The most used raw materials for bioethanol production can be classified into three main types: sugars, starches, and cellulose materials [4] . Because banana residues are made up of lignocellulosic fibers, they could be used as raw material for obtaining cellulose or in obtaining bioethanol, in this way, an added value would be provided to say residues.

Initially the raw material was washed with tap water, then dried at 60˚C for 48 h, and finally mechanical milled using a conventional mill (Corona brand) to obtain a particle size between 1 - 5 mm. Subsequently, the determination of the percentage of cellulose, hemicellulose and lignin was carried out, in establishing the potential for the raw material in bioethanol production [20] .

The experimental variables evaluated were fermentation time, hydrolysis pH, hydrolyzing agent, concentration and hydrolysis time [3] [21] . This work was carried out in duplicate by a 2⁴ experimental factorial design, where two levels and four factors were obtained, thus having an arrangement of 16 combinations that are observed in Table 1.

Initially, a pretreatment to the raw material was carried out, as described in Figure 1(a). In the fermentation process, the reducing sugars were converted to alcohol using yeast Saccharomyces cervisiae as the fermenting agent, which was activated prior to use (see Figure 1(b)). Finally, the raw material subjected to pretreatment with yeast activated was mixed and kept in an amber bottle in the incubator for 7 or 15 days at a temperature of 25˚C [22] . The fermented product was manually filtered [23] , and fractional distillation was performed according to ASTM method D2892-16.

The different bioethanol samples obtained were characterized by the refractive index using a PZO Warszawa Poland refractometer with temperature control of +/− 0.1, specific gravity (20˚C) and infrared spectrum through a Shimadzu FTIR spectrophotometer, model Prestige-21, following the procedure established in the equipment manual for volatile liquid samples.

In order to evaluate the simple and combined influence of critical variables in the production of bioethanol, a statistical ANOVA factorial analysis was performed using the Statgraphics Plus program, through graphical representation with a Pareto diagram.

The chemical characterization of banana agro-industrial waste (Table 2) indicated that, the raw material can be considered as a potential source for the production of bioethanol as it has a high content of cellulose. When comparing its composition with lignocellulosic materials, studies in the literature find a low percentage of hemicellulose and lignin, which increases the production of reducing sugars, and that the optimal hydrolytic pretreatment could include hydrolyzing agents in relatively low concentrations [3] [6] [24] .

Figure 2 shows that the density values closest to anhydrous ethanol (0.789 mg/L to 20˚C) come from samples obtained through basic hydrolytic pretreatment with 30 minutes hydrolysis time. The lignin degradation of the lignocellulosic

biomass allows the conversion of cellulose and hemicellulose into simpler sugars more easily digested by the fermentation agent, further decreasing the crystallinity of the cellulose and increasing the surface area [5] . In contrast, the samples obtained with 15 minutes hydrolysis time or acidic hydrolyzing agents represent densities close to the one from water, indicating a composition which water percentage is higher, and therefore, the existence of hydrogen bridges intra and intermolecular [26] . That fact is corroborated when establishing the concentration of bioethanol in each of the samples through the interpolation of the experimental data in theoretical curves of ethanol-water mixtures [27] . Figure 3 shows the increase in the percentage of ethanol as the density of the distillate decreases and approaches the one from pure alcohol.

The refractive index can indicate the purity of a substance or quantify the amount of a component in a binary mixture [28] . Figure 4 shows that the pretreatment with NaOH showed that the bioethanol had less water, which is reflected in refractive indices closer to the value registered in the literature for ethanol [29] .

The experimentally obtained bioalcohol can contain traces of compounds specific to the fermentation process, as well as water, given the characteristic formation of the ethanol-water azeotrope [30] [31] . This bioalcohol cannot be destroyed through separation methods such as fractional distillation.

Figure 5 shows the differences with respect to the anhydrous compound showing existence of absorption bands, characteristic of the organic alcohol functional group. For the particular case, they correspond to ethanol, and are located between 3050 - 3600 cm⁻¹ for O-H bond, 2950 - 3000 cm⁻¹ C-H bond and 1000 - 1100 cm⁻¹ C-O bond [32] [33] .

The ANOVA statistical analysis showed significant differences between treatments when combinations of AB, AD, BC and CD variables were performed (Table 3) because the P-Value was higher than the λ set at 0.05. However, it should be mentioned that the individual influence of variables (A, B, C and D)

does not prove to be significant between treatments, which would indicate that these alone would not have a significant influence on the production of bioethanol.

Figure 6 represents the Pareto diagram obtained through the statistical program Statgraphics. The most influential variable in the production of bioethanol is the pH of hydrolysis, since the best results in general were obtained with a basic pre-treatment (NaOH) because as it was mentioned before, it produces a rupture of the structure of the lignin increasing the internal surface, thus reducing the degree of polymerization and crystallinity [10] . The second most consequential variable was the concentration, as the experimental conditions using a 0.1M concentration represented a higher yield. At a higher concentration, the cellulose might convert to other molecules that may not be fermentable [22] [34] . The combination of fermentation time and concentration of the hydrolyzing agent showed a greater influence than the independent variable of fermentation time, although this last one proved to have a positive effect, directly proportional to the yield of ethanol. This result could be attributed to the essential role of hydrolysis in the release of cellulose and formation of bioethanol [35] .

Figure 7 shows the yield of bioethanol obtained under different hydrolytic pretreatment conditions. It is seen as a general tendency that NaOH as a hydrolyzing agent represented better results; tests with hydrolysis time of 30 minutes at basic pH (NaOH), with high fermentation time and low concentrations tend to have higher yields, the most representative being the test carried out with 15 days of fermentation in a 0.1 M concentration. The aforementioned behavior could be attributed to the fact that hydrolysis at high concentrations may form inhibitory agents that affect the fermentation process, which for the particular case would be present in greater proportion when the pH of hydrolysis is acidic and for short periods [25] . Although, in the present study the bioethanol obtained is not higher than 35% in yield, which is low compared with recent researches on banana residues that report ranges from 40% to 80% of bioalcohol

[22] [23] [24] . It is important to take into account that the analysis conditions play a fundamental role since the aforementioned investigations evaluate enzymatic hydrolysis, involving strongly acid pretreatments and high temperatures, while this study was carried out with acid-base hydrolysis, using temperatures and moderate times, which facilitates the use of waste byproducts as part of the solution to the energy demand in remote rural areas and in the development process, giving the population the possibility of using a versatile and much less polluting biofuel [36] [37] .