high-alumina multifunctional refractory castables

8 downloads 4176 Views 231KB Size Report
originally designed PSD. Fast-mixing castables that require no high-energy mixers are usually the most suitable ones. Castable mixing was accomplished in this ...
HIGH-ALUMINA MULTIFUNCTIONAL REFRACTORY CASTABLES A.R. Studart, R.G. Pileggi, J. Gallo♦, V.C. Pandolfelli DEMa - Department of Materials Engineering, UFSCar - Federal University of São Carlos, São Carlos, SP, Brazil ♦

Alcoa Aluminum S.A., Brazil

Two of the main requirements for refractories used in steelmaking and in related industries are high performance and easy installation. In order to meet these needs, the thermomechanical properties of high-alumina castables have been significantly improved through the reduction of liquid-phase forming oxides

(1)

and several techniques have been

developed to facilitate and automate the installation process of refractory linings (2,3). When an installation method is chosen, the aim is not only to select one that minimizes the number of flaws introduced in the refractory but also the technique that best fits the equipment in which the castable is to be applied. Several features may determine what the most suitable technique to be used is, including the size and type of equipment to be repaired, the lining region, easy access to this region, the availability of trained workers and special apparatus (for vibrating, pumping, gunning, etc.), the cost involved, and mold shape, among others. A large number of categories of castables have been developed to fit this wide variety of features, of which the main ones are ramming, vibrating, self-flowing and gunning castables. In order to properly line and repair the great diversity of refractory-consuming equipment, a significant number of different castables usually has to be kept in storage on a company’s premises. In addition to the high cost involved, the storage of these materials for long periods may affect some of their properties, especially those containing cement or other hydraulic binders that may deteriorate over time. This need has led to a growing demand for more versatile castables in recent years

.

(4)

Expectations focus on the development of a type of multifunctional castable that can be installed using most of the current methods and changing only the water content added during mixing. Another advantage of a multifunctional castable would be to offer the user greater freedom in selecting the placing technique that meets his needs in terms of time, cost and performance. Paper to be presented in 103rd Annual Meeting of the ACS – Indianapolis - 2001

Considering the fact that each class of castable is usually characterized by a specific particle size distribution (PSD), the main objective of this paper is to establish a PSD that can render multifunctional features to castables. After selection of the PSD, the next goal was to determine whether the versatility of multifunctional castables during the installation process extended to benefits during the mixing step prior to application and during exposure to high temperatures under load. This was accomplished through a comparison of the mixing characteristics, rheological behavior and creep resistance at high temperature of multifunctional castables with those of castables installed with specific application techniques. Preparation of castables Andreasen’s packing model was used to design the different PSDs evaluated. Three q values, 0.21, 0.26 and 0.31, were chosen for Andreasen’s equation. The PSDs of castables were adjusted to the theoretical packing curves using different combinations of white fused alumina as aggregates and calcined aluminas as fine particles (matrix). In fact, only the ratio (aggregates)/(matrix) was changed to reach the different q values. As described by Andreasen’s equation, the amount of matrix increases as the q value decreases. All the compositions were prepared with 1 wt% of calcium aluminate cement (CA 270). Citric acid was used as an additive to properly disperse particles

(2)

and to control

cement setting. An optimum citric acid content of 0.36 mg/m2 was previously established and used thereafter in all the castables. Compositions with q values of 0.21 and 0.26 were prepared using total water contents of 14, 15 and 16 vol%. Smaller amounts of water were used to prepare castables with q = 0.31 (13, 14 and 15 vol%), chiefly owing to their reduced matrix content and specific surface area (SSA). Mixing process The mixing process significantly influences the rheological properties of castables during installation

. Therefore, the rheology determines not only the power of the mixers

(5)

that should be used but also the efficiency of the installation process. When castables are not properly mixed, particle agglomerates remain in the matrix, making the adsorption of dispersants around particles difficult and preventing the castables from achieving the

originally designed PSD. Fast-mixing castables that require no high-energy mixers are usually the most suitable ones. Castable mixing was accomplished in this work using the two-step water addition method suggested elsewhere by the authors

(5)

. The water content added in the first step (at a

constant rate of 8 g/s) varied according to the composition. Contents of 11, 12 and 13 vol% were used in this first step for castables with q values of 0.31, 0.26 and 0.21, respectively. These water contents correspond to the amount of liquid necessary to cover the surface of the particles and to fill in the interparticle pores. This condition maximizes the capillary forces between particles, leading to the mixing stage known as the castable turning point

.

(5,6)

Castables displaying superior SSA are expected to require higher water contents to reach the turning point, which explains the tendency observed in this study for compositions with varying q values. The second fraction of water (2 nd step) was introduced into the castables only after a maximum torque value was attained (5). The mixing process was evaluated with the help of a rheometer to detect torque variations as a function of time during the addition of water. The torque values obtained indicate the castables’ resistance to mixing and shearing. Although it does not correspond to an energy measure, the area under the curves of torque as a function of time was interpreted as an indication of the energy the mixer must apply to accomplish mixing. Results reveal that both torque level and “mixing energy” are reduced with the increase of the q value. This tendency may be attributed to the reduction of the castable SSA as the q value increases. The SSA is directly related to the number of contacts between particles in the castable, which determines the number of liquid bridges built up among particles at the castable turning point and, thus, the magnitude of attractive capillary forces in this condition. Because the magnitude of capillary forces is directly proportional to a castable’s resistance to mixing/shearing, the superior torque levels obtained for compositions with lower q values can be ascribed to the higher surface area of these castables. It can be observed that the coefficient q also affects the mixing time necessary to achieve the castable’s turning point and, therefore, the total duration of mixing. This also seems to be related to the castable SSA. As the SSA value increases (q decreases), the amount of water required to cover the particles’ surfaces also increases. Thus, longer periods are necessary to add the water content and properly distribute it throughout the castable. Another fact that may have contributed toward the reduction of the mixing time of compositions with higher q values is the presence of a larger amount of coarse particles in

these castables, which are expected to facilitate the breakup of agglomerates through a kind of “ball milling effect” (5). Rheological behavior The rheological behavior of castables prepared with distinct q values and water contents was first evaluated by measuring the free-flow value obtained according to the procedure adapted from the ASTM C-860 standard (2). To facilitate our analysis of the results obtained, three different free-flow value ranges were arbitrarily selected and related to different classes of castables. The self-flow behavior is usually characterized by free-flow values in the range of 80 to 110%. Castables that require vibration during installation, on the other hand, may present free-flow values between 30 and 80%. Finally, compositions presenting free-flow values of less than 30% are usually installed using more energetic techniques such as ramming. It is worth noting that castables exhibiting free-flow values above 110% may be susceptible to segregation, since the matrix density may not be sufficiently high to prevent the sedimentation of coarse particles. Segregation in this case can only be prevented if the castable matrix displays a markedly dilatant behavior. It has been verified that the castable free-flow value can be correlated to the mean distance between aggregates

. As this distance increases, less interference between

(7)

aggregates may be expected during the castable flow, leading to higher free-flow values. The MPT (Maximum Paste Thickness) parameter was used in this study to indicate the mean distance between aggregates

. The castable matrix content, the PSD and the SSA of

(7)

aggregates, as well as the amount of water used during castable preparation, were used to calculate the MPT values for the compositions evaluated. Different MPT ranges were obtained for the castables prepared, chiefly due to their distinct matrix contents. Assuming that the distance separating aggregates was the main influencing factor for castable flowability, the free-flow values obtained were analyzed according to the MPT range calculated for each composition. The q = 0.31 composition presented the lowest free-flow values, probably as a consequence of its reduced MPT range (72.3 – 79 µm). The flowability values obtained in this case suggest that such castables would require energetic methods, such as ramming, to be properly installed, regardless of the water content utilized. Castables with q = 0.21, on the other hand, displayed MPT values ranging from 132.6 to 140.5 µm, resulting in a markedly higher free-flow range. In this case, castables may be

installed either as a self-flow or a vibrating composit ion by simply changing the amount of water added (15 and 14 vol%, respectively). The addition of 16 vol% of water to this composition would likely favor segregation in the castable due to the reduced matrix density. It can also be inferred, from the free-flow curve obtained, that the reduction of the water content to 13 vol% (turning point amount), would not allow for the production of ramming castables with a coefficient q at 0.21. Finally, castables whose coefficient was equal to 0.26 presented intermediate MPT values (105.0 – 112.4 µm), leading to free-flow values between 24 and 82%. This wide range of flowability covers the free-flow values expected for the three types of installation techniques. Therefore, castables with a coefficient q = 0.26 may be installed either as a ramming, vibrating or self-flow composition by simply adjusting the water content between 14 and 16 vol%. This feature fulfills the primary requirements used to define a multifunctional composition, suggesting that the coefficient q = 0.26 is the most appropriate one to obtain this class of castable. Castables with q = 0.26 were also evaluated in terms of their pumping ability with the help of the rheometer previously used to investigate the mixing process

. In order to

(8)

simulate the shearing conditions imposed during pumping, castables prepared with different water contents (q = 0.26) were subjected to a 2 to 75 rpm shearing cycle in the rheometer. The results obtained reveal that the increase of the castable water content from 14 to 16 vol% significantly reduces the torque level needed to keep a certain rotation constant. A comparison of these results with the rheological behavior shown by commercial pumping castables revealed that the addition of 16 vol% of water into castables with q = 0.26 reduces the torque level to a range of values very similar to those at which optimum pumping conditions are achieved

(9)

. It should also be noted that castables with 16 vol% (q = 0.26)

display a slightly dilatant behavior under high shearing conditions, which would be unsuitable for pumping. However, further experiments with this composition showed that this dilatant tendency is easily eliminated by slightly adjusting the PSD matrix. The results obtained with regard to pumping characteristics strengthen the multifunctional feature of castables prepared with a coefficient q equal to 0.26. Creep behavior The free-flow results shown in previous sections revealed that castable flowability is enhanced when the distance between aggregates is increased (higher MPT). High matrix

contents (fine particles) are usually required to attain large distances between coarse particles. However, fine particles are known to undergo greater creep rates at high temperature than coarse particles, mainly due to grain sliding and several other mechanisms involving grain boundaries. Therefore, it is worth investigating whether the greater amount of matrix (inferior q values) required to achieve high flowability during installation would degrade the castable creep resistance at high temperature. To gain a better understanding of this aspect, the creep rate of castables prepared with distinct q values (matrix contents) and a fixed amount of water (14 vol%) was evaluated. Creep measurements were taken from unfired cylindrical samples with an external diameter of 50 mm and a central hollow of ~ 10 mm. These tests were carried out at 1600ºC using a compressive load of 0.2 MPa. The results of creep of the unfired samples demonstrate that increasing the matrix content (and MPT), which was accomplished by decreasing Andreasen's q value, actually increased the creep rate of castables at a high temperature. This effect appears to be more pronounced when the matrix content changes from 28 to 35 vol% (q from 0.31 to 0.26), whereas a slight increase in the creep rate is observed when the matrix amount is further increased to 42 vol% (q from 0.26 to 0.21). In principle, these results suggest that castables that would be easier to install after the mixing operation (higher matrix content and lower q) and would exhibit inferior creep resistance at high temperature. In other words, it can be stated that castables that flow properly during installation would also be expected to “flow” (or deform) at working temperatures. One must, nevertheless, keep in mind that castables with a higher matrix content (higher SSA) are also more likely to sinter and densify during firing. Therefore, densifying and coarsening mechanisms are expected to be more effective in castables with lower q values. This effect would probably compensate for the presence of a larger amount of fine particles in these castables by generating a coarser microstructure after firing. This hypothesis was investigated by evaluating the creep behavior of samples with different q values previously fired at 1650ºC for 24 hours. The results showed that the compositions displayed very similar creep rates after being subjected to this pre-firing treatment, confirming the above assumption. Porosity measurements before and after firing also corroborated this hypothesis, since a greater reduction in porosity was observed for castables with lower q values (higher matrix content). This suggests that densifying and coarsening mechanisms were probably more efficient in these compositions, leading to coarser microstructures after firing.

Based on these results, one can conclude that the distinct creep behavior at high temperatures expected for castables with different q values does indeed occur, but only in the first hours after the beginning of the firing process. During firing for 24 hours, the microstructures of castables develop in different ways so that a similar creep resistance is finally verified. Castables with q = 0.26, prepared with a higher water content (16 vol%), were also evaluated in terms of their creep behavior after firing samples at 1650ºC (24 hours). No significant difference was observed in these specimens in comparison with the composition prepared with 14 vol% water. This suggests that the water content also does not apparently affect the creep resistance of pre-fired castables. Therefore, despite its higher water content (16 vol%), a multifunctional castable (q=0.26) may be installed with self-flow characteristics and still exhibit creep behavior similar to that of a typical self-flow composition (q = 0.21 and 14 vol% water). Concluding remarks This paper demonstrated that PSD markedly affects the rheological behavior of castables, determining the techniques recommended for use during installation. PSD curves based on Andreasen's packing model with coefficient q equal to 0.26 allowed for the production of “so-called” multifunctional castables. These castables can be installed using different methods (ramming, vibrating, self-flowing) simply by changing the water content used during their preparation. This feature fulfills the primary requirement for more versatile castables concerning to the installation technique. In addition to these several installation methods, further rheological measurements revealed that multifunctional castables (q = 0.26) are also suitable for pumping. The Andreasen coefficient q also influences the castable’s behavior during the mixing process. Castables with lower q values (~ 0.21) usually require high-energy mixers and longer preparation times, whereas compositions with higher q values (~ 0.31) appear to mix more easily and rapidly. Due to the intermediate q value (0.26) of multifunctional castables, their preparation is quite fast and, in principle, requires no high-energy mixers. The creep behavior of unfired castables was also affected by the coefficient q. The inherently higher matrix content (higher MPT) of compositions with lower q values led to a superior creep rate of the unfired specimens at high temperatures. However, when subjected to a pre-firing treatment at 1650ºC (24 hours), castables with distinct q values exhibited very similar creep resistance. Therefore, a similar behavior is expected for these castables during

most of their working life. These results suggest that, in addition to their advantages during installation, multifunctional castables are also likely to display high temperature behavior similar to that of compositions normally installed using specific methods (i.e., ramming, vibration). Acknowledgements The authors are grateful to the Brazilian research funding agencies FAPESP and CNPq and to ALCOA-Brazil for their support to this research . References 1. A.R. Studart and V.C. Pandolfelli; “Thermomechanical Behavior of Zero-Cement, HighAlumina Castables”, Am. Ceram. Soc. Bull., 79 [10] 53-60 (2000). 2. A.R.Studart, W.Zhong, V.C.Pandolfelli; “Rheological Design of Zero-Cement Self-Flow High-Alumina Refractory Castables”, Am. Ceram. Soc. Bull., 78 [5] 65-72 (1999). 3. N. Cassens, R.A. Steinke, R.B. Videtto; “Shotcreting Self-Flow Refractory Castables”, in Proceedings of the UNITECR’97, New Orleans, USA, p. 531-544 (1997). 4. J. Mosser and G. Karhut; “Refractories at the Turn of the Millennium” in Proceedings of the UNITECR’99, Berlin, Germany, p. 25-30 (1999). 5. R. G. Pileggi, A. R. Studart, J. Gallo, V. C. Pandolfelli; “Mixing Effects on the Rheology of Refractory Castables”, submitted to the Am. Ceram. Soc. Bull. (2000). 6. A.R. Studart, W. Zhong, R.G. Pileggi, P. Bonadia, V.C. Pandolfelli; “Dispersion of Microsilica-Containing Zero-Cement High-Alumina Castables”, Am. Ceram. Soc. Bull., 79 [2] 49-55,84 (2000). 7. P. Bonadia, A.R. Studart, R.G. Pileggi, S.L. Vendrasco, V.C. Pandolfelli; “Applying MPT Principle to High-alumina Castables”, Am. Ceram. Soc. Bull., 78 [3] 57-60 (1999). 8. R.G. Pileggi, A.E. Paiva, J. Gallo, V.C. Pandolfelli; “ Novel Rheometer for Refractory Castables”, Am. Ceram. Soc. Bull., 79 [1] 54-58 (2000). 9. R.G. Pileggi, A.R. Studart, C. Pagliosa, V.C. Pandolfelli; “Rheological Behavior of Pumpable Refractory Castables” (in Portuguese), presented in the 44 th Brazilian Ceramic Conference, Águas de São Pedro-S.P., 12 p. (2000).

TABLES: Table I: Content of raw materials, total specific surface area and MPT range for compositions prepared. © Alcoa/Brazil,

grade EK-8R, mesh size: from 4/10 to –200.

³ Alcoa/U.S.A. ª

Calculated from the specific surface area of individual raw materials based on BET experimental data

(Gemini, Micromeritics, model 2370). ² Calculated

(7)

according to the model proposed by Bonadia et al. .

Table II: Apparent porosity of castables (14 vol-% water) before and after firing at 1650ºC (24h). A parameter defined as the degree of densification was calculated in order to account for the percentage change in castable porosity during firing. © Degree

of densification = 100%×[(PBF - PAF)/PBF]

FIGURES

Figure 1: Particle size distributions (PSDs) of castables and the respective theoretical (Andreasen) curve to which they were adjusted. Figure 2: Torque applied to castables as a function of time during the mixing proce ss (rotation set at 33 rpm). The inset table shows the water content required to achieve the castable turning point (indicated by arrows), as well as the area under the curves obtained. Figure 3: Free-flow of castables as a function of water content for different q values, indicating the multifunctional behavior of compositions with q = 0.26. The figure was arbitrarily divided into regions where castables are expected to present ramming, vibrating and self-flowing characteristics. Figure 4: Torque applied to castables (q = 0.26) with different water contents when subjected to a shearing cycle in the rheometer. The figure also indicates (blue region) the level of torque normally obtained with commercial pumping castables under optimum conditions (9).

Figure 5: Creep rate of castables at 1600ºC (under a compressive load of 0.2 MPa) for samples subjected or not to a pre-firing treatment at 1650ºC (24h). Figure 6: Creep rate of castables at fixed temperature, compressive load and strain (1600ºC, 0.2 MPa and 0.4%, respectively) as a function of the matrix volume and MPT value obtained from the original PSD composition.

TABLES: Table I: Content of raw materials, total specific surface area and MPT range for compositions prepared. Vol-% (wt-%) Aggregates

q = 0.21

q = 0.26

q = 0.31

75.1 (75.2)

79.2 (78.3)

83.3 (82.6)

A-1000 SG

8.4 (8.4)

6.9 (6.9)

5.5 (5.5)

A-3000 FL

15.3 (15.4)

12.6 (13.7)

10.0 (10.9)

1.25 (1.00)

1.20 (1.00)

1.24 (1.00)

8.37

7.51

6.55

13 vol-% of water

-

-

72.3

14 vol-% of water

132.6

105.0

75.6

15 vol-% of water

136.5

108.6

79.0

16 vol-% of water

140.5

112.4

-

White fused alumina Calcined

Matrix

aluminas

³ ³

Cement CA 270 Total specific surface area

MPT

² ²

© Alcoa/Brazil,

( µm)

ª ª

© ©

³ ³

(m 2 /g)

grade EK-8R, mesh size: from 4/10 to –200.

³ Alcoa/U.S.A. ª

Calculated from the specific surface area of individual raw materials based on BET experimental data

(Gemini, Micromeritics, model 2370). ² Calculated

(7)

according to the model proposed by Bonadia et. al. .

Table II: Apparent porosity of castables (14 vol-% water) before and after firing at 1650ºC (24h). A parameter defined as the degree of densification was calculated in order to account for the percentual change in castable porosity during firing. Composition

© Degree

Apparent porosity (%)

Degree of densification (%)

Before firing (P BF )

After firing (P AF )

q = 0.21

11.95

8.22

31.2

q = 0.26

11.89

9.28

22.0

q = 0.31

11.94

9.98

16.4

of densification = 100%×[(PBF - PAF)/PBF]

© ©

FIGURES

100

CPFT (%)

Theoretical curves:

10

q = 0.21 q = 0.26 q = 0.31

Compositions: q = 0.21 q = 0.26

1

q = 0.31

0.1 0.01

0.1

1

10

100

1000

10000

Diameter (µm) Figure 1: Particle size distributions (PSD) of castables and the respective theoretical (Andreasen) curve to which they were adjusted.

30 q

Torque (N.m)

25 0.21 0.26 0.31

20

Water vol% at turning point 13 12 11

Area under the curve (N.m.s) 3796 2766 944

q = 0.21

15 0.26

10 5

0.31

0 0

50

100

150

200

250

300

350

Time (s) Figure 2: Torque applied to castables as a function of time during the mixing process (rotation fixed at 33 rpm). The inset table shows the water content necessary to achieve the castable turning point (indicated by arrows), as well as the area under the curves obtained.

Free-flow (%)

140 120

Susceptible to segregation

100

Self-flowing

q value: 0.21

80 60

0.26

Vibrating

0.31

40 20

Ramming

0 12

13

14 15 Water content (vol%)

16

Figure 3: Free-flow of castables as a function of water content for different q values, indicating the multifunctional behavior of compositions with q = 0.26. The figure was arbitrarily divided into regions where castables are expected to present ramming, vibrating and self-flowing characteristics.

10 q = 0.26

Torque (N.m)

8

Optimum condition (7) for pumping (9)

Water content (vol%) 14

6

15 16

4 2 0 0

20

40

60

80

Rotation (rpm) Figure 4: Torque applied to castables (q = 0.26) containing different water contents, when subjected to a shearing cycle in the rheometer. The figure also indicates ( blue region) the level of torque normally obtained with commercial pumping castables under optimum conditions (9).

1.E-05 10-5

q = 0.26

1.E-06

-1

Creep rate (s )

q = 0.21

10-6

q = 0.31 1.E-07 10-7

1.E-08 10-8 With pre-firing

Without pre-firing

-9

10 1.E-09

0

0.25

0.5

0.75

1

1.25

1.5

Strain (%) Figure 5: Creep rate of castables at 1600ºC (under a compressive load of 0.2 MPa) for samples subjected or not to a pre-firing treatment at 1650ºC (24h).

Volume of matrix (%) 28

30

32

34

36

38

40

42

Creep rate (s

-1

)

1.E-05 10 -5

1.E-06 10 -6

With pre-firing Without pre-firing Pre-firing at 1650ºC (24h)

1.E-07 10 -7

10 -8 1.E-08 70

80

90

100

110

120

130

140

MPT (µm) Figure 6: Creep rate of castables at fixed temperature, compressive load and strain (1600ºC, 0.2 MPa and 0.4%, respectively) as a function of the volume of matrix and MPT value obtained from the original PSD composition.