Anais do 45º Congresso Brasileiro de Cerâmica 0803201 30 de maio a 2 de junho de 2001 - Florianópolis – SC. ASPECTOS FLUIDODINÂMICOS E TÉRMICOS DURANTE REMOÇÃO DE ÁGUA LIVRE EM CONCRETOS REFRATÁRIOS DE ALTA ALUMINA TRATADOS ATÉ 1350ºC M.D.M. Innocentini, C. Ribeiro, J. Yamamoto, L.R.M. Bittencourt, R.P. Rettore, V.C. Pandolfelli Rodovia Washington Luis, km 235, São Carlos, SP, CEP: 13565-905, Tel: (16) 260-8250 ramal 2067, Fax: (16) 261-5404, e-mail: pmmi@iris.ufscar.br ou vicpando@power.ufscar.br Universidade Federal de São Carlos (UFSCar) Departamento de Engenharia de Materiais (DEMa) Magnesita S.A., Cidade Industrial - 32210-050 - Contagem – MG. ABSTRACT This article reports on an experimental investigation of the dewatering process of cement-free high-alumina refractory castables thermally treated up to 1350ºC. Simultaneous fluid dynamic, thermal and mass loss effects were investigated during the removal of physically absorbed water at temperatures of 25 to 700 C. It was found that the release of steam was decisively affected by the castable’s permeability level and the heating rate applied. The analysis of fluid dynamics revealed that, at 1C/min, the main bulk of physical water was released as steam under saturated conditions at 100C. However, at 5C/min, steam was trapped within the pores, with chaotic water loss and a shift to higher temperatures. A thermal analysis showed that the endothermic boiling of water may result in a critical thermal shock in the castable’s structure. Both steam entrapment and thermal shock were more severe with the reduction in the castable’s permeability level. Key words: drying, refractory castables, concretes, permeability, particle size distribution. Anais do 45º Congresso Brasileiro de Cerâmica 0803202 30 de maio a 2 de junho de 2001 - Florianópolis – SC. INTRODUCTION The dewatering behavior of refractory castables is highly complex and has been the subject of many theoretical and experimental studies found in the literature (1-5). Much of this complexity lies in the fact that water may vary in content and nature in the castable prior to dewatering. Physical water, or moisture, refers to the component of added water that remains chemically uncombined in a castable’s pores after mixing and curing. Chemical water, on the other hand, includes hydrates and gels of calcium aluminate cements (CACs) and hydratable alumina binders (HABs) (2). The relative amount of physical and chemical water depends on the total amount of water added for mixing and on the content and type of cement. Nevertheless, the relative importance of each fraction of water on the drying process is still a controversial question in the literature. The elimination of physical water is the main cause of weight loss at temperatures approaching 100°C (at ambient pressure). The decomposition of hydrates and gels, on the other hand, occurs according to complex kinetics at temperatures ranging from 100ºC to 550C and above, releasing a specific number of water molecules in proportion to the degraded phase. However, because temperature plays an important role in the buildup of steam pressure, it is not yet clear whether boiling at 100C is more deleterious than dehydration at higher temperatures to the castable’s mechanical strength. Furthermore, thick bodies may undergo an internal thermal gradient during fast heating programs, causing the release of steam to shift to higher temperatures and making dewatering even more complex. Due to experimental difficulties, the literature still shows a lack of simultaneous fluid dynamic and thermal data on the dewatering process, particularly insofar as the differences between boiling of physical water and dehydration of binders during the thermal treatment of high-alumina refractory castables are concerned. The present work focuses on the optimization of the dewatering process of highalumina refractory castable compositions based on hydratable alumina binders (HABs). In this context, this report discusses the fluid dynamics, heat transfer and mass loss involved in the elimination of physical water from refractory castables subjected to thermal treatment from 25ºC to 700C. Anais do 45º Congresso Brasileiro de Cerâmica 0803203 30 de maio a 2 de junho de 2001 - Florianópolis – SC. EXPERIMENTAL PROCEDURE The castable composition chosen for this study was similar to one previously described by Innocentini et al. (6) and consists of a high-alumina refractory castable prepared according to Andreasen’s model for particle size distribution (q = 0.21) and molded in the form of 7.5 cm diameter 2.5 cm thick cylinders. An amount of 2-wt% of hydratable alumina (Alphabond 300, ALCOA – Pittsburgh, U.S.A.) was used as binding agent and 4.12 wt% of water was added to the castable mixture. Samples were cured at room temperature for 24 hours. Samples were previously calcined in a furnace to decompose all the hydrates and gels and eliminate the chemical water. Two different temperatures were chosen (900C and 1350C with a 12-hour dwell time) in order to provide the specimens with distinct permeability levels (7). Heating rates were kept slow (1C/min up to 300C, 2C/min up to 900C and 3C/min up to 1350C. In order to investigate the fluid dynamic and thermal effects of the removal of physical water, the open pores of the two pre-fired castable samples were saturated with water under vacuum. Dewatering tests were carried out in a hot air permeameter (HAP) described by Innocentini et al (8-9). The test specimen was tightly fixed in a sample-holder and placed in an electronically controlled electric furnace (7500 W). Dry air supplied by a 2 HP compressor was pre-heated in the furnace according to a preset heating program and allowed to flow vertically upward through the sample at a constant pressure. The flow rate was measured at room temperature at the sample’s exit to capture not only air but also any gas or vapor released by the castable. Temperature and pressure were monitored at both the sample’s inlet and outlet surfaces exposed to airflow. The temperature, pressure and flow rate data were electronically recorded by computer at 1 second intervals throughout the test. The samples subjected to drying tests were heated at a rate of 1C/min from ambient temperature to 700C. The air pressure at the sample’s entrance was set at an absolute value of 1.021x105 Pa (1.10 bar) and kept constant throughout the test. After this first run, the samples were again saturated with water and subjected to a further test at a heating rate of 5C/min. Anais do 45º Congresso Brasileiro de Cerâmica 0803204 30 de maio a 2 de junho de 2001 - Florianópolis – SC. RESULTS AND DISCUSSION Figure 1-a shows the primary parameters obtained by a blank test run in the HAP device with a HAB castable sample pre-treated at 1350C for 12 hours and free of both moisture and hydrates. The heating rate was 5C/min from 25C to 700C, with a 100-minute dwell time at 700C. The samples were allowed to cool naturally inside the furnace. Figure 1-b displays the flow rate profiles through the blank sample as a function of the outlet air temperature. The curve obtained at room temperature indicates a smooth cyclic behavior, with a decrease in flow rate during the heating phase and an increase during cooling, returning to the initial value at ambient temperature. As expected, no mass decomposition peak was observed in the temperature range analyzed. However, considering that air expands volumetrically with increasing temperature, the original flow rate curve was corrected to the exit temperature, according to the ideal gas law, to evaluate the actual permeability changes that occurred in the castable structure (6). The corrected curve observed in Figure 1-b revealed that, at the air pressure applied, thermal expansion caused a temporary microstructural rearrangement, opening pores and intensifying fluid flow (8). Anais do 45º Congresso Brasileiro de Cerâmica 0803205 30 de maio a 2 de junho de 2001 - Florianópolis – SC. Figure 1. Results of a blank test run with a castable sample pre-treated at 1350C for 12 hours and free of both moisture and hydrates. Heating rate was 5C/min. (a) Primary parameters obtained by the test. (b) Fluid dynamic curve. (c) Thermal curve. As shown in Figure 1-c, no detectable local thermal effect was found within the temperature range analyzed, except for the heat transferred from the air to the sample (verified by the lower temperature at the sample’s outlet surface). After the blank test, the same sample was saturated with water and a new test was run in the HAP device at a heating rate of 1C/min. The results are illustrated in Figure 2. Anais do 45º Congresso Brasileiro de Cerâmica 0803206 30 de maio a 2 de junho de 2001 - Florianópolis – SC. Figure 2. Dewatering behavior of a castable sample pre-treated at 1350C and saturated with water. The heating rate was 1C/min. (a) Fluid dynamic curve. (b) Thermal curve. The flow rate curve in Figure 2-a displayed a single distinct peak around 100C, which coincides with the boiling point of free water under ambient pressure. The flow rate then slowly decreased during the remainder of the heating stage. The thermal curve at 1C/min (Figure 2-b) indicates that the sample absorbed energy during the water loss period, as the temperature at the exit of the sample remained constant at 100C, while the air temperature at the entrance steadily increased. This behavior represents the endothermic boiling of free water. The hysteresis in both flow rate and temperature curves confirmed the occurrence of an irreversible change, typical of mass loss phenomena in the heating stage. Extraordinary changes in the dewatering behavior were found when the same sample was again saturated with water and tested at 5C/min. As shown in Figure 3a, although the main peak again reached its maximum near 100C, the flow rate was disturbed up to 700C, indicating that steam was generated faster than it could be eliminated from the castable bulk and its release shifted to higher temperatures. The thermal curve displayed in Figure 3-b showed a much greater discrepancy between entrance and exit temperatures than the one observed at 1C/min, with a thermal Anais do 45º Congresso Brasileiro de Cerâmica 0803207 30 de maio a 2 de junho de 2001 - Florianópolis – SC. gradient of almost 150C between both castable surfaces during elimination of the main bulk of free water. The obvious explanation, not covered by traditional thermogravimetric analyses (TGA), is that liquid water retains all the heat it receives while it evaporates. The castable bulk is kept at a constant temperature even though its inlet surface is continuously heated. When the water stops boiling, the resulting thermal gradient rapidly disappears, causing a thermal shock in the castable structure. Figure 3. Dewatering behavior of a castable sample pre-treated at 1350C and saturated with water. The heating rate was 5C/min. (a) Fluid dynamic curve. (b) Thermal curve. In order to evaluate the influence of the castable’s permeability level on the removal of physical water, a new set of tests was carried out with a sample pretreated at 900C. Prior permeability characterization at room temperature according to Forchheimer’s equation (7) showed that this sample was less permeable than the one treated at 1350C (k1 = 0.33x10-15 m2 and k2 = 0.36x10-13 m for the sample at 900C and k1 = 1.51x10-15 m2 and k2 = 28.9x10-13 m for the sample treated at 1350C). Figure 4 shows the fluid dynamic and thermal changes that occurred during dewatering at 1C/min for the sample pre-treated at 900C. The trends are similar to Anais do 45º Congresso Brasileiro de Cerâmica 0803208 30 de maio a 2 de junho de 2001 - Florianópolis – SC. those shown in Figure 3 for the sample treated at 1350C, indicating that, for the samples analyzed, permeability plays a minor role in the dewatering of physical water when slow heating rates are applied. The only perceptible difference was the lower background flow rate level, reflecting the lower permeability of the sample pre-treated at 900C. Figure 4. Dewatering behavior of a castable sample pre-treated at 900C and saturated with water. The heating rate was 1C/min. (a) Fluid dynamic curve. (b) Thermal curve. Figure 5 reveals the severe thermal and mass loss consequences of a dewatering process in a less permeable castable when the heating rate increased from 1 to 5C/min. As shown in Figure 5-a, although the main peak was still located at 100C, steam was intermittently and chaotically released at temperatures up to 700C, and the airflow was even blocked for a few moments. However, Figure 5-b reveals that the more deleterious effect caused by the decrease in the permeability level was a noticeable increase of almost 300C in the thermal gradient between both castable surfaces during elimination of the main bulk of free water. Anais do 45º Congresso Brasileiro de Cerâmica 0803209 30 de maio a 2 de junho de 2001 - Florianópolis – SC. Figure 5. Dewatering behavior of a castable sample pre-treated at 900C and saturated with water. The heating rate was 5C/min. (a) Fluid dynamic curve. (b) Thermal curve. From this standpoint, this experiment provided one of the most important insights to understand the occurrence of explosive spalling in HAB refractory castables. If a large amount of physical water is retained within the refractory body when it is subjected to drying at rapid heating rates, the stress caused by the thermal gradient associated with the high pressure due to the entrapment of steam at higher temperatures may suffice to overcome the fracture stress, causing failures or even bursting the structure. Furthermore, in an actual dewatering process, the absence of airflow through the structure causes the steam to be released even more slowly and, depending on the heating rate applied, the amount of physical water retained and the body’s thickness, the thermal gradient generated between cold and hot faces may be decisive to the occurrence of failures in the castable structure. Lastly, this scenario is even more complex if one considers that the permeability level and the mechanical strength are even lower in a green sample subjected to drying and that the elimination of physical water may occur concurrently to the cement’s dehydration in most common castable compositions. Anais do 45º Congresso Brasileiro de Cerâmica 0803210 30 de maio a 2 de junho de 2001 - Florianópolis – SC. CONCLUSIONS This investigation demonstrated that the steam release that occurs during the dewatering process of refractory castables pretreated up to 1350ºC is dramatically affected by the permeability level and heating rate applied. The fluid dynamic analysis revealed that the main bulk of physical water is released as steam under saturated conditions at 100C. However, if the castable sample is subjected to a higher heating rate (5C/min), steam is generated faster than it can be released and water loss chaotically shifts to higher temperatures. Simultaneous thermal analyses showed that liquid water retains all the heat it receives while it evaporates. When the water stops boiling, the resulting thermal gradient rapidly disappears, causing a thermal shock in the castable structure. The less permeable the castable structure, the higher the steam entrapment and thermal shock. ACKNOWLEDGEMENTS The authors are grateful to the Brazilian research funding institutions FAPESP and CNPq, and to Alcoa S.A. and Magnesita S.A. for their support of this work. REFERENCES (1) W.E. Lee and R.E. Moore, “Evolution of in-situ refractories in the 20th century”, J. Am. Ceram. Soc., 81 [6], 1385-1410 (1998). (2) R.E. Moore; J.D. Smith; W.L. Headrick Jr and T.P. Sander, “Monolithic dewatering theory testing and practice: new challenges”, In: 32nd Annual Symposium on Refractories, The American Ceramic Society, St. Louis Section, 26p (1996). (3) W.H. Gitzen and L.D. Hart, “Explosive spalling of refractory castables bonded with calcium aluminate cement”, Bull. Am. Ceram. Soc., 40 [8] 503-507, 510 (1961). (4) Z.P. Bazant and W. Thonguthai, “Pore pressure and drying of concrete at high temperature”. Proc. of the Am. Soc. of Civil Engineers. 104[EM5] 1059-1079 (1978). Anais do 45º Congresso Brasileiro de Cerâmica 0803211 30 de maio a 2 de junho de 2001 - Florianópolis – SC. (5) D.L. Hipps and J.J. Brown Jr., “Internal pressure measurements for control of explosive spalling in refractory castables”. Am. Ceram. Soc. Bull. 63[7] 905-910 (1984). (6) M.D.M. Innocentini; M.G. Silva; B.A. Menegazzo and V.C. Pandolfelli, “Permeability of refractory castables at high-temperatures”, J. Am. Ceram. Soc. 84[3], 645-647 (2001). (7) A.R.F. Pardo, “Permeability of high-alumina self-flow refractory castables”, M.Sc. Thesis, Federal University of São Carlos, Brazil, 2000 (In Portuguese). (8) C. Ribeiro; M.D.M. Innocentini and V.C. Pandolfelli, “Dynamic permeability behavior during drying of refractory castables based on calcium-free alumina binders”, J. Am. Ceram. Soc., 84[1] 248-250 (2001). (9) M.D.M. Innocentini; C. Ribeiro; J. Yamamoto; A.E. Paiva; L.R.M. Bittencourt; R.P. Rettore and V.C. Pandolfelli, “A novel technique to access the drying behavior of refractory castables”, accepted for publication in the Bull. Am. Ceram. Soc. (2001).