at 25 o C. It is possible that the percentage of the slag used in this work (5%) is the major reason for that difference, and can be attributed perhaps to the dilution effect. It is to be noted that the water to cement ratio of.4 constitutes an optimal value promoting the hydration at high temperatures. The hydration of the samples with water to cement ratio of.35 and.45 appears to be complete after three days while with the w/b =.4 sample still continues to evolve heat. 4. CONCLUSIONS A partial replacement of PC by blast furnace slag resulted in a reduction of the total heat generated which is very useful for mass concrete to minimise the risk of cracking. While the contribution of heat from the slag is not noticeable at low curing temperatures, with an increase in temperature, the presence of the slag contributes a significant rise in the heat of hydration indicating that slag hydration is more thermally activated than PC. A 4 o C curing temperature is found to be the optimum temperature to promote the acceleration of the hydration of the cement incorporating 5% of the slag. At higher temperatures, the total heat released after 72 hours is proportional to the water to cement ratio. At 6 o C the optimal value for the water to binder ratio giving the most hydration is equal to.4. The total heat released from slag cements is affected by variations in slag fineness, particularly at early hydration. The results found demonstrate that the heat output is proportional to the slag fineness at elevated temperatures. 5. REFERENCES ACI Committee , Effect of restraint, volume change and reinforcement on cracking of massive concrete. ACI journal, Bamforth P.B. 198, In-situ measurement of the effect of partial Portland cement using fly ash, ground granulated blast furnace slag, on the performance of mass concrete, Proc. Institut. Civil Eng., Part 2, Benghazi Z. & Zeghichi L. 28, Le ciment au laitier sans clinker :Activation mixte, Séminaire National de Génie Civil "SNGC8" Chlef, Bougara A., Lynsdale C. & Ezziane K. 29, Activation of Algerian slag in mortars. Construction and Building Materials, [23], Cembureau. 1977, Recommended procedure for the measurement of the heat of hydration of cement by the conduction method, Paris, Cembureau, January, pp.8. Claire A. U. 26, The encapsulation of a BaCO 3 waste in composite cements, PhD thesis, The University of Sheffield, UK. Douglas E. & Zerbino R. 1986, Characterisation of Granulated and Pelletized Blast Furnace Slag, Cem. Concr., Res., 16 [15], Escalante-Garcia J.I. & Sharp J.H. 2, The effect of temperature on the early hydration of Portland cement and blended cements. Adv. Cement Res., 12[3], Hwang C.L. & Shen D. H. 1991, The effects of Blastfurnace slag and fly ash on the hydration of Portland cement, Cem. Concr. Res, [21], Lang E. 22, Blastfurnace cements in Structure and performance of cements, 2 nd ed. Bensted J and Barnes P, Spon Press, Ma W., Sample D., Martin R., & Brown P.W. 1994, Calorimetric study of cement blends containing fly ash, silica fume and slag at elevated temperatures. Cement and Concr. and Agg., [16], Milestone N. B. & Rogers D E. 1981, The use of an isothermal calorimeter for determining heats of hydration at early ages, World Cement Technol., [12]; 376, Wu X, Roy D M. & Langton C A. 1983, Early stage hydration of slag-cement. Cem. Concr. Res., [13],