Cow dung is used as a binding agent. Cow dung is the dung of cattle and oxen, made up of 80% - 90% water and dry matter. Cow dung is the most productive product in terms of energy, with around 2 kW for 20 kg. Once dried, dung can be used as a fuel for heating [13] . This was obtained from a farm in Dakar, where 2 × 50 kg bags were collected and sent to the laboratory for pre-treatment. Coconut, the fruit of the coconut palm, is widely available and can be used for a variety of purposes, offering an alternative to charcoal [14] . Charcoal was obtained from coconut vendors, and a 50 kg bag was purchased. Groundnut shells are used as fuel in oil mills or as fertiliser, and represent a potential of 175,000 t/year, according to the programme for the promotion of renewable energy, rural electrification and the sustainable supply of domestic fuel [15] . The hulls came from SONACOS, a groundnut oil producer based in Senegal, and we were given a 50 kg bag of groundnut hulls. The sawdust is used as fuel, litter or adsorbent. A 50kg bag of sawdust is bought from a local sawmill in Dakar. Potato peelings are edible tubers with several uses. Potato peelings are one of the best organic fertilisers, and were collected at the market from women vendors. 30 kg of potato peelings were purchased. The different raw materials used are shown in Figure 1 .

The extruder used to compress the briquettes consists of a 0.075 kW three-phase motor weighing 16.29 kg. Inside the cylinder is an internal screw propeller which rotates at a speed of 280 rpm, compressing the material. After this compression stage, the fuel briquettes emerge in a cylindrical shape thanks to a cone placed at the outlet of the extruder. The production capacity of this machine is around 200 kg/h, and it is illustrated in Figure 2 with its various component parts.

It consists of defining the ranges for the study of raw materials for 100 g of coal, then varying the factors in steps of 10 g in order to obtain the range of variation of the factors. These are: the ratio of coconut shell mass to potato peel mass (X₁), the ratio of groundnut shell mass to sawdust mass (X₂) and the percentage of cow dung (X₃). The experimental matrix consists of 17 trials, with calorific value (HCV) and ash content (ASH) as the variables of interest. Based on the ratios in the experimental matrix, the masses of cow dung, coconut and peanut shells, sawdust and peelings were calculated for each experiment. The quantity of water used for the different tests varied to ensure good adhesion between the materials. The different experiments were carried out using the production diagram shown in Figure 3 .

It is used to obtain a mathematical model linking the responses (PCS and ASH and the factors X₁, X₂ and X₃) and to validate the mathematical model on the basis of the statistical results provided by the Design-Expert version 8.0.6 software. The interpretation of the results obtained is based on the analysis of variance (ANOVA), which assesses the significance of one or more factors by comparing their probabilities with the threshold value (α = 5%), and the search for the optimum operating conditions to satisfy two objectives, maximising the PCS and minimising ASH. The desirability function is used to obtain an optimal pair of factors leading to the best performance.

The analyses carried out included ash content, moisture, fixed carbon, volatile organic matter and gross calorific value. They were carried out by DANGOTE Cement’s analysis laboratory using standard NF B55-101.

These were applied to extruded optimum briquettes to achieve friability and compressive strength. These were carried out using ASTM D440. The briquettes were air-dried until they reached a mass of 14% and their initial masses were recorded. They were dropped from a regular height of 2 m onto a concrete floor. The number of broken pieces (weighing >5% of their initial mass) was recorded. The number of drops was set at 10 for each sample. The friability and compressive strength are calculated using Equation (1) and Equation (2) respectively:

$\text{F}=\frac{{\text{M}}_{\text{i}}-{\text{M}}_{\text{f}}}{{\text{M}}_{\text{i}}}\times 100$ (1)

$\text{R}=\frac{{\text{F}}_{\text{app}}}{\text{S}}$ (2)

with:

${\text{M}}_{\text{i}}$ the initial mass of the briquette before dropping in g.

${\text{M}}_{\text{f}}$ the mass of the briquette after dropping in g.

${\text{F}}_{\text{app}}$ the force applied in N.

$\text{S}$ cross-section in mm².

$\text{R}$ compressive strength in MPa.

This test consists of determining the burning time, flammability and boiling point of optimal briquettes. These tests will also be carried out on charcoal sold in Senegal. This will enable a real comparison to be made and confirm the quality of the briquettes produced.

It represents the range of variation of the factors, which is between a low and a high value. Table 1 shows the levels of these three factors chosen for the formulation of fuel briquettes.

The results of the physico-chemical analyses obtained are presented in Table 2 .

Peanut shells have a moisture content, ash content and calorific value of 9.72%, 11.47% and 4699.43 kcal/kg respectively. These values show that peanuts have a very high energy capacity. This is why they are widely used in industry as a fuel for boilers [16] . Coconut shells have a moisture content of 4.94%, a low ash content of 1.2% and a High Calorific Value (HCV) of 4443 kcal/kg. Sawdust, with a High Calorific Value (HCV) of 4399.57 kcal/kg, a low ash content of 6.31% and a slightly high moisture content of 11.23%, also has real energy potential, which is why it has long been used directly in rural areas as a fuel. However, the problems associated with these residues are transport, storage and handling [17] . Apple peelings, with their low calorific value of 713 kcal/kg and high ash content of 82.4%, are added to the formulation process in small quantities in order to contribute to the density of the briquette. In addition, large quantities of peelings in the briquettes will slow down combustion because of their high ash content, which represents the mineral matter [18] . The cow dung used acts as a binder, enabling the various components to adhere together. Because of its calorific value of 2930 kcal/kg, it is also used as a source of energy in some localities. Analysis of the volatile organic matter content of peanut shells, coconut shells, sawdust, potato peelings and cow dung gave values of 67.88%, 76.13%, 75.76%, 17.46% and 69.65% respectively, which would explain the fumes given off by the briquettes during the combustion process.

The operating conditions of the planned tests and the results of the analyses obtained are shown in Table 3 . These results will be used to carry out a statistical analysis in order to determine the mathematical models that will link the experimental factors to the responses. On the other hand, the models obtained will be used to optimise the formula to find the optimum one.

The two responses (HCV and ASH) are linked to the various study factors by quadratic models represented by the following equations:

$\begin{array}{c}\text{HCV=}+\text{47452}\text{.675}-\text{680}\text{.890}\times {\text{X}}_{\text{1}}-\text{2293}\text{.9036}\times {\text{X}}_{\text{2}}-\text{153594}\text{.3781}\times {\text{X}}_{\text{3}}+\text{78}\text{.6069}\times {\text{X}}_{\text{1}}\times {\text{X}}_{\text{2}}\\ \text{+1046}\text{.6666}\times {\text{X}}_{\text{1}}\times {\text{X}}_{\text{3}}-\text{3208}\text{.955}\times {\text{X}}_{\text{2}}\times {\text{X}}_{\text{3}}-\text{1}\text{.16049}\times {\text{X}}_{\text{1}}{}^2\text{+ 1480}{\text{.285*X}}_{\text{2}}{}^2+\text{138850}\times {\text{X}}_{\text{3}}{}^{\text{2}}\end{array}$ (3)

$\begin{array}{c}\text{ASH=}-\text{392}\text{.9302}+\text{16}\text{.1795}\times {\text{X}}_{\text{1}}-\text{15}\text{.0782}\times {\text{X}}_{\text{2}}+\text{1536}\text{.9259}\times {\text{X}}_{\text{3}}-\text{1}\text{.6484}\times {\text{X}}_{\text{1}}\times {\text{X}}_{\text{2}}\\ -\text{25}\text{.9333}\times {\text{X}}_{\text{1}}\times {\text{X}}_{\text{3}}-\text{33}\text{.1343}\times {\text{X}}_{\text{2}}\times {\text{X}}_{\text{3}}+\text{0}\text{.0977}\times {\text{X}}_{\text{1}}{}^2+\text{12}\text{.3635}\times {\text{X}}_{\text{2}}{}^2-\text{1261}\times {\text{X}}_{\text{3}}{}^2\end{array}$ (4)

This analysis enables us to study the influence of the factors on the variables of interest. This influence is all the more significant when the p.value of the effect of the factors is less than the threshold value of 5%. Terms with a p.value greater than 5% are not significant. The analysis of variance for the HCV and ASH is given in Table 4 and Table 5 respectively.

The results in Table 4 show that X₃ is a significant term since the p-value is less than 0.05. This can be explained by the large quantity of cow dung used in production, which has a major influence on the HCV. This can be explained by the large quantity of cow dung used in production, which has a major influence on the HCV. Increasing the cow dung content to certain values will result in a decrease in the HCV, as shown in Figure 4(a) . This decrease can be explained by the fact that cow dung comes with sand during the sampling process, which constitutes mineral matter and therefore has a negative impact on the HCV. Those in Table 5 show that X₂ (ratio of groundnut shell mass to sawdust mass) and X₃ (percentage of cow dung) are also significant terms. This is due to the high ash content of potato peelings and cow dung, which increases the overall ash content of the briquette. This can be seen in Figure 4(b) and Figure 4(c) .

The desirability function was used to maximise the calorific value and minimise the ash content of the briquettes in order to obtain the best formula. The possible solutions provided by this function are given in Table 6 . The experimental conditions predicted from this function were reformulated in order to validate the mathematical model. The latter effectively showed that formula No. 3 was better with a HCV of 3031 kcal/kg and a composition of 50% binder (cow dung), 10.17% coconut shell, 19.83% potato peelings, 12.42% peanut shell and 7.58% sawdust.

The physico-chemical results obtained for optimal green charcoal and charcoal are shown in Figure 5 . The calorific value of the green charcoal obtained is lower than that of the charcoal. This can be explained firstly by the significant presence of one of the materials making up the green charcoal (19.83% potato peelings), which has a very low calorific value, and also by the absence of carbonisation during production. The moisture content of green charcoal is within the stated limit. The moisture content is highly dependent on the type of biomass selected, ranging from less than 15% for cereal straw to more than 90% for algal biomass [19] . It is comparable to that obtained by Huang et al., with values of between 3% and 7% [20] . The high ash content of green charcoal is due to the binder, which makes up half of the dry matter. As indicated by Supatata et al. [21] , an ash percentage of around 4% for agricultural residues is the most appropriate for biomass briquette to avoid an increase in particulate emissions during the combustion process. The high ash content is thought to be due to the high content of inorganic compounds that can have an intrinsic catalytic effect on overall thermal conversion technologies. However, a high ash content poses huge problems as its accumulation obstructs the flow of flue gases inside biomass boilers, causing corrosion and abrasion [22] . The rather lower volatile organic matter content of charcoal compared with green coal is justified by the carbonisation it has undergone, which eliminates a large proportion of the gasifiable matter. Green charcoal that has not undergone a carbonisation stage therefore has a high volatile matter content, which corroborates the studies by Ajimotokan et al. [23] , who showed that the volatile matter content of biomass was between 41% and 78%. As stated by Cuvilas et al. (2014) that species with less volatile matter burn more slowly than those with more volatile matter [24] . In addition, a low volatile matter content promotes clean combustion of charcoal, which is therefore important according to Luxán Jimenez (2003) to be used in thermoelectric power plants. The amount of fixed carbon influences the calorific value of the briquette, equal to 30.23% for the optimum briquette, which is low compared with charcoal, which contains 50% carbon. The briquettes studied by Huang et al. have low fixed carbon values, less than 50%. Even lower values (36.6% - 43.3%) were obtained by Dias Júnior et al. [25] . Consequently, the higher the fixed carbon content, the greater the heat produced during combustion [26] . The European market requires more than 75% fixed carbon for industrial applications, although values < 75% are accepted for other industrial applications and many non-industrial applications, for example, cooking and smoking [27] .

The briquettes coming out of the extruder and then dried in the oven are very strong, cylindrical in shape, around 5 - 6 cm high and 4 cm in diameter ( Figure 6 ). Similar drop tests were carried out on charcoal and green coal in order to obtain their average friability ( Table 7 and Table 8 ). The fall on the height side and on the section side was carried out in two series.

The friability of green charcoal is very low and much lower than that of charcoal. Using an extruder coupled with a cutting machine or harmonising the cuts after extrusion should reduce friability [28] . This is a good indicator that the briquette is clean and resistant. The higher the fixed carbon content, the better the briquette will resist handling forces. On the other hand, a low fixed carbon content increases the friability of charcoal [29] .

This test was carried out using 150 g of green charcoal for 1/2 L of water. A thermometer was used to measure the temperature of the water until it boiled ( Figure 7 ). The test was carried out in a well-ventilated area to ensure good combustion, and the results are shown in Table 9 .

The green charcoal ignites after the indicated time, but its flame is smoky, which may be due to the high quantity of volatile matter. It should be noted that the fire was started with pieces of paper. The smell is due to the large quantity of binder, cow dung. The briquettes give off a little smoke when they burn. It varies, decreasing with the time it burns. The smoke is white and odourless. Its origin may lie in the species of organic matter or may be due to the large quantity of volatile matter present in the briquette. Smoke can be reduced by designing a suitable, well-ventilated firebox and sieving and adjusting the proportions of the briquette components. The high ash content of green coal could be justified by the very high ash content of potato peelings (82.40%). This raw material, which accounts for 19.83% of the mass of the briquette, has a huge influence on its characteristics. The briquettes were left to burn in order to record the ignition time. It should be noted that the heat is retained even inside the ashes. This means that there is an advantage when cooking food, in that fewer briquettes have to be used in terms of kilograms. According to Carnaje [30] , the factors that could be responsible for the rate of briquette combustion are the volatile matter content and the geometry of the briquettes. It took 30 min to boil the water. The heat released by briquettes is less intense but more even. This is why they are favoured for recipes that require long, constant cooking or for ovens. The temperature was recorded every 5 minutes for both green charcoal and charcoal in order to follow its development until boiling ( Figure 8 ).