The raw materials used in this study were gray clay from the Thicky deposits, located in the Thies region, in western Senegal at 14˚50'N, 17˚06'W, and coal fly ash, which came to us from a power station. Some physical and particle size parameters of these raw materials are presented in ( Table 1 ). The chemical characteristics of the clay and ash were determined using an X-ray fluorescence spectrometer (XRF) of the Malvern Panalytical type (Epsilon1). The chemical element results expressed in % were the average of three readings.

Clay platelets containing 0; 5; 10 and 15 wt% coal fly ash were designed. The compositions of each mixture are given in ( Table 2 ).

The different quantities of each constituent were mixed using a BMF type mixer. The quantity of water used during mixing represented 19% of the total mass of the two raw materials. Indeed, this quantity of water corresponded to the appropriate humidity for shaping by molding. This shaping was carried out using a VERDES type extruder where the material came out with a certain cross-section continuously, which was subsequently cut into sized plates (length 150 mm; width 85 mm and thickness 15 mm). In the wet state, the wafers were weighed and measured with a caliper to control the shrinkage that would occur after drying and cooking. After shaping, the wafers were placed in a Memmert type oven at a temperature equal to 40˚C and which was gradually increased every 2 hours by 20˚C. The temperature of the oven was subsequently set at 105˚C for a period of 24 hours. After drying the wafers, they were put in an oven for 3 days to cook. To avoid cracking problems and to eliminate all residual water, cooking was done gradually from 100˚C to 880˚C. 880˚C was the maximum temperature that allowed the bricks to perform well. Thus, for reasons of energy savings, it was not necessary to exceed 880˚C. The cooked platelets are presented in ( Figure 1 ). The physical and mechanical properties of the platelets were determined by standard methods. For each analytical parameter, 3 tests were carried out to get closer to the normal value. The apparent density and the loss on ignition were determined respectively according to the standards NF EN 1936 [20] and NF EN 15169 [21] . The plasticity index before cooking was determined according to the ASTM D4318 standard [22] . The Atterberg plasticity index (PI) is given by equation 1. The compressive strength properties were carried out according to the ASTM-C674 standard [23] . Water absorption was determined according to the ABNT NBR 15270-2 standard [24] . Shrinkage after drying (Rs) was calculated by applying Equation 2.

$I_P=W_L-W_P$ (1)

with:

$W_P$: the plasticity limit (LP);

$W_L$: The liquidity limit.

$Rs=\frac{Lh-Ls}{Ls}\ast 100$ (2)

with:

R_S: Shrinkage after drying;

Lh: Length of wet pads;

Ls: Length of dry platelets.

A principal component analysis (PCA), which is the most widespread of the factorial methods, was carried out in order to highlight the information carried by the platelets according to the physico-mechanical properties. Therefore, all analyzes were carried out with R software (version 4.1.1, 2021).

The results of X-ray chemical characterization of gray clay and coal fly ashwere presented in Figure 2 , and Figure 3 , respectively.

The results of ( Figure 2 and Figure 3 ) showed that in clay, the most abundant oxides were respectively SiO₂, Fe₂O_3, MgO and Al₂O₃ with contents estimated at 55.034; 10.155; 7.560 and 4.109%. While in coal ash, SiO ₂ is predominant followed by CaO, Al₂O_3, MgO and Fe₂O₃. The respective contents of these oxides in the coal fly ash were estimated at 38.574; 9.271; 7.717; 7.298 and 6.293%. The higher SiO₂ content in the clay than in the coal fly ash indicates a greater presence of quartz in the clay. On the other hand, the very high ratio (13.39) of SiO₂/Al₂O₃ in the clay shows that this material contained a greater quantity of kaolinite compared to the ash. In general, the oxides of Si and Al are mainly associated, thus forming the structures of aluminosilicates such as kaolinite [23] . However, kaolinitic clay very often contains a small quantity of fluxing oxides, which leads to the slow formation of the liquid phase of the designed bricks. Thus, the addition of alkaline oxides or alkaline earth oxides such as MgO and CaO which are fluxing oxides is necessary to facilitate the formation of the liquid phase during the baking of the wafers [25] . Thus, the relatively satisfactory CaO content (9.271%) in the coal fly ash could play the role of facilitator of the formation of this liquid phase. Note also that the two raw materials had relatively high iron oxide (Fe₂O₃) contents_. The latter are mainly responsible for the reddish color of the bricks after firing. In addition, the clay and coal fly ash used in this study do not present dangerous elements and are chemically stable. The combination of these two materials would increase the bond between the particles, thus ensuring weather resistance.

So, these two materials could be used well in making clay bricks. This ability is confirmed by the presence of kaolinite, quartz and iron oxides in these materials.

The results of the physico-mechanical characterization of the platelets are presented in Table 3 .

According to ( Table 3 ), the mechanical strengths of the different plates varied from 1.206 MPa for plate C (15% ash) to 6.358 MPA for plate A. The latter thus presents the best compressive strength. On the other hand, beyond 5% by weight of coal ash, we found that the resistance had decreased. This result could be explained by the fact that a progressive increase in the coal content breaks the bonds of the clay material but also reduces the plasticity of the material [26] . This hypothesis was confirmed by the plasticity indices, which also decreased with the increase in coal fly ash content. However, these strength values comply with Adobe legislation, which certifies the acceptance of compressive strength between 1.2 and 2.1 MPa [27] . These results on compressive strength are better than those obtained in the work of (Monica Castoldi et al. 2019) [24] and (Mekbel, Debieche et al. 2023) [28] .

Withdrawal tests were also carried out. Indeed, shrinkage is an important parameter that could cause tensions thus generating defects likely to compromise the quality of the bricks during the cooking and drying process [29] . The shrinkage results gave values between 0.04 to 0.32%, respectively for wafers A and D. Thus, wafer A (5% coal ash) lost less organic matter.

These results also showed that the loss on ignition in the different chips increased with the addition of coal ash. This trend could be explained by the volatilization of organic matter present in coal fly ash [30] . On the other hand, plates (A) and (D) had practically the same loss on ignition values. The latter were evaluated at 5.39 and 5.13%, respectively. The difference is not significant. Thus, they lost less material during cooking compared to the others.

Concerning water absorption, it is a very important parameter for the durability of terracotta bricks and depends largely on the firing temperature. Furthermore, to improve the density and decrease the water absorption of clay bricks, the firing temperature should be increased [30] . According to the sintering phenomenon of ceramic firing, when the temperature increases, the porosity between the seeds decreases, so the absorption rate also decreases. From ( Table 3 ), the water absorption rates of different platelets were approximately equal and ranged from 10.45 to 11.19%. In addition, the wafers were all fired at the same temperature and on the same wagon, which could explain the small difference between these values. However, the reference wafer (D) had the greatest absorption rate compared to the others. This characteristic constitutes an advantage because the plates, without addition, absorb a large part of this water and cause the brick to crumble, thus leading to poor adhesion. These values are also lower than the ASTM C62 specifications, which stipulate that the absorption rate should not exceed 17% [31] . Previous studies [12] , carried out on rice husk ash and sewage sludge as degreasers, have given somewhat similar compressive strengths but with higher water absorption rates compared to the incorporation of coal fly ash as a degreaser. Thus, the long-term exposure of these bricks to adverse weather elements such as water could reduce their performance. In addition, compared to coal fly ash, sewage sludge could also contain dangerous chemical contaminants, such as heavy metals, that could have adverse effects on the environment.

Thus, from all these results, it appears that plate A (5% by weight of ash) presented the best physico-mechanical properties.

To better appreciate the quality of the designed wafers, the data obtained from the physical and mechanical characterization were processed in the form of principal component analysis (PCA). The latter also made it possible to calculate the eigenvalues, the variances expressed for each factor and their accumulation ( Table 4 ). The analysis on the Dim1-Dim2 factorial plan highlighted the general trends. Indeed, dimension 1 (Dim 1) has an expressed variance of 71.1%, the largest, followed by dimension 2 (Dim 2) with 27.3% of the expressed variance. The cumulative variance expressed is 98.4% for the two dimensions. These factorial axes retained for this statistical analysis were supposed to be representative of the variance of the entire data set ( Figure 4 ). The correlation coefficients obtained between the different physical parameters of the platelets are grouped in ( Table 5 ). The latter had shown on the one hand a good correlation between shrinkage and the water absorption coefficient (R² = 0.91), a positive correlation between the resistance of the platelets and the plasticity index which is acceptable (R² = 0.78) and a fairly weak positive correlation between loss on ignition and removal of platelets. On the other hand negative correlations were observed, this is the case of the loss on fire with the resistance and with the plasticity index (R² = −0.94) and the water absorption coefficient with the resistance (R² = −0.636).

In order to classify the different variables under different groups, an Ascending Hierarchical Classification (AHC) was carried out, and this made it possible to observe three groups of variables through the dendrogram obtained ( Figure 5 ). These three groups are, on the one hand, the resistance (I), the plasticity index and the water absorption (II), and on the other hand, the shrinkage and loss on ignition (III). The k-means clustering obtained from ( Figure 6 ) allowed us to

have three classes. Class 1 consisted of sample A (5% by weight ash) with the highest compressive strength. Class 2 formed by sample D (reference) had a greater plasticity index and water absorption coefficient and finally class 3 which grouped samples B and C. This class was mainly made up of platelets with very high shrinkage and loss on ignition compared to other classes. Thus, this ascending hierarchical classification confirmed that wafer A (with 5% by weight of coal ash) presented the best physico-mechanical qualities.