Glycerol-Free Biodiesel via Catalytic Interesterification: A Pathway to a NetZero Biodiesel Industry

Abstract

Conventional biodiesel manufacturing uses alcohol as an acyl acceptor, resulting in glycerol as a side product. The increased demand for biodiesel has led to the production of a substantial surplus of glycerol, exceeding the market need. Consequently, glycerol is now being regarded as a byproduct, and in some cases, even as waste. The present study aims to suggest an economically viable and ecologically friendly approach for maintaining the viability of the biodiesel sector. This involves generating an alternative byproduct of higher value, rather than glycerol. Triacetin is produced through the interesterification of triglycerides with methyl acetate, and is a beneficial ingredient to biodiesel, reducing the need for extensive product separation. The primary objective of this research is to improve the interesterification reaction by optimising process parameters to maximise biodiesel production while using sulphuric acid as an economically viable catalyst. The study utilised the Box–Behnken design (BBD) to investigate the influence of various process variables on biodiesel yield, such as reaction time, methyl acetate to oil molar ratio, and catalyst concentration. An optimisation study using Response Surface Methodology (RSM) focused on key process reaction parameters, including the methyl acetate to oil (MA:O) molar ratio, catalyst concentration, and residence time. The best conditions produced a biodiesel blend with a 142% yield at a 12:1 MA:O molar ratio, with 0.1 wt% of catalyst loading within 1.7 h. The established technique is deemed to be undeniably effective, resulting in an efficient biodiesel production process.

1. Introduction

With the escalating global energy consumption and the rapid depletion of fossil fuel supplies, along with growing environmental concerns, there is an urgent need for a sustainable alternative energy source to solve these difficulties [1]. Biodiesel is composed of fatty acid methyl ester (FAME) chains, which are produced through the chemical processing of triglycerides found in various sources. Transesterification of triglycerides found in vegetable oils is a typical method for producing them, with alcohols such as methanol and ethanol serving as reactants in the presence of a catalyst [2,3,4]. Notably, this process results in the generation of glycerol as a byproduct in vast amounts. The resulting biodiesel has advantages such as lower carbon dioxide emissions, improved lubricating qualities, and the absence of sulphur. Furthermore, biodiesel outperforms petroleum-based diesel in terms of density, flashpoint, and viscosity. Its feedstocks are mostly renewable oils and animal fat wastes, making it a potential and sustainable energy source [5,6].

Biodiesel feedstocks, which are historically sourced from edible oils, have aroused concerns about competitiveness with the food industry, which has an impact on prices and pricing. As a result, non-edible oils are becoming more prevalent as alternative feedstocks for biodiesel syntheses. Non-edible oils include waste cooking oil, linseed oil, jatropha oil, and neem oil [7]. Although numerous techniques to produce biodiesel from different feedstocks have been established, any source containing fatty acids, such as animal or plant lipids, can be used as a substrate. Waste oils, including leftover frying oil and animal fat, are also added to the biodiesel feedstock pool [8,9,10].

The rising demand for biodiesel has resulted in major breakthroughs in both development and production. However, a recurring difficulty in the biodiesel market is the high cost of feedstocks, as well as the significant output of glycerol. Excess glycerol poses a serious challenge because further separation and treatment are expensive and time-consuming [11].

To address these issues, various technologies in biodiesel production have been investigated, including supercritical processes and interesterification. These methods have the potential to produce valuable byproducts such as triacetin, glycerol carbonate, or glycerol-tert-butyl ether, offering alternatives to the traditional crude glycerol byproduct [10,12,13,14,15]. These technologies not only improve the overall efficiency of biodiesel synthesis, but they also provide the possibility of addressing the issues connected with expensive feedstocks and glycerol disposal [12]. Recently, glycerol produced from biodiesel process was used as a cosolvent for hydrothermal liquefaction of kraft lignin for the production of bio-oil with antioxidant properties. The produced bio-oil was used as an additive to enhance the oxidation stability of biodiesel [13].

Waste Cooking Oil (WCO) is valued in a variety of methods, including transesterification, interesterification, hydrolysis, hydrodeoxygenation, hydrocracking, and hydrogenation. Transesterification stands out as the principal method for creating biodiesel and biolubricants. Interesterification provides an alternate approach that produces biodiesel and triacetin without producing glycerol, which differs from typical transesterification. Hydrolysis of WCO produces free fatty acids and glycerol as a byproduct. Hydrogenation, on the other hand, is the conversion of WCO into free fatty acids, aldehydes, alcohols, and alkanes via hydrodeoxygenation and hydrocracking [14].

Adopting green and sustainable approaches for valorising WCO increases its efficiency, paving the way for more widespread use in high-value chemical manufacture. This strategy is consistent with environmental concerns, and considers WCO to be an important resource in the development of sustainable and environmentally friendly chemical processes [15].

The interesterification process emerges as a feasible alternative to classical transesterification for biodiesel production, providing the benefit of glycerol-free biodiesel while also producing triacetin (C₉H₁₄O₆) as a desirable byproduct. The interesterification of waste cooking oil with methyl or ethyl acetate offers significant potential for the future of biodiesel production [16]. In contrast to transesterification, which takes place when esters react with an alcohol, interesterification does not involve the addition of an alcohol. One of the key reasons for contemplating interesterification is the opportunity to produce glycerol triacetate, or triacetin, as instead of crude glycerol, which requires extra processing. Given that crude glycerol output surpasses worldwide demand, making it an undesired byproduct in biodiesel production [17,18].

Several recent studies have investigated the viability of interesterification as an alternative to typical transesterification reactions, focusing on different reaction parameters [12,19]. These studies include a variety of catalytic systems, including homogeneous [20], heterogeneous [21], and enzymatic techniques [12], as well as non-catalytic systems [22]. Most of the published research has focused on homogeneous catalysts, which are known for their simplicity and cost-effectiveness in terms of materials and operation [12]. Supercritical methyl acetate has also been investigated for non-catalytic triglyceride conversion. However, using supercritical conditions requires elevated temperatures and pressures, which presents scalability issues and safety concerns [23]. A study was conducted to investigate the kinetics of interesterification under supercritical conditions of methyl acetate, utilizing alumina as a catalyst. The study revealed that the reaction is endothermic, and that both esterification and interesterification can occur simultaneously when acidic catalysts are used [24]. Alkaline hydroxides and/or alkoxides have been used as catalysts in interesterification reactions conducted at room temperature. However, the poor solubility of the alkaline compounds in the reaction medium has prompted the use of complexing agents such as tetrahydrofuran and polyethylene glycol to improve solubility and reaction kinetics [25]. A recent study demonstrated the effectiveness of a calcined Mg-Al hydrotalcite catalyst in achieving a 95.9% conversion rate of soybean oil into biodiesel, with a high level of selectivity for triacetin [21]. These various approaches and catalyst systems highlight the ongoing efforts to improve interesterification processes.

Interesterification has been intensively explored, with a focus on producing the oxygenated compound triacetin as a byproduct. Triacetin has a wide range of applications, including fuel and diesel additives, as well as economic benefits from reducing the need for separation and purification processes. Notably, triacetin can be produced through esterifying glycerol. Its miscibility in fatty acid methyl esters (FAME) enables it to operate as a gasoline additive without requiring additional product separation [26,27]. Triacetin, which is a secondary product of the production of biodiesel, can be mixed with biodiesel in small amounts (up to 10%) without affecting the standards of fuel quality. Nevertheless, the present constraints on monoglyceride levels impose restrictions on the quantity of triacetin that can be utilized, primarily because existing analytical techniques are unable to differentiate between closely related molecules [20]. Efforts were made to esterify and interesterify triglycerides using microwave-assisted processes and ultrasonic radiation, resulting in a high yield of biodiesel and selectivity of triacetin [28,29]. A new technoeconomic analysis has shown that interesterification is a cost-effective method for producing biodiesel, offering a viable alternative to conventional transesterification [30]. The authors suggested using acidic catalysts for large-scale biodiesel production and emphasized that the absence of glycerol in biodiesel has enhanced the economic performance of the process.

The purpose of this study is to investigate parametric optimization for the synthesis of biodiesel from waste cooking oil using a distinctive catalytic interesterification process. This process produces triacetin as a byproduct rather than glycerol, which adds value. Triacetin is useful in a variety of applications, including plasticizers and gasoline additives. Importantly, interesterification does not require product separation and can be used directly as a biodiesel additive to improve fuel qualities. This study attempts to optimize the parameters of the production process, demonstrating its potential as an economical and sustainable technique for producing biodiesel from waste cooking oil. The innovation of this work is summarized in process optimization of an environmentally benign methodology for glycerol-free biodiesel production.

2. Materials and Methods

2.1. Materials

Waste cooking oil was collected from a variety of residences. Merck KGaA, EMD Millipore corporation provided 99% purity methyl acetate (C₃H₆O₂), while Piochem Chemicals, Giza, Egypt, provided 96–98% concentration sulfuric acid (H₂SO₄). Deionized water has been employed to clean the equipment, while tap water was used for the condensation process. The reaction setup encompasses a 2-neck round bottle reactor vessel, a hot plate magnetic stirrer, a condenser, and a filtering system.

2.2. Experimental Setup and Procedure

2.2.1. Pretreatment

Pretreatment of WCO is a crucial step in biodiesel production to ensure the removal of water and other contaminants. First, the WCO was allowed to settle in a container for a period of 24 to 48 h. Gravity settling helps to separate water, silt, and other heavy contaminants from the oil. After settling, the top was carefully decanted to obtain a clearer layer of oil. Next, to reduce the oil’s viscosity, it was gently heated to a moderate temperature (usually between 60 and 70 °C). Heating also improves the separation of water and other contaminants. To avoid thermal degradation, do not exceed the recommended temperature range. After that, gravity separation was used to remove any remaining water droplets, and the WCO was stored in clean, dry containers. Finally, the sulfuric acid solution was prepared and combined with the waste cooking oil and the methyl acetate.

2.2.2. Operational Upstream Procedures

The combined mixture of WCO, methyl acetate, and sulfuric acid solution was placed in the reactor vessel (100 mL) and stirred thoroughly to ensure homogeneity. The reaction mixture was heated up to the appropriate reaction temperature, which was fixed at 55 °C, and a condenser was fitted with a thermocouple to monitor the temperature.

2.2.3. Operational Downstream Procedures

After the reaction period has passed, the mixture was left to cool before separating the biodiesel from the triacetin and unreacted methyl acetate.

The biodiesel was filtered to eliminate any solid contaminants, and then was washed with water to remove remaining catalyst and contaminants. Lastly, it was left to settle to separate the water layer. Figure 1 demonstrates the experimental setup.

2.3. Experimental Design

The independent variables for the catalytic interesterification of WCO with methyl acetate and sulfuric acid were selected based on a review of the literature and the reactant characteristics [31]. The experimental design included three independent variables: reaction time (coded as A), methyl acetate to oil molar ratio (coded as B), and sulfuric acid concentration (coded as C), as shown in

Table 1. Experimental design variables and their coded levels.

Independent Variable	Code	Levels
		−1	0	1
Time	A	1	2.50	4
MA:Oil (Molar ratio)	B	6	9	12
H2SO4 Concentration (vol%)	C	0.10	0.30	0.50

. Two other parameters, the stirring rate (500 rpm) and the reaction temperature (55 °C), held constant. The experimental runs were anticipated using Response Surface Methodology (RSM) and a Box–Behnken Design (BBD). The systematic experimental design enables a complete investigation of the catalytic interesterification process and the determination of optimal parameters. RSM and BBD results should be extensively evaluated and validated before being applied in practice.

The experimental runs for the catalytic interesterification process were constructed using Response Surface Methodology (RSM) and the Box–Behnken Design (BBD). A total of seventeen random runs using Design Expert Software V14 were generated. The experiment focused on the influence of three independent variables on biodiesel yield. The variables were reaction time, methyl acetate to oil molar ratio (MA:Oil), and catalyst loading. Each variable was studied at three distinct levels to fully explore the parameter space. Using the previously stated upper and lower limits from Table 1 for the independent variables, the experimental design matrix with the predicted yield is shown in

Table 2. Experimental design matrix with actual and predicted yield.

Run	Time (h)	MeAc: Oil Molar Ratio	Catalyst Loading (%)	Actual Yield (%)	Predicted Yield (%)
1	4	9	0.5	87.2	89.15
2	4	6	0.3	97	92.78
3	2.5	12	0.1	140	137.72
4	1	12	0.3	105	109.22
5	4	9	0.1	97	98.50
6	1	6	0.3	61	60.23
7	2.5	6	0.1	85	87.73
8	2.5	12	0.5	96	93.28
9	1	9	0.5	97	95.50
10	2.5	9	0.3	99	103.90
11	2.5	9	0.3	106	103.90
12	2.5	6	0.5	116	118.28
13	2.5	9	0.3	103	103.90
14	2.5	9	0.3	106.3	103.90
15	2.5	9	0.3	105.2	103.90
16	4	12	0.3	68	68.78
17	1	9	0.1	102	100.05

.

2.4. Statistical Analysis

A regression analysis was conducted using a general quadratic polynomial equation to establish the model, as represented by Equation (1).

$$Y=b_o+{{\displaystyle \sum }}_{i=1}^n{b}_i{x}_i+{{\displaystyle \sum }}_{i=1}^n{b}_{ii}{x}_i^2+{{\displaystyle \sum }}_{i=1}^{j-1}{{\displaystyle \sum }}_{j=2}^n{b}_{ij}{x}_i{x}_j+\mathsf{\varepsilon }$$

(1)

where Y is the biodiesel yield, b_o is the model coefficient constant, b_i, b_ii, b_ij, are coefficients for the intercepts of the linear, quadratic, and interactive terms, respectively, while x_i and x_j are independent variables (i ≠ j). n is number of independent variables, and ɛ is the random error.

The model’s accuracy was assessed using the coefficient of correlation (R²), adjusted coefficient of determination (R²_adj), and predicted coefficient of determination (R²_pred). The statistical significance of the investigation has been assessed by conducting an analysis of variance (ANOVA), calculating the Fisher’s F-test at a 95% confidence level. The experiments were designed, and regression analysis, graphical analysis, and numerical optimisation were performed using Design Expert 14 software developed by Stat-Ease Inc., based in Minneapolis, MN, USA.

2.5. Chromatographic Analysis

Analysis of the samples’ chemical composition was carried out using a Trace GC-TSQ mass spectrometer (Thermo Scientific, Austin, TX, USA) equipped with a TG-5MS direct capillary column (30 m in length, 0.25 mm in diameter, and 0.25 m in film thickness). We started with a temperature of 50 °C. In the column oven, it was raised at a rate of 5 °C per minute to 250 °C, and then was held it there for two minutes. Helium was used as the carrier gas at a constant flow rate of 1 mL/min, and the injector and MS transfer line temperatures were maintained at 270 and 260 °C, respectively. An Autosampler AS1300 coupled with a GC in split mode was used to automatically inject 1 l diluted samples with a solvent delay of 4 min. Full-scan EI mass spectra were taken from m/z 50–650 at ionisation voltages of 70 eV. The ion source was heated to a comfortable 200 °C. Components were named after their mass spectra, and were compared to the databases WILEY 09 and NIST 14.

3. Results

3.1. Model Development and Statistical Analysis

Design Expert software has fitted four models for each response, including linear, two factors interactions (2FI), quadratic, and cubic polynomials. From the various fitted models for each response, a single model has been chosen using different statistical tests, such as lack of fit analysis, adjusted coefficient of determination (R²_adj), predicted coefficient of determination (R²_pred), and associated aliased coefficients. The software recommended using the quadratic model to predict the biodiesel yield responses. Equation (2) represents the quadratic models that have been developed to describe the empirical relationships between responses and reaction variables at specific levels, as indicated by the coded factors in Table 1.

Y = 103.9 − 2.1 A + 6.38 B − 3.47 C − 18 AB − 1.2 AC − 18.75 BC − 17.43 A² − 3.98 B² + 9.33 C²

(2)

Y represents the yield of biodiesel. A, B, and C represent the process variables, including reaction time, MA:Oil molar ratio, and catalyst concentration.

The regression equation illustrates the influence of the reaction variables on the response variable. The positive sign of each term represents a harmonious effect, while the negative sign indicates an opposing effect. The linear coefficient represents the influence of the reaction variable on the response, while the coefficient of variables interaction measures the collective impact of the process variables. The quadratic coefficient represents the effect of an increase in the variable on the response.

Equation (2) demonstrates that the reaction time has a negative impact on biodiesel yield, as indicated by the negative sign coefficients. Increasing reaction time variables has a diminishing effect on biodiesel yield. Nevertheless, methyl acetate to oil molar ratio and catalyst concentration have positive signs, indicating that an increase in these process variables leads to an increase in glycerol yield.

3.2. Statistical Validation

The accuracy of the predicted models has been examined to identify any errors related to the assumptions of normality. Several analyses have been employed to assess the suitability of the predicted model. The coefficient of correlation (R²) assesses the accuracy of the predicted model. A value of R² approaching unity indicates a high degree of similarity between the predicted values of the model and the actual experimental value. The R², R²_adj, and R²_pred values for the predicted biodiesel yield model have been assessed as 0.9837, 0.9627, and 0.8453, respectively. The adequacy precision value is a metric that measures the predicted response value in relation to its relative error, also known as the signal-to-noise ratio. It provides an indication of the likely range within which the predicted value will fall. A value exceeding 4 is deemed desirable. The adequacy precision value for biodiesel yield has been evaluated as 29.8.

The statistical data obtained from variance analysis has been utilized to ascertain the significance of the predicted models. Furthermore, the impact of reaction parameters and their interactions was determined. The parameter values obtained from the analysis of variance (ANOVA) are presented in

Table 3. Analysis of variance (ANOVA) of the developed model.

Source	Sum of Squares	df	Mean Square	F-Value	p-Value
Model	4882.98	9	542.55	46.84	<0.0001
A-Time	49.00	1	49.00	4.23	0.0787
B-Methyl Ac:Oil	364.50	1	364.50	31.47	0.0008
C-H2SO4:Oil	96.60	1	96.60	8.34	0.0234
AB	1332.25	1	1332.25	115.00	<0.0001
AC	5.76	1	5.76	0.4972	0.5035
BC	1406.25	1	1406.25	121.39	<0.0001
A2	1260.17	1	1260.17	108.78	<0.0001
B2	62.41	1	62.41	5.39	0.0533
C2	356.38	1	356.38	30.76	0.0009
Residual	81.09	7	11.58
Lack of Fit	44.41	3	14.80	1.61	0.3197
Pure Error	36.68	4	9.17
Cor Total	4964.07	16

.

Table 3 presents an evaluation of the model’s importance by utilising both the p-value and F-test at a confidence level of 95%. A p-value less than 0.05 signifies a greater level of significance for the corresponding parameter. The model has been determined to be highly significant, as evidenced by p-values below 0.0001. The significance of the model in accurately depicting the experimental results has been ensured by these factors. Lack-of-fit analysis is an ANOVA method employed to measure the difference between the regression model and the actual data points collected during an experiment. The insignificant p-value for the lack of fit test indicates that the model fits well. The lack-of-fit value has been measured to be 0.319. The absence of statistical significance in the test suggests that the models have effectively portrayed the majority of the experimental data.

In an attempt to simplify the developed model, and due to the fact that both reaction time (A), interaction between reaction time and catalyst to oil ration (C), and the excess of MA:Oil ratio (B²) were not significant with p-value > 0.05, C and B² were excluded from the model. Although parameter A is not significant, it was kept to maintain the hierarchy of the model. The newly developed simplified model is demonstrated in Equation (3).

Y = 102.28 − 1.98 A + 6.25 B − 3.47 C − 18.25 AB − 18.75 BC − 17.5 A² + 9.33 C²

(3)

Similar to Equation (2), Y represents the yield of biodiesel. A, B, and C represent the process variables including reaction time, MA:Oil molar ratio, and catalyst concentration as shown in

Table 4. Analysis of variance (ANOVA) of the simplified developed model.

Source	Sum of Squares	df	Mean Square	F-Value	p-Value
Model	4745	7	677.86	34.07	<0.0001
A-Time	31.20	1	31.20	1.57	0.242
B-Methyl Ac:Oil	312.50	1	312.50	15.71	0.0033
C-H2SO4:Oil	96.61	1	96.60	4.86	0.045
AB	1332.25	1	1332.25	66.96	<0.0001
BC	1406.25	1	1406.25	70.68	<0.0001
A2	1293.44	1	1293.44	65.01	<0.0001
C2	341.80	1	341.80	17.18	0.0025
Residual	179.06	9	19.90
Lack of Fit	142.38	5	28.48	3.11	0.147
Pure Error	36.68	4	9.17
Cor Total	4924.07	16

. The R², R²_adj, and R_2pred values for the simplified biodiesel yield model have been assessed as 0.9847, 0.9659, and 0.8489, respectively, with an adequacy precision value for biodiesel yield evaluated as 28.7.

The normality of residuals for the simplified model was assessed by examining a normal plot, which demonstrated that they approximately follow a straight line, as depicted in Figure 2. This test verifies the assumption that the residuals of the biodiesel model follow a normal distribution, thus ensuring its validity.

Additionally, Figure 3 illustrates a visual representation of the actual values obtained from the experiment, as well as the predicted values derived from the developed model. The model’s accuracy in predicting the response variable has been ensured by the similarity between the actual and predicted values.

The investigation of error randomization has been conducted by plotting residuals against actual response values. Figure 4 demonstrates that the residuals were distributed in a random manner, without adhering to any discernible pattern. These randomised distributions confirm the assumption used in the ANOVA.

3.3. Effect of Process Parameters

The influence of various process parameters on biodiesel yield was statistically analysed using Analysis of Variance (ANOVA). Table 3 summarizes the p-values for each parameter and their interaction effects. As expected, reaction time (2.5 h) itself did not exhibit a significant impact on biodiesel yield (p-value = 0.0787). This is further confirmed by Figure 5, where the yield remains relatively constant between 1 and 4 h before a slight decrease at longer times. However, the high A² value (p-value > 0.0001) in the ANOVA table indicates a significant curvature effect for reaction time. This suggests that, while overall time may not be a major factor within the chosen range, extremely short or long reaction times could negatively impact yield. A similar observation of reaction time having a non-linear effect on biodiesel yield by previous researchers [32]. They attributed this effect to the interaction between reaction time and FAME degradation at extended times.

The methyl acetate to oil (MA:O) molar ratio and catalyst concentration were found to have significant individual effects on biodiesel yield (p-values of 0.0008 and 0.0234, respectively), as shown in Table 3. Figure 5 and Figure 6 depicts the interaction between the reaction time and MA:O ratio. While increasing the MA:O ratio generally leads to higher yields, the effect is dependent on reaction time. At shorter reaction times (1 h), a higher MA:O ratio enhances the mass transfer, promoting faster conversion of the oil. However, at longer reaction times (4 h), excess methyl acetate can react with the already formed biodiesel esters, generating undesirable side products. This highlights the importance of optimizing both reaction time and MA:O ratio for maximum yield. Recent studies by Casas et al. [32] echo these findings, emphasizing the need for a balanced MA:O ratio to achieve optimal yield and minimize side reactions.

The interaction plots similar to Figure 6 represent the impact of a certain process parameter at different levels of another parameter. If the effect of the first parameter on the response is similar at a higher and lower level of the second parameter, the lines will seem parallel. However, if there is an impact on the level of parameter 2 on the impact of parameter 1 on the response, there will be an interaction. Red lines represent the effect of response 1 at a higher level of parameter 2. However, the black line represents the effect of response 1 at a lower level of parameter 2.

A similar interaction effect was observed between the MA:O ratio and catalyst concentration (Figure 7 and Figure 8, and p-value < 0.0001 for BC in Table 3). At a low MA:O ratio (6:1), increasing the catalyst concentration has a detrimental effect on yield. Conversely, at a higher MA:O ratio (12:1), a higher catalyst concentration leads to a positive impact. This can be explained by the catalyst’s role in accelerating the reaction. When the MA:O ratio is low, excess catalyst can promote side reactions. However, at a higher MA:O ratio, the additional catalyst can effectively utilize the abundant methyl acetate for efficient biodiesel production. References supporting this interaction effect include Sendzikiene et al. [31], where the optimal catalyst loading was found to be dependent on the reactant ratio used in biodiesel production via interesterification.

These findings highlight the importance of using response surface methodology (RSM) to optimize the interesterification process for biodiesel production. By considering the interactive effects between reaction parameters, RSM allows for the identification of conditions that maximize biodiesel yield, while minimizing side-product formation. Further research could explore the influence of additional factors, such as temperature and type of catalyst, on the interesterification process using waste cooling oil and methyl acetate.

3.4. Product Analysis

The resulting fatty acid methyl ester (FAMEs) content that was analysed using GC-MS is shown in Figure 9, whilst the fatty acid profiling of the biodiesel is expressed in

Table 5. Fatty acid profiling of the biodiesel.

RT	Area %	Compound Name
5.40	0.69	4,4-Dimethyl-2-pentanol
24.41	13.84	Hexadecanoic acid, methyl ester
25.85	2.17	Hexadecanoic acid, ethyl ester
27.21	55.61	Octadecadienoic acid, methyl ester
27.60	2.46	Methyl stearate
28.63	18.87	9,13-Octadecadienoic acid methyl ester
36.74	6.37	2-hydroxy-1-(hydroxymethyl)ethyl ester

.

The results showed that the ester content of the produced biodiesel is 99.3%, which is satisfying European biodiesel standard EN14214 (>96.5 wt%) [33]. In addition, the produced biodiesel recorded 3.8 mm²/s, 0.03% water content, 142 °C flash point, and 7.6 h of oxidation stability, based on the Rancimat method. On the other hand, the produced biodiesel recorded a cetane number of 52, which is 7% lower than biodiesel without triacetin (produced at similar conditions using methanol). However, the produced biodiesel–triacetin blend is still fulfilling the European biodiesel standard (EN14214) for Cetane number (>51). Finally, the triacetin ratio in the produced mixture is 10 wt%, which is satisfactory according to the EN14214 standard [32].

3.5. Process Optimisation

Previous studies have reported the use of RSM to optimise the reaction variables that impact biodiesel production. To enhance yield of biodiesel, a numerical feature has been utilised through the implementation of Design Expert 14 software. This feature allows for the evaluation of the optimal combination of conditions to achieve the desired target. The maximum target for biodiesel yield response has been established. The software generated 20 solutions for optimal conditions, and the solution with the highest desirability was chosen. The optimal conditions yielded a biodiesel yield of 142% yield at 12:1 MA:O molar ratio and 0.1 wt% of catalyst loading within 1.7 h.

To verify the predicted optimal conditions, three experiments were conducted under these conditions, and the average result was taken as the experimental outcome. The experimental validation confirmed a biodiesel yield of 141.2%, demonstrating that the predicted optimum conditions were accurate within a relative error of 0.83% compared to the experimental results.

4. Discussion

The influence of various process parameters on biodiesel yield was statistically analysed using Analysis of Variance (ANOVA).

4.1. Reaction Time

At short reaction times, the conversion of triglycerides to biodiesel esters might be incomplete. Conversely, extended reaction times can lead to product degradation due to over-reaction with the catalyst or undesirable side-reactions between the formed biodiesel and reactant molecules. Reference [20] observed a similar phenomenon in their study on biodiesel production from waste vegetable oil. They attributed this effect to the interplay between reaction progress reaching a plateau and product degradation at extended times.

4.2. Methyl Acetate to Oil (MA:O) Molar Ratio

The MA:O molar ratio significantly affects biodiesel yield (p-value of 0.0008). Figure 6 depicts the interaction between reaction time and MA:O ratio. Generally, increasing the MA:O ratio leads to higher yields due to several factors, including enhanced mass transfer and favourable equilibrium shift. A higher concentration of methyl acetate provides more reactant molecules available for collisions with triglycerides, accelerating the interesterification reaction. The reaction reaches equilibrium when the concentration of products (biodiesel and triacetin) is comparable to the reactants (triglycerides and methyl acetate). Increasing the MA:O ratio pushes the equilibrium towards product formation.

However, the impact of the MA:O ratio is dependent on reaction time. At longer reaction times (4 h), excess methyl acetate can have a detrimental effect. This can be explained by potential side-reactions. Excess methyl acetate can react with the already formed biodiesel esters, generating undesirable byproducts like fatty acid methyl esters (FAMEs) with shorter chain lengths.

Therefore, optimising the MA:O ratio is crucial for maximising biodiesel yield while minimizing side reactions.

4.3. Catalyst Concentration

Catalyst to oil molar ratio also significantly impacts biodiesel yield (p-value = 0.0234). The catalyst plays a critical role in accelerating the interesterification reaction. However, the optimal catalyst concentration depends on the MA:O ratio, as shown in Figure 7 and Figure 9, and the p-value for interaction (BC) in Table 3 (p-value < 0.0001).

At a low MA:O ratio (6:1), an excessive amount of catalyst can be detrimental. This is because the limited availability of methyl acetate molecules can lead to catalyst overconsumption through unwanted side-reactions with the triglycerides. However, at a high MA:O Ratio (12:1), the abundant methyl acetate molecules can effectively utilize the additional catalyst, promoting a faster and more efficient interesterification reaction.

The findings emphasise the importance of RSM to optimise the interesterification process for biodiesel production. By accounting for the interactive effects between reaction time, MA:O ratio, and catalyst concentration, RSM allows for the identification of conditions that maximize biodiesel yield, while minimizing side product formation. Further research could explore the influence of additional factors, such as temperature and type of catalyst, on the interesterification process using waste cooling oil and methyl acetate.

5. Conclusions

This study investigated the viability of interesterification as an environmentally friendly substitute for traditional transesterification in the production of biodiesel from waste cooking oil. Interesterification, unlike transesterification, yields triacetin, a valuable biofuel additive, in addition to biodiesel, without generating glycerol as a byproduct. The interesterification process was optimised using RSM to identify the crucial factors that affect biodiesel yield.

The findings indicated that the relationship between reaction time and yield was non-linear. Optimal reaction times facilitated conversion, but excessively prolonged durations could result in product degradation. The yield was significantly influenced by the molar ratio of methyl acetate to oil (MA:O). By increasing the ratio of the MA:O molar ratio, the yield is generally improved. This is because it helps with the movement of substances and pushes the reaction towards the formation of the desired product. Nevertheless, when the reaction time is extended, an excessive amount of methyl acetate may engage in unfavourable side reactions, thereby diminishing the overall yield.

An important discovery in this study was the notable interaction between reaction time and MA:O ratio. A higher molar excess of the main reactant over the other reactant improved the yield by increasing the number of available reactant molecules at shorter reaction times. On the other hand, when exposed for long periods, having a higher molar ratio of MA:O methyl acetate impacted the yield negatively. Likewise, the correlation between the MA:O ratio and catalyst concentration were crucial. Increasing the concentration of the catalyst had a positive effect on the yield when the ratio of MA to O was high. However, at a lower ratio, it could have a negative impact due to excessive consumption of the catalyst inside reactions.

By taking into account these interdependent factors, the optimal operating conditions were determined, resulting in a biodiesel production rate of 142% (which surpasses 100% due to the presence of triacetin) at a molar ratio of 12:1 of methyl acetate to oil, and a 0.1 wt% of catalyst loading ratio within 1.7 h. The results indicate that interesterification, optimised using RSM, has the ability to produce biodiesel from WCO in an efficient and sustainable manner. Finally, the produced biodiesel is fulfilling European biodiesel standard EN14214 for ester content, viscosity, water content, flash point, and Cetane number.