See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/229622153

Drying Stages During the Heating of High-Alumina, Ultra-Low-Cement

Refractory Castables

Article  in  Journal of the American Ceramic Society · December 2004

DOI: 10.1111/j.1151-2916.2003.tb03438.x

CITATIONS READS

69 2,647

4 authors, including:

Murilo D.M. Innocentini Victor C Pandolfelli

Universidade de Ribeirão Preto Universidade Federal de São Carlos
162 PUBLICATIONS   2,165 CITATIONS    618 PUBLICATIONS   6,166 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Development and optimization of physical and/or chemical processes for recovery of potassium dissolved in vinasse effluent from sugar cane industry View project

Ceramic foam filters - Processing and Properties View project

All content following this page was uploaded by Victor C Pandolfelli on 13 October 2017.

The user has requested enhancement of the downloaded file.


journal
Drying Stages during the Heating of High-Alumina, Ultra-Low-Cement
Refractory Castables
Murilo D. M. Innocentini, Fábio A. Cardoso, Mario M. Akyioshi, and Victor C. Pandolfelli*
Department of Materials Engineering, Federal University of São Carlos, São Carlos, São Paulo 13565-905, Brazil
The purpose of this work was to investigate the drying kinetics
of high-alumina, ultra-low-cement refractory castables under
continuous heating conditions. Three main drying stages were
identified during the castable heat-up and were related to the
phase change of free water and to the decomposition of
hydrated products present in the body. A clear correlation was
found between the actual heating profile inside the castable
and the dewatering stages under various heating schedules.
Thermal analysis was used to assess the drying temperature
that represents the highest risk of steam pressure buildup and,
thus, of explosive spalling.
I. Introduction
R EFRACTORY castables are hydrated materials that require
special attention during their first heat-up. Depending on
the heating schedule applied, the steam that is generated may
become pressurized within the ceramic structure, occasionally
causing cracking or even explosive disintegration of the prod-
uct.1–5 For this reason, the dry-out schedules of preshaped
refractories are often divided into steps, aiming for safe removal
of both free casting water at around 100°C and cement
decomposition products at higher temperatures. However, the
choice of suitable heating rates and dwell times is based mostly
on empirical knowledge that takes into account the ceramic
composition, and the geometry and size of the product. In fact,
lack of reliable data on the duration and intensity of transfor-
mations that take place inside the heated structure make
long-term schedules preferable. This minimizes the risks of
explosive spalling but makes the dry-out of green castables an
expensive and time-consuming operation.
In this context, this study focused on the optimization of the
dewatering process of high-alumina, ultra-low-cement refractory
castable compositions. The drying kinetics of physically absorbed
water was assessed during the castable heat-up at various heating
rates. The water loss pattern was directly related to the temperature
gradients measured in the sample, allowing for the identification of
different drying stages and providing guidelines to optimize
refractory castable heating schedules.
II. Experimental Procedure
The castable composition chosen for this study consisted of a
mixture of 98 wt% alumina, 2 wt% calcium aluminate cement
(CAC; CA-14, Alcoa, Pittsburgh, PA), and 4.52 wt% water (dry
basis). Alcoa (Brazil and U.S.) supplied the matrix powders
W. E. Lee—contributing editor
Manuscript No. 186688. Received September 17, 2002; approved December 30,
2002.
Supported by the Brazilian research funding institution FAPESP, by Alcoa S.A.,
and by Magnesita S.A.
*Member, American Ceramic Society.
1146
J. Am. Ceram. Soc., 86 [7] 1146 – 48 (2003)
(calcined aluminas A1000 SG and A3000 FL, 22 wt%, dp ⬍ 100
␮m, where dp is the particle diameter) and the aggregate grains
(white fused aluminas, grades 4/10, 8/20, 10/36, 20/40, and 200F,
76 wt%, dp ⬍ 4.5 mm). The particle size distribution of this
composition was adjusted to a theoretical curve based on Andreas-
en’s packing model to obtain a potentially self-flow castable, with
a coefficient of distribution (q) of 0.21.
Samples were cast as 4 cm diameter ⫻ 4 cm thick cylinders,
cured in the mold at 50°C (relative humidity of ⬃100%) for 48 h
and heat-treated for 6 h at 800°C (heating rate of 1°C/min). During
casting, two thin K-type thermocouples (diameter of 0.2 mm) were
inserted into some of these samples at a height of 20 mm, the first
at a depth of 1 mm (lateral surface) and the second at 20 mm
(center of the body). To focus the analysis on the removal of
physically absorbed water, the samples were saturated with dis-
tilled water under a vacuum for 2 h before each test.
Dewatering tests were conducted in an electric furnace con-
trolled by a proportional-integral-derivative (PID) system to a
maximum temperature of 700°C and according to various heating
schedules. The test sample was suspended in the center of the
furnace to enable heat to reach every face of the body simulta-
neously. No forced air convection was provided in the furnace
during the heat-up.
In the first set of experiments, the temperatures at the surface
and center of the saturated specimen were monitored during each
heating program. New tests were then conducted with similar
samples without thermocouples to monitor the sample’s water
loss under the same heating conditions. Data on mass and
temperature were computer-recorded at 5 s intervals throughout
the experiments.
Mass loss during drying was assessed through the normalized
parameter W, which measures the cumulative fraction of water
expelled during the heat-up per total amount of water initially
present in the body. W varies from 0%–100% and is defined as:
W共%兲 ⫽ 100 ⫻ 冉 Mo ⫺ M
Mo ⫺ Mf 冊 (1)
where M is the instantaneous mass recorded at time ti during the
heating stage, Mo the initial mass, and Mf the final mass of the
tested sample.
The experimental drying rate during heat-up was obtained
through the derivative dW/dt by
冉 冊
dW
dt i
⫽
Wi⫹1 ⫺ Wi⫺1
ti⫹1 ⫺ ti⫺1
(2)
III. Results and Discussion
(1) Drying Stages
The dewatering of a moistened porous medium under ambient
conditions occurs in stages that are driven by different transport
mechanisms.6 – 8 Initially, water evaporates adiabatically from the
surface of the body in a stage known as the constant rate period
(CRP).6 In this stage, the evaporation rate is greatly influenced by
external conditions, such as temperature and partial vapor pres-
sure. In an isothermal environment, the evaporation rate remains
July 2003 Drying Stages of High-Alumina, Ultra-Low-Cement Refractory Castables
constant as long as the transport forces maintain the flux of water
from the pores toward the body’s surface. However, as drying
proceeds, the evaporation front eventually recedes into the inter-
particle voids, when the water transport, initially occurring by fluid
flow, becomes driven by vapor diffusion through increasingly
longer path lengths. This process is characterized by gradual
mass-loss rate decreases with time, in a sequence of two stages
known as the falling rate periods (FRP).6
To ensure complete removal of moisture and hydrate decom-
position, the dry-out of refractory castables must be conducted at
temperatures above the normal boiling point of water. Therefore,
in the range of ambient temperature to 100°C, the drying process
usually occurs at a constant heating rate rather than under
isothermal conditions, and the drying stages, as previously de-
scribed, undergo remarkable changes.
The influence of continuous heating on the drying kinetics of a
moistened castable is shown in Fig. 1. Instead of the typical CRP
and FRP established for an isothermal environment, dewatering
occurs in three stages, identified by the peaks in the drying rate
curve during the furnace heat-up.
Although peaks appear to shift to higher temperatures as the
heating rate increases, the two first drying stages shown in Fig. 1
actually involve the elimination of physically absorbed water. The
cause of this misleading result is that the temperature of reference
for the peaks’ location is set in the furnace rather than in the
sample.
Thus, a much more consistent pattern is observed when the
drying profiles involve the actual castable temperature, as illus-
trated in Fig. 2. The temperature range for each drying stage
becomes clear, and the main influence of the heating schedule is on
the intensity and no longer on the location of the peaks.
In the first stage, which takes place from ambient temperature to
⬃100°C, the main drying mechanism is evaporation of free water,
which also occurs under adiabatic conditions. However, both CRP
and FRP are affected by the temperature increase. Thus, although
the evaporation rate is enhanced by the greater water vapor
pressure, it is simultaneously reduced by the progressive retreat of
the drying front into the porous network. The temporary equilib-
rium between these two opposite effects is represented by the first
peak in the drying rate curve at 50°– 60°C.
As the castable continues to be heated, the temperature at the
liquid/vapor interface eventually reaches the boiling point, and the
second important drying stage begins with the ebullition of free
water. At this stage, the generation of steam is substantially
enhanced by increase in the castable temperature, when the steam
pressure becomes the main driving force for the flux of mass
exiting the body. The peak in the range of 100°-170°C results from
the balance between the increase of the rate at which steam is
eliminated and the progressive shrinkage of the drying front
toward the center of the body.
The third stage in the drying profiles of Fig. 2 occurs in the
temperature range of 200°– 400°C and is apparently not related to
free water, but instead to hydrated products formed during the
moistening of the sample or even during the dewatering process.
Fig. 1. Drying profiles as a function of the furnace temperature for
various heating schedules.
1147
Fig. 2. Drying stages related to the actual sample temperature for various
furnace heating schedules.
For the composition tested here, the pretreatment temperature of
800°C is insufficient to start the sintering process. Hence, the
calcined phases of the calcium aluminate cement and the fine
alumina particles in the castable matrix can still be partially
reactive, rehydrating when in contact with liquid water or with the
pressurized steam generated in the previous drying stages.
The aforementioned hypothesis that compares the dewatering
profiles of three different samples heated at 10°C/min to 700°C is
corroborated in Fig. 3. Sample A refers to a green, 2 wt% CAC
castable cured in the mold at 50°C and tested with its original
moisture content; sample B, with a composition similar to that of
sample A, was prefired at 800°C and then moistened with distilled
water for 2 h under vacuum; sample C refers to a cement-free
composition also prefired and moistened before the test.
As depicted in Fig. 3, the lack of a third drying stage in the
cement-free composition confirms that dehydration is in fact the
cause of the minor water loss between 200°– 400°C for both
green and fired CAC-containing samples. According to the
literature, this temperature range is typical for the decomposi-
tion of gibbsite (Al(OH3)) and hexahydrated tricalcium alumi-
nate (3CaO.Al2O3.6H2O).10
Figure 3 also shows some changes in the drying rates in the two
first stages according to the type of sample tested. The fact that
evaporation and ebullition occur in consecutive stages during the
drying process implies that as less water is evaporated, more of it
remains during ebullition and vice versa. For this reason, the
intensities of peaks in both stages usually display opposite trends
when a variation occurs in the heating rate (as seen Fig. 2) or in
some property of the material. In the case of curves in Fig. 3, the
differences in the permeability level among the samples would
explain the different drying rates observed in each stage. For
instance, the green sample, displaying a lower permeability level
than the prefired one, had a higher resistance to dewatering in the
initial stage (lower evaporation peak), consequently leaving a
larger amount of water to be boiled (higher ebullition peak).
Fig. 3. Drying profiles for CAC-containing and cement-free samples at a
furnace heating rate of 10°C/min.
1148 Journal of the American Ceramic Society—Innocentini et al.
Nevertheless, the similarities among the drying profiles of the
different tested samples confirmed that removal of unbound water
should, in fact, be the main concern during the heat-up process of
green ultra-low-cement compositions, requiring further attention to
avoid excessive pore pressurization and explosive spalling.
(2) Thermal Changes during Dewatering
Dewatering is associated with important thermal transforma-
tions that take place inside the heated castable. Although part of
the incoming heat is conducted into the body through the solid
phase, some of it is also absorbed by the liquid water retained in
the pores, first enhancing evaporation under 100°C and then
promoting boiling above that temperature.9
The profiles in Fig. 4 indicate that these endothermic phase
change processes are remarkably intense during the first and
second dewatering peaks, causing a temporary slowdown in the
heating rate inside the castable. This effect is further accentuated
by the flux of steam leaving the body, which decreases the
efficiency of heat absorption and increases the thermal gradient
between the body and the surrounding atmosphere. The result is
that the end of both dewatering stages is immediately followed by
an acceleration of the castable heating rate.9
The heating schedule also has a marked effect on the thermal
changes inside the castable, as can be observed in Fig. 4. For the
sample size tested here, the furnace heat-up at 1°C/min is suffi-
ciently slow to produce no important internal thermal gradient,
with both the center and surface of the sample becoming equally
heated during the dewatering stages.
This behavior changes with heating rates of 5°C/min and
10°C/min, especially during the second stage when there is a
temporary discrepancy between the heating profiles at the surface
Fig. 4. Correlation found between the actual heating profiles inside the
castable and the dewatering stages for various furnace heating schedules
((a) heating rate of 1°C/min, (b) heating rate of 5°C/min, and (c) heating
rate of 10°C/min).
View publication stats
Vol. 86, No. 7
and center of the body. In this case, the abrupt increase in the
heating rate at the center signals the end of the boiling process and
of the second dewatering stage, reestablishing similar heating
profiles for the center and surface for the remainder of the test.
Therefore, although heated to over 100°C, the body retains
water in liquid form during most of the second drying stage. This
may occur if the structure’s permeability is sufficiently low to
increase the vapor pressure inside the pores, shifting the boiling of
water to higher temperatures. This effect is dangerously enhanced
during the time interval in which the discrepancy in the heating
profiles inside the body is maintained, when the heating rate
increases rapidly on the surface while water is still boiling in the
center.
Figure 4(c) shows that, for the castable heated at 10°C/min, the
critical displacement interval refers to a temperature increase from
160°–220°C. According to Antoine’s equation,4 this implies a
theoretical steam pressure buildup from 0.6 –2.3 MPa, which can
be sufficiently high to cause the explosion of a green castable.
These values obviously refer to the maximum pressure buildup
achievable in the center of the body if no release is provided for
steam. In practice, the castable’s permeability and the heating
schedule determine the amount of water retained in the pores and,
hence, the actual intensity of the pressure buildup. Nevertheless,
the values calculated by Antoine’s equation for the critical dewa-
tering interval represent the upper limit for steam pressurization
and may be reasonably compared with the mechanical tensile
strength of green castables with the purpose of evaluating the risk
of explosive spalling.
IV. Conclusions
This investigation demonstrated that the dewatering process of
a continuously heated high-alumina, ultra-low-cement castable
occurs in stages and in different temperature ranges. The two first
stages involve the removal of free water, first by evaporation under
100°C and by ebullition thereafter. The third stage takes place at
higher temperatures and involves the elimination of hydrated
phases in the castable matrix. Thermal analysis revealed that the
sample’s center and surface were both remarkably sensitive to
endothermic processes involving the conversion of water into
steam and to the heating schedule applied. The heating profile
inside the castable is therefore clearly related to the dewatering
stages, providing an indirect assessment of the maximum steam
pressure buildup developed inside the pores, thus guiding the
choice of heating schedules that minimize the risk of explosive
spalling.
References
1
M. Velez, A. Erkal, and R. E. Moore, “Computer Simulation of the Dewatering of
Refractory Concrete Walls,” Taikabutsu Overseas, 20 [1] 5–9 (2000).
2
R. E. Moore, J. D. Smith, W. L. Headrick Jr., and T. P. Sander, “Monolithic
Dewatering Theory Testing and Practice: New Challenges”; presented at the 32th
Annual Symposium on Refractories, St. Louis, MO, Mar. 1996. American Ceramic
Society, St. Louis Section, 1996.
3
M. D. M. Innocentini, C. Ribeiro, J. Yamamoto, A. E. M. Paiva, L. R. M.
Bittencourt, R. P. Rettore, and V. C. Pandolfelli, “Drying Behavior of Refractory
Castables,” Am. Ceram. Soc. Bull., 80 [11] 47–56 (2001).
4
D. A. Bell, “First Heat-up of Ladle Linings”; pp. 292–94 in UNITECR ’97
Proceedings. American Ceramic Society, Westerville, OH, 1997.
5
P. Kalifa, F. D. Menneteau, and D. Quenard, “Spalling and Pore Pressure in HPC
at High Temperatures,” Cem. Concr. Res., 30, 1915–1927 (2000).
6
G. W. Scherer, “Theory of Drying,” J. Am. Ceram. Soc., 73 [1] 3–14 (1990).
7
W. C. J. Wei and C. L. Hsieh, “Drying Kinetics of Ultrafine Alumina Cake with
Drying Control Chemical (DCC),” J. Ceram. Soc. Jpn., 107 [4] 313–17 (1999).
8
K. Oda, H. Itoh, T. Matsumoto, and Y. Ono, “Drying Characteristic of High-
Dense Castable,” Taikabutsu Overseas., 13 [2] 21–28 (1992).
9
M. D. M. Innocentini, C. Ribeiro, L. R. M. Bittencourt, and V. C. Pandolfelli,
“Fluid Dynamics and Thermal Aspects of the Dewatering of High-Alumina Refrac-
tory Castables: Removal of Physically Absorbed Water,” J. Am. Ceram. Soc., 84 [11]
2696 –98 (2001).
10
N. Schmitt, J. F. Hernandez, V. Lamour, Y. Berthaud, P. Meunier, and J. Pourier,
“Coupling Between Kinetics of Dehydration, Physical and Mechanical Behaviour of
High Alumina Castable,” Cem. Concr. Res., 30, 1597– 607 (2000). 䡺