1. Introduction
Ceramics are a class of inorganic and non-metallic materials that are obtained after heat treatment at high temperatures [
1]. These materials are obtained from firing clay in a specific format (tiles, bricks, blocks, floors, artistic and decorative artifacts, etc.) and are widely used in civil construction [
2].
Depending on the type of product, clay is mixed with other raw materials, such as kaolin, feldspar rocks (pegmatite, granite, and leucogneiss), phyllite, talc, limestone, zirconium and quartz [
3,
4,
5]. Such mixtures are intended to obtain a mass that provides the desired properties for each product type, as well as guaranteeing its quality (absence of defects) and reducing energy costs during processing.
In Brazil, the ceramic sector is formed by a large number of industries with different levels of technological development and productive capacities. Although some companies stand out in the sector, it is characterized by a considerable number of micro and small companies, generally with a family structure, distributed throughout the country [
4]. As for the technological level, some have a high degree of technological development throughout the production process. However, this is not the general rule, and many companies are still at a very primitive stage in terms of modernization, with outdated and inefficient production processes and technology [
6].
While the red ceramic sector is characterized by a decentralized industrial structure and national financial investment, the sanitary ware segment is concentrated in a few companies and has a large share of foreign financial investment. National companies are responsible for 60% of the national production, and the two leading companies, Dexco (Brazilian industry) and Roca (Spanish industry), hold more than 60% of the market [
5]. Currently, Brazil is among the five largest world producers of sanitary ware, has 26 medium to large-scale manufacturing units, and produces around 22 million pieces of kitchen and bathroom ware, including toilets with attached tanks (37%), basins (25%), conventional toilets (20%), column washbasins (10%), tanks (5%), and urinals (3%) [
7].
In the sanitary ware manufacturing process, in general, the non-plastic raw materials undergo a dry grinding process until they reach the appropriate granulometry, to then be blended with the plastic raw materials, water, and deflocculant in tanks with mechanical agitation [
8]. The pulp thus obtained, known as barbotine, is sieved and pumped to the casting sector, where sanitary parts are cast into plaster or resin molds [
5]. After being molded, the pieces go through the steps of drying at low temperatures (below 40 °C), drying at high temperatures (above 50 °C), glazing, and firing before being stored and delivered to the customer [
9].
Drying is a thermodynamic process in which the moisture contained in the solid is removed by evaporation, providing an appreciable amount of thermal energy, which is considered a critical stage in the ceramic material manufacturing process [
10].
It is essential that the ceramic material be subjected to the drying process to ensure that it has the necessary strength and consistency during firing [
11]. If the moisture from the part is not removed during drying, the high temperature in the furnace will force the water out during the firing process, increasing the chances of defect formation and even the explosion of the product [
12,
13,
14].
By carrying out experiments on the drying of sanitary ware in a controlled environment, it is possible to understand the drying phenomenon and obtain important information for the optimization of the process. The ultimate goal is to obtain a superior-quality product, reduce the waste of raw materials in processing and energy consumption, promote economic gain, and, consequently, play an important role for industries that wish to remain competitive in the national and international markets [
15].
In addition to the financial concern, there is the environmental impact, as the energy required during the thermal processing steps is normally provided by burning fuels, which, during the combustion reaction, emit polluting gases into the atmosphere. It is well known that the consequences of global warming and climate change caused by greenhouse gas emissions are catastrophic for the ecosystem: melting of the polar ice caps, which causes rising sea levels and floods [
16,
17], species extinction [
18], worsening food security [
19], various health problems, and even premature deaths from asthma, cardiovascular, and pulmonary diseases [
20,
21].
Silva et al. [
10] conducted a study on the drying process of industrial ceramic blocks in an oven with forced air recirculation, exploring temperature variations from 50 °C to 100 °C. The research confirmed that drying ceramic products at elevated temperatures leads to substantial mass loss, intense heating, and significant volumetric shrinkage. These conditions result in high thermo-hydraulic stresses, leading to the formation and spread of cracks, compromising the product’s overall quality.
In a recent study by Zaccaron et al. [
22], the feasibility of fast-drying clay ceramics was explored through the utilization of various clay mixtures. Their research focused on the influence of plastic clay, sandy clay, and claystone compositions processed via extrusion and subjected to rapid drying cycles. The findings emphasized the pivotal role of plastic clay, with mixtures containing more than 33% of this component exhibiting significant moisture loss during drying. Remarkably, the introduction of claystone proved instrumental in stabilizing shrinkage, effectively minimizing defects such as cracking during the drying process.
Zhao et al. [
23] investigated the drying behaviors of calcium silicate and ceramic brick through experimental and numerical methods. The authors found that the surface velocity and the surface area of the sample increase the drying rate in the first drying stage (constant drying rate). Additionally, it was observed that the calcium silicate sample has a relatively longer initial drying phase compared to the ceramic sample.
Araújo et al. [
24] reported an investigation on the convective drying process of industrial hollow clay bricks within an oven. Utilizing computational fluid dynamics (CFDs), the study involved a meticulous analysis of various variables, including the behavior of air inside the oven and the distribution of water mass and temperature fields within the industrial hollow structure throughout the entire drying process.
Arvanitidis et al. [
25] developed a 1D diffusion model and performed a parametric analysis to investigate the influence of shrinkage on the drying kinetics of ceramic tiles. After validating the mathematical model by comparing the results with experimental data, the authors concluded that the magnitude of shrinkage directly influences alterations in porosity, thereby significantly impacting both the drying kinetics and the distribution of properties within the material.
In light of the scarcity of research that studies the drying process of sanitary ware, the purpose of this work is to evaluate the drying process at low temperatures in sanitary toilets. Three drying experiments were conducted in an oven with the same temperature (35 °C) and different values of drying air relative humidity. Several results of moisture content, surface temperature, and linear retraction are presented and discussed. The idea is to assist production engineers, specialists, and academics in making decisions about this interesting topic.
3. Results
Table 2 indicates the temperature, relative humidity, and drying air velocity for each analyzed case, as well as the initial mass (
m0), initial mass of water (
mw0), equilibrium mass (
me), and dry mass (
md) of the samples. It is observed that even after obtaining the same temperature for the three cases, the drying air relative humidity values were different.
While Cases 1 and 2 were carried out using forced convection, Case 3 was carried out using natural convection and, consequently, a low renewal of the air inside the oven. In this way, a greater part of the water that evaporates from the sample during the drying process ends up being trapped inside the oven, increasing the air’s absolute humidity and, consequently, the water vapor pressure and the drying air’s relative humidity. This fact is corroborated by the water vapor that is constantly condensed on the internal parts of the roof and the door of the oven, observed only in Case 3, as shown in
Figure 4.
Analyzing
Table 2, it is observed that there exists a certain variance in the initial sample masses (
m0), even respecting the same time interval between extracting the sanitary toilet from the mold and the start of the drying experiment. A possible reason for this fact is the dimensional variations in the molds of the samples.
Figure 5 shows the average moisture content of the sample over time for each case. It is observed that drying occurred more quickly for the experiment with lower relative humidity (Case 1). It is also observed that the drying curve for the experiment with intermediate relative humidity (Case 2) was very close to that obtained for the experiment with lower relative humidity. Finally, it was noticed that Case 3 presented a drying curve slower than the other experiments. This occurred because Case 3 was carried out with natural convection and, consequently, low air renewal inside the oven, keeping the air relative humidity at a high level during most of the drying experiment.
These results highlight the significant role of air relative humidity in the drying process of sanitary ware. It is evident that in situations in which the relative humidity of the drying air is higher, it becomes more advantageous to focus on reducing air humidity instead of exploring solutions to raise its temperature. As for intermediate values of relative humidity (ranging from 40 to 65%) of the drying air, variations in its value have a smaller influence on the drying removal of the product.
Figure 6 illustrates the drying rate curves as a function of time for each of the experiments. It is observed that the first stage of drying, or constant rate period, does not occur for any of the analyzed cases; that is, the drying rate decreases with time already at the beginning of the process, continues to decrease its value even more during the course of the experiment, and approaches zero as it nears the hygroscopic equilibrium state. It is evident the effect of the air relative humidity on the drying rate intensity.
It is notable that the drying rate curve for natural convection (Case 3) presents large oscillations throughout the experiment. This occurred due to the large fluctuations in the air temperature inside the oven, as shown later, due to the low air renewal. For the cases with forced convection, the maximum standard deviation observed for the drying air temperature was 0.2 °C, while for the experiment with natural convection, the standard deviation was 0.7 °C. These values are in accordance with the accuracy of the measurement devices.
It is noticed that for the experiments with forced convection (Cases 1 and 2), oscillations in the drying rate also occur, being more evident in the initial instants of the process. Factors that may have influenced these oscillations are the frequent removal of samples from the oven for measurements of mass, temperature, and dimensions; scale accuracy (5 g); and the time interval between two successive measurements. As mentioned earlier, from 18 h until the end of the drying experiment, the interval between the two measurements was only two hours. If the adopted interval were greater, oscillations in the drying rate would be less evident.
Additionally, it can be seen that the disparity in drying rates between the experiments with forced convection is greater in the initial moments of the drying process and that after 50 h, this difference decreases with a more accentuated decay of the drying rates in relation to time.
Figure 7 illustrates the drying rates as a function of the average moisture content for each case. Note that for cases with forced convection, there is a more pronounced decay in its value as the average moisture content nears 0.03 kg/kg. It is also observed that for average moisture contents above 0.02 kg/kg, the drying rate for Case 1 (
RH = 20%) is always higher than for Case 2 (
RH = 41%), which, in turn, is always greater than for Case 3 (
RH = 63%).
Figure 8a,b illustrates images of the sample used in Case 1 (the most severe drying case) at the beginning and end of the drying experiment, respectively. There is a change in the color of the piece due to the reduction in moisture content during the drying process. Because the experiment was carried out at low temperatures, no cracks or fissures were observed.
Figure 9 illustrates the surface temperature of the sample as a function of time for all analyzed cases. Similar behavior is observed for all the experiments carried out: an accentuated increase in temperature from its initial value to an intermediate value; it remains oscillating close to this value for a certain time interval; and near the end of the process, its value increases again and approaches the drying air temperature (35 °C).
This increase in the surface temperature of the sample after the intermediate temperature is explained by the change in the color of the sample, which modifies the material emissivity, directly influencing the temperature read by the digital infrared thermometer.
Analyzing
Figure 9, it is observed again that the first stage of drying does not occur for any of the analyzed cases since this process is characterized by occurring at constant temperature and with a value equal to the wet bulb temperature of the drying air. In contrast, in the experiments, it was observed an accentuated increase in the surface temperature of the samples during the initial moments of the process. It is noticed that the wet bulb temperature for Case 3 is 28.8 °C.
It is observed that the highest value of the intermediate temperature was observed for the experiment carried out with intermediate relative humidity (Case 2), while Cases 1 and 3 presented similar values for the intermediate temperature. It is also noticed that the upper-temperature level was reached first for the experiment carried out with lower relative humidity (Case 1), followed by the experiment with intermediate relative humidity (Case 2), and much later for Case 3.
From
Figure 10, which illustrates the temperature of the drying air over time for all the cases analyzed, it can be seen that for the experiments carried out with forced convection (Cases 1 and 2), the oscillations in the temperature of the drying air around the initially defined value (35 °C) were minimal. These fluctuations exhibited maximum variations of 0.3 °C and a standard deviation of approximately 0.2 °C. As for the experiment carried out with natural convection (Case 3), the air temperature inside the oven showed greater oscillations. A minimum temperature of 34 °C and a maximum temperature of 36.3 °C were observed during the experiment, and the standard deviation was 0.7 °C. This is a characteristic of the oven used, which works with smaller fluctuations in the temperature of the internal air when operated with forced convection (fan on).
Analyzing
Figure 11, which illustrates the relative humidity of the drying air as a function of time for all experiments, it can be seen that the greater oscillations in the values obtained for the air relative humidity in Cases 2, carried out in the TECNOMAT, are justified due to the fact that it rained in some moments of the execution of this experiment. Since a door providing access to the exterior of the building remained open during the experiments conducted in the TECNOMAT, it happened that during the time it rained, the air relative humidity in the laboratory increased, influencing the increase in the relative humidity of the drying air.
Oscillations in the drying air relative humidity were also observed for Case 1, performed at LABFILM. This occurred because, during the experiment, there was a greater circulation of people in the laboratory in some periods. As people exhale more humid air than they breathe in, and also because of the constant air renewal whenever the laboratory door is opened for people to enter or exit, it is natural that there is an increase in the relative humidity of the ambient air and, consequently, in the relative humidity of the drying air during periods when there is a greater circulation of people in the laboratory.
It is observed that for the experiment with natural convection (Case 3), the relative humidity of the drying air remained at a level between 65 and 70% during a good part of the experiment, and after 250 h of experiment, it reduced its value to something between 45 and 50%. This occurred because, at this final stage of the drying process, the moisture content of the piece already had values close to equilibrium, and consequently, the loss of moisture from the product to the air was greatly reduced. In this final stage of the process, the absolute humidity of the air inside the oven tended to approach the absolute humidity of the air outside the oven, as was observed in the experiments with forced convection.
Figure 12 illustrates the surface temperature of the sample over the average moisture content for each case. It is observed that at the beginning of the process, the surface temperature of the samples experiences a rapid increase from its initial value to an intermediate value, with minimal fluctuations in the average moisture content. This temperature level remains relatively constant until the average moisture content approaches approximately 0.03 kg/kg, when the temperature rises again, approaching the temperature of the drying air.
Figure 13 shows the thermograms obtained at the intermediate temperature level (
t = 36 h) and equilibrium condition (
t = te), respectively, for Case 1 (LABFILM;
T = 35.0 °C;
RH = 20%). When analyzing the thermogram, small temperature gradients are observed in the external region of the sample, with a maximum variation of 1.5 °C in the equilibrium condition. Furthermore, it is observed that the temperatures obtained by the thermogram are a little lower than the values previously presented for the surface temperature of the sample, which is justified by the heat loss from the sample between these two measurements.
Table 3 presents the initial moisture content (
M0), equilibrium moisture content (
Me), time required to reach the equilibrium condition (
te), equilibrium surface temperature (
Te), and time intervals required to reduce the average moisture content of the sample from 20% to 15, 10, 5, and 1% for each analyzed case. It appears that there is a notable discrepancy in the initial moisture content of the sample in Case 2 when compared with the values obtained for the samples in Cases 1 and 3, despite the same time interval between extracting the sample from the mold and starting the oven drying experiment for all cases.
The Case 1 sample exhibited the highest initial moisture content, registering a value of 0.22156 kg/kg, whereas the lowest initial moisture content was observed in Case 2, with a value of 0.20000 kg/kg, a difference of 0.02156 kg/kg in absolute terms and 10.78% in relative terms. Comparing these results with the data presented in
Table 2, it is noticed that the lowest value of initial moisture content is not necessarily correlated with the lowest value of initial mass.
Possible justifications for the lower value of initial moisture content obtained for Case 2 are variations in the air conditions within the company’s casting sector, dwell time of the part in the mold, and efficiency of mold moisture absorption, given that the collections of the samples were performed in different weeks and from different molds.
Table 3 illustrates that, at the same temperature, the higher the drying air relative humidity, the longer the time required to reach the hygroscopic equilibrium condition (
te) and the greater the equilibrium moisture content (
Me). It is observed that by reducing the drying air relative humidity from 63% to 41%, there is a reduction of 210 h in the time to reach the equilibrium condition (
te), while reducing the drying air relative humidity from 41% to 20% provides a reduction in the total drying time (
te) of only 12 h. Then, it is verified that the reduction in the air’s relative humidity will be more efficient for the reduction in the total drying time in situations in which its value is higher than 40%. In situations where the drying air’s relative humidity is already low, further reducing its value will not provide significant efficiency gains. Obviously, this information is related to the drying air temperature values.
It is also noticed that the time required to reach the equilibrium condition (te) for the experiment carried out with natural convection (Case 3) is 2.85 times the value obtained for the faster experiment with forced convection (Case 1) and 2.59 times the value obtained for the slower experiment with forced convection (Case 2). The equilibrium moisture content (Me) for Case 3 corresponds to 2.11 times the value obtained for Case 1 and 1.56 times the value obtained for Case 2.
A good way to define the time the product must remain in the pre-drying sector (drying at low temperatures) before going to the drying stage at high temperatures is by defining an ideal moisture content, such as 0.15 kg/kg. Knowing the product drying curves for different ambient air conditions and the ideal moisture content, it becomes possible to estimate the required pre-drying duration before transitioning to high-temperature drying.
Table 3 also provides information on the time intervals needed to reduce the moisture content from 0.20 to 0.15 kg/kg (Δ
tM = 0.20→0.15), from 0.20 to 0.10 kg/kg (Δ
tM = 0.20→0.10), from 0.20 to 0.05 kg/kg (Δ
tM = 0.20→0.05), and from 0.20 to 0.01 kg/kg (Δ
tM = 0.20→0.01).
The same increasing order observed for the total drying time is also evident for the time intervals needed to reduce the average moisture content from 0.20 to 0.15, 0.10, 0.05, and 0.01. It is also observed that the time intervals needed to achieve these moisture content reductions in the experiment conducted with natural convection (Case 3) correspond to 3.67, 3.65, 3.69, and 4.34 times the values obtained for the fastest experiment with forced convection (Case 1), respectively, and 2.76, 2.77, 2.81, and 3.55 times the values obtained for the slower experiment with forced convection (Case 2), respectively.
Table 4 provides the results of the final linear retraction and final mass variation for each of the analyzed cases. The final mass variation is related to the amount of water evaporated during the drying process, and the final linear retraction is related to the total shrinkage of the part. It is observed that the greater the amount of evaporated water, the lower the final linear retraction value. However, the shrinkage of the piece during the drying experiment is small, about 3% only, while the amount of water evaporated relative to the initial mass of water (
mw0) is about 97%. This is a stronger indication that the product has high rigidity and is very porous after drying.
Figure 14 shows the linear retraction of the samples over time for all cases. One notable observation is that the lower the drying air relative humidity, the greater the slope of the linear retraction over time. It is also noticed that there is a greater proximity between the linear retraction curves for Cases 1 and 2. As it happened for the average moisture content, the linear retraction curve over time for the experiment carried out with natural convection (Case 3) shows a smoother decay compared to the drying experiments with forced convection.
Figure 15 presents the linear retraction plotted against the average moisture content of the sample. It is evident that there is an approximately linear relationship between these two variables for all analyzed cases, proving the stronger rigidity of the sanitary ware after the drying process.