The Effects of Biodiesel on the Performance and Gas Emissions of Farm Tractors’ Engines: A Systematic Review, Meta-Analysis, and Meta-Regression

Abstract

The need for the decarbonization of heavy-duty vehicles requires a deep understanding about the effects of biofuels, which represent a viable pathway to cut the emissions in the hard-to-abate sectors, like agricultural tractors. A novel meta-analysis approach can help to thoroughly investigate the effects of biodiesel blends on farm tractor engines in terms of performance and emissions. Studies were identified using the main keywords related to internal combustion engines in prominent scientific databases. Standardized mean differences were calculated for each study to evaluate engine performance and gas emissions. Mixed-effects regression models were developed to investigate performance and environmental pollution changes over different biodiesel blending ratios, biodiesel sources, and engine types. The analysis revealed significant effects of biodiesel blending ratio on decreasing torque [−13.0%, CI 95% (6.7%–19.3%); I² = 97.67; p = 0.000; Q = 129.94], engine power [−15.0%, CI 95% (10.0–20.0%); I² = 54.82; p = 0.000; Q = 101.81], CO₂ emissions [−24.1%(15.0–32.0%); I² = 0.198; p = 0.000; Q = 20.04], and CO emissions [−17.5%, CI 95% (16.0–18.0%); I² = 98.62; p = 0.000; Q = 97.69], while increasing specific fuel consumption [+5.2%, CI 95% (1.0–9.0%); I² = 95.94; p = 0.000; Q = 129.74] and NO emissions [+11.0%, CI 95% (6.0–15.0%); I² = 98.51; p = 0.000; Q = 157.56]. The same analysis did not show any influence of the sources of biodiesel and the engine type. Finally, meta-regression found a significant positive association between increasing ratios of biodiesels and decreasing torque, engine power, CO and CO₂ emissions, and increasing fuel consumption and NO emissions in terms of linear equations. Although through these equations it is not possible to individuate an optimal range of blending ratios able to lower the emissions and not affect the engine parameters, the range from 9.1% to 13.0% of biodiesel is a good tradeoff. Within it, the only decrease in engine performance is in charge of the power, however contained within 4%, while CO and CO₂ emissions are reduced (respectively by 0.0%/−2.8% and −3.6%/−6.0%) without using any specific pollutant abatement systems.

1. Introduction

Diesel engines are utilized in powertrains across the industrial, transportation, and agricultural sectors due to their distinctive thermodynamic, economic, and structural characteristics [1]. Nonetheless, despite recent advancements in engine technologies aimed at mitigating the polluting effects of these energy systems on air quality and human health, the optimization of this engine’s operation remains a persistent challenge. Fossil fuels are primary energy sources that cause gas emissions and global warming. These fuels cause pollution in the environment and cause emissions of carbon dioxide that were previously stored underground [2]. One simple and affordable way to reduce hazardous emissions, even for existing diesel engines, is through fuel reformulation. Another possibility, more radical as far as the architectural implications on a vehicle and on the refueling infrastructure are concerned, is the shift towards electrification. While a large share of passenger vehicles and light-duty transportation systems are moving towards electric powertrains [3], other industrial, agricultural, and heavy-duty sectors, also known as “hard-to-abate sectors”, face difficulties in adopting this shift, as the typical workday energy demand and the average engine load are too intense to be profitably met using the current technology of storage batteries and charging times [4]. Therefore, heavy-duty engines, and specifically in the agricultural sector [5], can still benefit from the use of traditional powertrains if powered with alternative and “clean” fuels, having a neutral CO₂ balance. Using blends of diesel fuel, biodiesel, and bioethanol can significantly reduce exhaust emissions while maintaining compatibility with existing fuel-supply systems designed for low-viscosity fossil fuels. This approach leverages the cleaner combustion characteristics of biofuels, leading to improved air quality and reduced greenhouse gas emissions without necessitating major modifications to current fuel infrastructure [6]. As a clean alternative fuel, biodiesel is produced when oil sources react with methanol and/or ethanol alcohol in the presence of a catalyst, for example: CaO, CaTiO₃, CaZrO₃, CaO-CeO₂, CaMnO₃, Ca₂Fe₂O₅, and KOH/Al₂O₃ [7]. Biodiesel holds a unique position due to its physicochemical properties, which are similar to those of mineral diesel, such as cetane number [8]. Its high viscosity, limited volatility, and inadequate cold flow properties at lower temperatures significantly affect both its combustion characteristics and emissions outputs. High viscosity and low volatility hinder fuel atomization, whereas poor cold flow characteristics lead to the solidification of fatty acid compounds, forming crystals at lower temperatures and compromising the engine’s operation [9,10,11,12]. Based on a review of the literature, many studies suggest blending different fuels, such as gasoline, biogas, different types of alcohols, and fuel additives with biodiesel to mitigate the drawbacks of the high viscosity of biodiesel without raising too much the total cost per liter and decreasing the engine performance [13]. As a renewable alternative fuel boasting a high oxygen content, ethanol blended with biodiesel can reduce fuel density, viscosity, cold filter plugging point, and freezing susceptibility of the resulting blend [14,15]. A study has advised against using biodiesel or pure vegetable fuels in diesel engines without adding alcohols to lessen their high density and viscosity [16].

Off-road internal combustion engines, such as the ones used in the agricultural sector [17], contribute to environmental pollution through the emission of nitrogen oxides, carbon monoxide, and other pollutants. Although the engine’s global efficiency and technology have advanced significantly in the agriculture sector, there is still opportunity for further developments [18]. In addition to technological advancements, incorporating biodiesel blends can aid in adhering to exhaust gas emissions limits; nevertheless, it is crucial to accurately identify blends to mitigate potential drawbacks [19]. Biodiesel blends derived from different biomasses and formulated with varying ratios possess unique chemical and fluid dynamic properties that affect their energetic content, efficiency, and combustion in comparison to diesel fuel and other fuel blends [20]. Injection systems play a crucial role in maximizing the performance and minimizing the emissions of diesel engines. Conventional injection systems, directly or indirectly, contribute to environmental pollution [21]. Injection systems are affected by the use of biodiesel or biofuel mixtures [22]. The injector’s response varies due to the different fluid dynamic properties and bulk modulus of the fuels, resulting in changes to the injection timing (delay or advancement) and a coarser spray atomization leading to the production of larger fuel droplets [23]. These factors, in turn, influence combustion and emissions. Cutting-edge injection and emissions control systems, such as advanced common rail systems and selective catalytic reduction (SCR) technology, have the potential, respectively, to boost the efficiency of diesel engines by up to 15% and notably decrease their environmental footprint by reducing emissions by up to 50% [24,25]. This advancement renders them capable of meeting current rigorous emissions regulations and fostering cleaner air quality. However, they also present practical challenges such as increased costs, maintenance demands, technical complexity, etc. [26,27].

Unfortunately, studies using different ratios of biodiesel and diesel fuel in tractor engines reported conflicting results, thus making a complete understanding of the action of biodiesel-based blends even more complex. Tomić et al. mixed ratios ranging from 15% to 100% of biodiesel derived from sunflower and studied their impact on engine performance and emissions. They observed that the inclusion of biodiesel (100%) led to a decrease in engine power (−9.55%), as well as a reduction in CO₂ (−9.10%) and CO emissions (−8.75%), while also increasing specific fuel consumption (+8.80%) [28]. Emaish et al. assessed the effects of blends containing varying percentages (0%, 5%, 20%, and 100%) of biodiesel and noted adverse impacts on engine performance at higher biodiesel ratios (max torque: −9.0%) [29]. However, Allami et al. investigated the effects of biodiesel blends on tractor engines and reported that biodiesel could enhance maximum power (+9.10%) [30]. Bavafa et al. assessed the impact of a biodiesel blend derived from poultry fat on tractor engine performance. Their findings indicated reduced specific fuel consumption at higher biodiesel ratios (−8.70%) [31]. Conflicting findings regarding the impact of biodiesel on engine performance and gas emissions highlight the need for further investigation. Discrepancies may arise from factors such as biodiesel source, blending ratio (to be intended as the share of biofuel in a fuel blend including also a conventional fuel), engine type (naturally aspired or turbocharged), study duration, and the conditions under which the engine is tested. The conflicting results in the literature and the complexity of the many phenomena involved in the problem (comprising the injection systems and the emissions abatement solutions) make it difficult to identify a clear indication. Moreover, the tractors assessed in these studies show a diverse range of emissions abatement solutions. While some tractors lack such systems entirely, others feature basic catalysts, advanced SCR systems, and particulate matter filters, all of which can significantly impact the ultimate findings [32,33]. Meta-regression considers moderator factors and can retrieve the data. To provide comprehensive insights, a meta-analysis and meta-regression are essential. There are some review meta-analyses on the effects of energy consumption and environmental degradation nexus [34], the effects of environmental CO₂-based chemical production [35], and the energy use, carbon emissions, and environmental burdens of industry [36].

Indeed, a meta-analysis statistically combines the results emerging from multiple scientific studies addressing a similar research question, evidencing convergences/similarities and divergences/differences in the obtained findings. Meta-regression is the best tool that could be used to shed light on the large scientific production of these last years, as it would evaluate both positive and negative effects of biodiesel on tractor engine performance and emissions, considering variables like biodiesel source, blending ratio, and engine type/specifications. Despite our review of existing literature, we have not encountered a comprehensive study addressing all these aspects at the same time. Therefore, this preliminary systematic review and meta-analysis aims at filling this gap by examining the effects of biodiesel on tractor engine performance and emissions. This study investigates the impact of biodiesel on engine parameters, including torque, specific fuel consumption, and power, as well as emissions of CO, CO₂, and NO in tractor engines, based on previous studies. It also considers the roles of blending ratio, engine type, and biodiesel sources as moderating factors.

2. Materials and Methods

2.1. Study Protocol Registration

The study adhered to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines during its execution [37,38]. The PRISMA checklist consists of 27 questions, covering every aspect of the systematic review process, including the title, abstract, introduction, methodology, results, discussion, and funding. Additionally, it includes a flow diagram to illustrate the procedure used to select the studies included in the review [37,38]. The PRISMA guidelines have been meticulously crafted to enhance the transparency, comprehensiveness, and precision of systematic reviews and meta-analyses. As it comprises a checklist of essential items and a flow diagram, PRISMA serves as a blueprint for researchers to adhere to when presenting their findings. By embracing PRISMA, researchers guarantee that their systematic reviews and meta-analyses are conveyed in a transparent and standardized way, thereby bolstering the credibility and replicability of their results (Supplementary Materials).

2.2. Eligibility Criteria

This study considered English-language articles examining the effects of biodiesel on tractor engine performance and gas emissions for inclusion. Studies without complete data, hence without a so-called “control group” (i.e., studies presenting also some experiments on the same tested engines but not utilizing biodiesel, to be used as a comparison), were excluded. Furthermore, unpublished papers, review articles, and studies in languages other than English (i.e., de facto, the international language for science) were omitted from the analysis.

2.3. Information Sources and Search Strategies

We conducted a comprehensive search using the following keywords, separated by the logical connector “OR” to maximize the search breadth: “diesel” OR “biodiesel” OR “torque” OR “engine performance” OR “tractor” OR “gas emissions” OR “combustion engines” OR “CO₂ and CO emissions” OR “NO emissions” OR “fuel consumption” OR “agriculture machines”. These keywords were queried across multiple databases, including Google Scholar (https://scholar.google.com/ accessed on 10 May 2024), Elsevier (https://www.elsevier.com accessed on 10 May 2024), Science Direct (http://www.sciencedirect.com accessed on 10 May 2024), Wiley Online Library (http://www.onlinelibrary.wiley.com accessed on 10 May 2024), Springer Link (http://www.springer.co.in accessed on 10 May 2024), and Scopus (http://www.scopus.com accessed on 10 May 2024). No limitation concerning the year of publication or the source (hence: the journal of publication of a study) was introduced in the query. Our search, concluded on 10 May 2024, resulted in a total of 415 articles. We included all articles in the initial review, excluding duplicates and selecting relevant ones for potential inclusion in our study.

2.4. Data Synthesis

The data were collected according to the following keys: authors and publication year, characteristics of the study sample, blending ratio of biodiesel, biodiesel source, and engine type. Following the above-declared eligibility criteria, at the end of the selection process, 23 papers out of the initial 415 articles (i.e., the ones resulting from the first selection) were included in the current study. The PRISMA flow diagram (Figure 1) illustrates the identification, screening, and selection process, as well as the number of final papers included.

2.5. Biases and Measures

Funnel plot and Egger’s regression intercept were employed to assess possible biases, while Comprehensive Meta-Analysis (CMA) V3 software (by Biostat, Englewood, NJ, USA; https://meta-analysis.com/ accessed on 10 May 2024) was utilized to evaluate both effect size and biases. Funnel plots serve as visual aids frequently utilized in meta-analysis for evaluating publication bias or detecting small study effects. Within a funnel plot, the effect size of each study is graphed along the vertical axis, juxtaposed against a measure of study precision on the horizontal axis. Egger’s regression intercept constitutes a statistical technique employed in meta-analysis to gauge publication bias. This method entails also conducting a regression analysis of effect estimates in relation to their precision measures. An effect size is an index that measures the magnitude of the effect of one variable (or a set of variables) in a group (i.e., a set of papers/experiments) compared with another group taken as a reference (in this case, findings referred to the use of biodiesel compared with findings obtained without using biodiesel). Given the heterogeneity among groups, random-effects models were employed. The pooled standard errors of the mean and statistical heterogeneity were examined using the Cochrane Q test and I² statistics, respectively [39]. The risk ratio was applied to assess the impact of weights, defined as the relative importance of each group compared with the control group. A risk ratio of 1 means no change. In the present case, it indicates that biodiesel sets show no difference compared to non-biodiesel sets. A tendency to the left or right indicates a change relative to the non-biodiesel set. For example, +1.23 and −1.45 indicate a +23% increase and a −45% decrease of output relative to the non-biodiesel set, respectively.

Table 1. Main indexes used in meta-analysis and meta-regression.

Statistic Result Parameter	Overall Range	Explanation
RR (risk ratio or effect size)	0–unlimited	It quantifies the magnitude of an effect or relation. For the same engine, it is numerically equal to the (percentage) level of a studied output referred to an engine fueled with petrol fuel. For example: 1 (or 100%): For a studied output (e.g., the max torque), there are no differences between engines fueled with biodiesel blends and engines fueled with petrol fuel. <1 (<100%) or >1 (>100%): For a studied output, engines fueled with biodiesel blends have a performance that is numerically equal to the risk ratio of the same engines fueled with petrol fuel (e.g., for maximum torque, the overall risk ratio is 0.857, hence the maximum torque in engines using biodiesel is in general 85.7% of the maximum torque of engines not using biodiesel).
I2 (I-squared)	0–100%	It quantifies the degree of inconsistency/heterogeneity/variability of results among the many considered studies, guiding the interpretation and generalizability of findings. The I2 is scaled as follows: 0–25% (low heterogeneity); 25–50% (moderate heterogeneity); 50–75% (high); 75–100% (very high heterogeneity).
Q (Cochrane parameter)	no lower or upper limits	It is used to assess the presence of heterogeneity among the effect sizes of individual studies included in the meta-analysis. It complements other measures, such as the I-squared, in quantifying and understanding the variability across studies. p-value < 0.05: significant heterogeneity. p-value > 0.05: non-significant heterogeneity.
Y (output variation)	no lower or upper limits	Variation of an output in a regression equation. The mathematical relation of this parameter with RR is Y = 1 − RR.

presents the main indexes used in meta-analysis and meta-regression.

2.6. Meta-Regression

The influence of possible “moderator factors” (i.e., factors able to further alter the final output of an experiment), such as the biodiesel source and the engine type, was taken into consideration. The slope coefficient and p-value were calculated for each covariate in relation to the outcome of interest. The p-value indicates the significance of a potential association, while the slope coefficient measures the strength of the association, indicating how much the outcome change for each unit changes in the covariate.

3. Results and Discussion

3.1. Descriptive Data

Table 1 presents the main characteristics and data extracted from the selected studies, which were collected from various years between 2006 and 2023. The distribution of the studies included in the meta-analysis is as follows: 2006 (n = 1), 2007 (n = 1), 2008 (n = 1), 2009 (n = 1), 2011 (n = 3), 2012 (n = 2), 2013 (n = 3), 2014 (n = 2), 2016 (n = 1), 2018 (n = 2), 2020 (n = 3), 2021 (n = 1), 2022 (n = 1), and 2023 (n = 1) (Figure 2). We excluded several studies in final screening due to a lack of groups without biodiesel [40,41], unclear data in the charts, the lack of precise values [42] and the use of languages different from English [43,44]. Biodiesel was blended with other fuels in concentrations ranging from 5% to 100%. The primary sources of biodiesel fuels were plant-based materials (vegetable oils) and their byproducts, with a small portion derived from animal sources (Figure 1). A diverse range of tractors featuring both turbocharged and non-turbocharged engines were examined, with a predominant focus on Kubota tractors in the majority of studies. Because the authors were from African, East Asian, and Middle Eastern countries, this trend may be more prevalent in those regions. Studies were conducted on various plant and animal sources (

Table 2. Characteristics of the studies (BR: blending ratio; EP: engine power; SFC: specific fuel consumption; Cont: control group without biodiesel; Bio: bio-based fuel/fuel blend group).

Authors, Ref.	Year	BR (%)	Biodiesel Source	Tractor Manufacturer and Type	Engine Specs.	Torque (N·m)		EP (kW)		SFC (kg·kW−1·h−1)		CO2 (%)		CO (ppm)		NO (ppm)
						Cont	Bio	Cont	Bio	Cont	Bio	Cont	Bio	Cont	Bio	Cont	Bio
Tomić et al. [28]	2013	15.0	Sunflower	Mahindra 6500 4WD	Direct injection, four-stroke			28.68	28.69	363.00	368.00	7.79	7.63	174	171	779	763
Tomić et al. [28]	2013	25.0	Sunflower	Mahindra 6500 4WD	Direct injection, four-stroke			28.68	28.36	363.00	369.00	7.79	7.40	174	167	779	740
Tomić et al. [28]	2013	50.0	Sunflower	Mahindra 6500 4WD	Direct injection, four-stroke			28.68	27.81	363.00	376.00	7.79	7.33	174	158	779	733
Tomić et al. [28]	2013	75.0	Sunflower	Mahindra 6500 4WD	Direct injection, four-stroke			28.68	27.51	363.00	394.00	7.79	7.19	174	154	779	719
Tomić et al. [28]	2013	100.0	Sunflower	Mahindra 6500 4WD	Direct injection, four-stroke			28.68	27.37	363.00	412.00	7.79	7.09	174	152	779	709
Emaish et al. [29]	2021	5.0	Frying oil	Kubota M-90	Direct-injection turbocharger	475.60	460.30	18.90	18.00	363.00	380.00	9.03	8.90	159	153	460	463
Emaish et al. [29]	2021	20.0	Frying oil	Kubota M-90	Direct-injection turbocharger	475.60	442.10	18.90	16.70	363.00	386.00	9.03	8.93	159	138	460	466
Emaish et al. [29]	2021	100.0	Frying oil	Kubota M-90	Direct-injection turbocharger	475.60	428.50	18.90	16.10	363.00	400.00	9.03	8.33	159	83	460	484
Venkatesan and Nallusamy, [45]	2020	PO100	Soapnut and pine oils	Simpson S217 twin cylinder	Inline direct injection diesel engine, naturally aspirated									250	400	700	780
Venkatesan and Nallusamy, [45]	2020	PO75:SO25	Soapnut and pine oils	Simpson S217 twin cylinder	Inline direct injection diesel engine, naturally aspirated									250	220	700	770
Venkatesan and Nallusamy, [45]	2020	PO50:SO50	Soapnut and pine oils	Simpson S217 twin cylinder	Inline direct injection diesel engine, naturally aspirated									250	270	700	800
Venkatesan and Nallusamy, [45]	2020	PO25:SO75	Soapnut and pine oils	Simpson S217 twin cylinder	Inline direct injection diesel engine, naturally aspirated									250	300	700	850
Venkatesan and Nallusamy, [45]	2020	SO100	Soapnut and pine oils	Simpson S217 twin cylinder	Inline direct injection diesel engine, naturally aspirated									250	350	700	820
Mohebbi et al. [46]	2012	5.0	Waste cooking oil	Motorsazan MT4.244	Turbocharged, four-cylinder direct injection diesel engine	330.00	330.00	65.00	65.00							2000	2000
Mohebbi et al. [46]	2012	20.0	Waste cooking oil	Motorsazan MT4.244	Turbocharged, four-cylinder direct injection diesel engine	330.00	320.00	65.00	63.00							2000	2100
Mohebbi et al. [46]	2012	50.0	Waste cooking oil	Motorsazan MT4.244	turbocharged, four-cylinder direct injection diesel engine	330.00	315.00	65.00	62.00							2000	2200
Mohebbi et al. [46]	2012	100.0	Waste cooking oil	Motorsazan MT4.244	Turbocharged, four-cylinder direct injection diesel engine	330.00	310.00	65.00	60.00							2000	2300
Aybek et al. [47]	2011	10.0	Canola oil	Massey Ferguson 3056 2WD	Perkins/direct injection	139.90	136.90	32.23	31.54	313.73	310.40
Aybek et al. [47]	2011	20.0	Canola oil	Massey Ferguson 3056 2WD	Perkins/direct injection	139.90	138.90	32.23	32.00	313.73	303.12
Aybek et al. [47]	2011	30.0	Canola oil	Massey Ferguson 3056 2WD	Perkins/direct injection	139.90	136.90	32.23	31.54	313.73	306.91
Li et al. [48]	2006	20.0	Soybean	-	-					371.20	350.10					180	170
Li et al. [48]	2006	50.0	Soybean	-	-					371.20	371.20					180	200
Li et al. [48]	2006	100.0	Soybean	-	-					371.20	371.20					180	210
Kim et al. [49]	2013	20.0	Animal fat	DAEDONG INS./4A220LWS	Four-stroke diesel engine, direct injection			34.80	34.67	391.21	415.21	8.51	8.30	93	90	503	570
Kim et al. [49]	2013	50.0	Animal fat	DAEDONG INS./4A220LWS	Four-stroke diesel engine, direct injection			34.80	34.00	391.21	429.32	8.51	8.10	93	88	503	544
Kim et al. [49]	2013	100.0	Animal fat	DAEDONG INS./4A220LWS	Four-stroke diesel engine, direct injection			34.80	32.93	391.25	445.12	8.51	7.80	93	86	503	519
Gomaa et al. [50]	2014	20.0	Castro oil	Kubota M1-100S-DT	Indirect injection Turbocharged	1000.00	650.00	50.00	43.00
Gomaa et al. [50]	2014	20.0	Palm oil	Kubota M1-100S-DT	Indirect injection Turbocharged	1000.00	800.00	50.00	41.00
Gomaa et al. [50]	2014	20.0	Frying oils	Kubota M1-100S-DT	Indirect injection Turbocharged	1000.00	800.00	50.00	35.00
Sasmito et al. [51]	2022	20.0			Turbocharger	42.00	42.00	23.26	23.55	273.00	256.00
Sasmito et al. [51]	2022	30.0			Turbocharger	42.00	44.00	23.26	23.53	273.00	425.00
Sasmito et al. [51]	2022	100.0			Turbocharger	42.00	40.00	23.26	22.37	273.00	283.00
Gokalp et al. [52]	2011	5.0	Soybean	Basak MR40	Naturally aspirated four cylinder and direct injection diesel engine	189.00	187.00	39.00	39.00					800	780	1100	1100
Gokalp et al. [52]	2011	20.0	Soybean	Basak MR40	Naturally aspirated four cylinder and direct injection	189.00	182.00	39.00	38.00					800	700	1100	1200
Gokalp et al. [52]	2011	50.0	Soybean	Basak MR40	Naturally aspirated four cylinder and direct injection	189.00	179.00	39.00	38.00					800	600	1100	1200
Gokalp et al. [52]	2011	100.0	Soybean	Basak MR40	Naturally aspirated four cylinder and direct injection	189.00	175.00	39.00	37.50					800	450	1100	1400
Müllerová et al. [53]	2011	20.0	Rape seed	Hürlimann H-488 DT	Naturally aspired									180	161	11	12
Müllerová et al. [53]	2011	20.0	Rape seed	Hürlimann XB Max 100	Naturally aspired									105	91	5	6
Soltani Nazarloo et al. [54]	2016	10.0		Massey Ferguson A4-248	Naturally aspirated, Direct-injection	450.00	320.00	25.00	17.00	500.00	481.00	8.50	6.50			248	275
Soltani Nazarloo et al. [54]	2016	20.0		Massey Ferguson A4-248	Naturally aspirated, Direct-injection	450.00	350.00	25.00	23.00	500.00	481.00	8.50	5.00			248	350
Soltani Nazarloo et al. [54]	2016	50.0		Massey Ferguson A4-248	Naturally aspirated, Direct-injection	450.00	340.00	25.00	20.00	500.00	481.00	8.50	2.50			248	400
Al-Aseebee et al. [55]	2023	20.0	Oleic acid methyl ester	Kubota M1-100S-DT	Four strokes, indirect injection, turbocharged, liquid cooled diesel							7.51	7.65	75	55	986	1899
Al-Iwayzy and Yusaf, [56]	2013	20.0	Microalgae	John Deere 4410 e-hydro	Naturally aspired	79.10	78.23					12.10	12.02	90	85	970	994
Neel et al. [57]	2008	20.0	-	John Deere 3203	Three-cylinder, four-stroke, naturally aspirated, compression-ignition engine	316.25	307.80	17.89	17.41	325.00	337.00					471	492
Neel et al. [57]	2008	100.0	-	John Deere 3203	Three-cylinder, four-stroke, naturally aspirated, compression-ignition engine	316.25	284.51	17.89	16.06	325.00	363.00					471	521
Neves et al. [58]	2018	5.0	Murumuru + Soybean							275.00	277.00
Neves et al. [58]	2018	15.0	Murumuru + Soybean							275.00	283.00
Neves et al. [58]	2018	25.0	Murumuru + Soybean							275.00	287.00
Neves et al. [58]	2018	50.0	Murumuru + Soybean							275.00	396.00
Neves et al. [58]	2018	75.0	Murumuru + Soybean							275.00	308.00
Neves et al. [58]	2018	100.0	Murumuru + Soybean							275.00	321.00
Volpato et al. [59]	2012	100.0	Olive oil	Massey Ferguson 275 Compact	Injection system with rotating pump, four cylinders	645.20	577.95	27.50	24.40	270.00	380.00
Barbosa et al. [60]	2018	20.0	Macaw palm oil	Valtra A950	Turbo aspirated injection system with rotary pump and direct injection	320.00	350.00	17.00	21.00
Barbosa et al. [60]	2018	50.0	Macaw palm oil	Valtra A950	Turbo aspirated injection system with rotary pump and direct injection	320.00	280.00	17.00	15.00
Barbosa et al. [60]	2018	80.0	Macaw palm oil	Valtra A950	Turbo aspirated injection system with rotary pump and direct injection	320.00	280.00	17.00	17.00
Barbosa et al. [60]	2018	100.0	Macaw palm oil	Valtra A950	Turbo aspirated injection system with rotary pump and direct injection	320.00	250.00	17.00	12.00
Ramesh [61]	2014	20.0	Karanja oil	Mahindra585DI 2WD	Naturally aspired							14.99	14.08	1309	1289	483	509
Ramesh [61]	2014	40.0	Karanja oil	Mahindra585DI 2WD								14.99	13.35	1309	1271	483	548
Ramesh [61]	2014	60.0	Karanja oil	Mahindra585DI 2WD								14.99	13.03	1309	1207	483	596
Shrivastava et al. [62]	2007	20.0	Karanja oil					24.21	23.77	271.00	282.00
Shrivastava et al. [62]	2007	40.0	Karanja oil					24.21	23.49	271.00	282.00
Shrivastava et al. [62]	2007	60.0	Karanja oil					24.21	23.30	271.00	289.00
Shrivastava et al. [62]	2007	80.0	Karanja oil					24.21	23.03	271.00	290.00
Shrivastava et al. [62]	2007	100.0	Karanja oil					24.21	22.76	271.00	291.00
Nuanual et al. [63]	2020	12.0	Jatropha podagrica	Kubota M7040						398.00	399.00	7.2	4.2	700	420
Nuanual et al. [63]	2020	88.0	Jatropha podagrica	Kubota M7040						398.00	399.00	7.2	2.6	700	250
Venkatesan et al. [64]	2020	10.0	Soapnut oil	Simpson S217	Inline direct-injection diesel engine, naturally aspirated					350.00	400.00			600	500	600	700
Venkatesan et al. [64]	2020	20.0	Soapnut oil	Simpson S217	Inline direct-injection diesel engine, naturally aspirated					350.00	400.00			600	400	600	750
Venkatesan et al. [64]	2020	30.0	Soapnut oil	Simpson S217	Inline direct-injection diesel engine, naturally aspirated					350.00	400.00			600	400	600	800
Venkatesan et al. [64]	2020	100.0	Soapnut oil	Simpson S217	Inline direct-injection diesel engine, naturally aspirated					350.00	400.00			600	200	650	820
Gravalos et al. [65]	2009	5.0	Mixed vegetable oil	ZETOR 7745	Direct fuel injection, turbocharged			58.00	57.00	270.00	280.00
Gravalos et al. [65]	2009	30.0	Mixed vegetable oil	ZETOR 7745	Direct fuel injection, turbocharged			58.00	56.00	270.00	285.00
Gravalos et al. [65]	2009	50.0	Mixed vegetable oil	ZETOR 7745	Direct fuel injection, turbocharged			58.00	55.00	270.00	290.00
Gravalos et al. [65]	2009	100.0	Mixed vegetable oil	ZETOR 7745	Direct fuel injection, turbocharged			58.00	54.00	270.00	300.00

). Only canola oil was examined in two studies while other sources were each investigated in a single study.

3.2. The Effects of Biodiesel on Torque

Figure 3 presents the forest plot results illustrating the impact of biodiesel on maximum torque. This type of plot visually represents effect size, lower and upper confidence limits, risk ratio, and p-values. In meta-analysis, the effect size is a crucial metric that quantifies the magnitude of an effect or relation, allowing for the synthesis and comparison of results from diverse studies. It enhances our understanding of the data, aiding in drawing more robust and generalizable conclusions. Confidence intervals (CIs), or lower and upper limits, provide essential information about the reliability and precision of the estimated effects, helping researchers interpret results within the broader context of uncertainty and variability.

The p-value indicates the significance of the effects observed in experimental setups (with biodiesel) compared to control setups (without biodiesel). A p-value less than 0.05 means that the biodiesel blends, which were experimented in the considered study, significantly affect torque.

In these plots, the value “1” at the center marks the boundary between the two data categories (without biodiesel and with biodiesel). A shift of the plot towards the left indicates that adding biodiesel results in a decrease in torque. The results showed that most of the studies are centered, while some are placed at the left. As depicted in the figure, in all studies, the effect size ranges indeed from 0.1 to 1 for the variable indicating the relatively significant effects of biodiesel in reducing torque (i.e., the risk ratio), with a leftward-leaning graph (without biodiesel). In the last row, the overall “risk ratio” (last line of Figure 3) is 0.868, indicating that the maximum torque in engines using biodiesel is 87% of that in engines not using biodiesel. In other words, using biodiesel reduced torque by 13.0% on average [CI 95% (6.7–19.3%); I² = 97.67; p = 0.000; Q = 129.94, p-value of Q = 0.001]. Most studies attributed the observed reduction in torque to several factors, particularly the lower heating value of biodiesels. The lower energy content of biodiesels results in a slight decrease in peak torque output, as noted, because this fuel provides less energy per unit volume [66]. Furthermore, the higher viscosity of biodiesel can affect the injection process, potentially leading to further changes in torque production [67]. The associated I² parameter is critical in meta-analysis as it quantifies the degree of inconsistency among study results, guiding the interpretation and generalizability of findings. The high I² value of this case (97.67%) indicates considerable heterogeneity, suggesting variability in torque effects across different studies. The Q statistic is fundamental in meta-analysis for detecting the presence of heterogeneity among study results. A high Q value (here: 129.94) indicates significant differences in effect sizes among the studies. A p-value < 0.05 for Q confirms significant heterogeneity.

Figure 4 presents three scatter plots illustrating the relations, respectively, between the blending ratios (numerical factor; on the top), the engine type (categorical factor; in the middle), the biodiesel sources (categorical factor; at the bottom), and the experimented torque changes (represented by the risk factor).

There is a significant linear relation (p = 0.0001) between increasing blending ratios and decreasing torque values, described by the following equation, where Y is the (percentage) variation of torque and BR is the (percentage) blending ratio:

Y [%] = + 6.89 − 0.53 × BR [%].

(1)

The R² for this equation is 0.74, indicating that the blending ratios account for 74.0% of the torque variation, highlighting the substantial predictive capability of the indicated regression equation, including only the blending ratio as an independent variable. It is worth underlying that the intercept of Y in this case (and in the following equations, too) is not zero due to the dispersion of values in the considered range (0–100%) of blending ratios. This caused a repositioning of the regression line to better fit the values (according to the least squares algorithm) and, hence, a not-null intercept (in particular with the sign opposite to the angular coefficient). The passage point of this regression line with the horizontal axis can be interpreted as a minimum threshold below which the presence of biodiesel in the fuel blend does not have any influence on the considered parameter. In this case, estimates suggested that biodiesel did not adversely affect torque until the blending ratio exceeded about 13%. This is also confirmed by the findings in Table 1 and the meta-analysis results in Figure 1, according to which it emerges that lower ratios did not significantly decrease torque. Instead, higher blending ratios significantly reduced torque compared to lower ratios. As shown in Figure 1, a 20% decrease in torque (precisely 19.6%, as it results from the regression equation) was observed at a 50% blending ratio. The decrease at a 100% biodiesel ratio was nearly 46% (precisely −46.1%, as it results from the regression equation), representing a substantial reduction. A safe range for no adverse effect on torque is below 13% of the blending ratio. However, increasing the blending ratio beyond this point resulted in decreased torque. This reduction in torque with higher blend rates was attributed by the many authors of the articles included in this analysis to the increased viscosity and lower heating value of biodiesel. It appears that lower blending ratios do not exhibit such a high viscosity and reduced heating value that significantly impact torque. Therefore, a maximum blending ratio of 20% may be considered optimal not to decrease sensibly the maximum torque.

The meta-regression results did not show significant effects of engine types (turbo and non-turbo) and of biodiesel sources on torque, as visible from the second and third pictures of Figure 4. Indeed, in both these graphs, the mean values of the represented categories (on the horizontal axis) are substantially aligned, and the (vertical) dispersion of points is due to other factors (mainly the blending ratio, as discussed above).

Therefore, it is possible to state that biodiesel sources and engine types do not significantly predict any variation on the torque, as their effects are negligible. It was expected that the type of tractor engine would be a good predictor of torque variations because non-modern tractors (often non-turbocharged) may encounter more substantial reductions in torque due to a higher, expectable biodiesel’s influence on combustion and energy content. Instead, modern tractors with electronic fuel injection systems can be tuned to optimize biodiesel performance. Through precise adjustments in fuel injection timing and quantity, tailored to biodiesel’s properties, these systems can mitigate some of the torque losses associated with biodiesel use [68]. However, the findings indicate that both non-turbo and turbocharged tractors exhibit similar effects on torque, confirming that engine type does not serve as a reliable predictor of torque.

3.3. The Effects of Biodiesel on Engine Power

Figure 5 presents a forest plot demonstrating the impact of biodiesel on maximum engine power. The analysis indicates a significant reduction of −15.0% on average in engine power due to biodiesel use [CI 95% (10.0–20.0%); I² = 54.82; p = 0.000; Q = 101.81, p-value of Q = 0.001]. The plot’s shift towards the left (hence, toward values less than 1) indicates that adding biodiesel results in a decreased maximum engine power. The results show that most studies are centered, with some skewed to the left. The overall risk ratio in the last row is 0.850, indicating that the maximum power in engines using biodiesel is 85% of that in engines not using biodiesel. In other words, using biodiesel reduces engine power by 15%. The findings suggest that biodiesel has adverse effects on engine power, primarily due to its lower heating value [50]. The I² parameter indicates moderate heterogeneity (54.82%), reflecting mean variations between 12% and 65%. Biodiesel blends with ratios up to 7.4% are considered safe (hence not affecting the maximum engine power). However, higher blends, with their lower energy content, lead to a notable decrease in engine power output, with more pronounced effects observed at higher blending ratios.

Figure 6 presents three scatter plots illustrating the relations, respectively, between the blending ratios (top), the engine types (middle), the biodiesel sources (bottom), and the torque changes. There is a significant linear relation (p = 0.0001) between increasing blending ratios and decreasing torque, described by the following equation, where Y is the (percentage) variation of engine power and BR is the (percentage) blending ratio:

Y [%] = + 2.94 − 0.40 × BR [%].

(2)

The results indicate that blending ratios account for 85.0% of the variation in engine power (R² = 0.85), highlighting their substantial predictive capability. In contrast, biodiesel sources and engine types do not significantly affect power variations, and it is possible to state that their effects are negligible. It was anticipated that modern tractors would be able to accommodate biodiesel blends or employ automatic compensation strategies to mitigate torque decreases, leading to minimal impact on power output. However, this expectation was not supported by the reviewed studies. Similar to the findings regarding the torque, the biodiesel sources were not effective predictors of engine power.

3.4. The Effects of Biodiesel on Specific Fuel Consumption

Figure 7 illustrates the forest plot results for the effects of biodiesel on specific fuel consumption (SFC). The analysis revealed significant effects of biodiesel, showing an average increase in SFC of +5.2% [CI 95% (1.0–9.0%); I² = 95.94; p = 0.000; Q = 129.74, p-value of Q = 0.001]. The plot’s shift towards the right indicates that adding biodiesel results in an increase in SFC. The results show that most studies are centered, with some skewed to the right. The overall risk ratio reported in the last row is 1.052, indicating that the SFC in engines using biodiesel is, on average, 105.2% of that in engines not using biodiesel. In other words, using biodiesel increases SFC by 5.2%. Higher biodiesel blending ratios, indeed, may slightly reduce the energy content per volume compared to pure diesel, potentially leading to increased consumption [69]. Biodiesels caused an increase in SFC to a relatively small extent (5.2%). The results showed that most studies had a size effect close to zero. The I² parameter was high, showing higher heterogeneity among data.

Figure 8 presents three scatter plots illustrating the relations, respectively, between the blending ratios (top), the engine types (middle), the biodiesel sources (bottom), and the SFC changes. There is a significant linear relation (p = 0.0001) between increasing blending ratios and fuel consumption, described by the following equation:

Y [%] = −3.55 + 0.18 × BR [%],

(3)

where Y stands for the (percentage) change in SFC and BR for the (percentage) blending ratio. The results indicate that blending ratios from 0% to 100% account only for 40% (R² = 0.40) of variation (increase, in this case) of the SFC, highlighting the effects of blending ratio on the SFC. At the highest ratios (100%), this increase in fuel consumption is approximately 14%. It shows that specific fuel consumption is affected by the presence of biodiesel in blends.

In contrast, the biodiesel sources and the engine types do not significantly predict any SFC variation, their effects being negligible. Both turbocharged and non-turbocharged tractors are not affected by biodiesel, and engine type was not a good predictor for specific fuel consumption. Interestingly, the SFC of tractors remains largely unaffected by the type of biodiesel used [19]. However, proper engine management and maintenance practices are crucial for optimizing fuel efficiency in both types of tractors when using biodiesel blends [19]. As observed in other studies, the biodiesel sources were not good predictors for possible increases in fuel consumption. It seems that the combinations and chemical compositions of the plant-based fuels do not have significant effects on SFC; instead, the blending ratios are the determining factors.

3.5. The Effects of Biodiesel on CO₂ Emissions

Figure 9 illustrates the forest plot results for the effects of biodiesel on CO₂ emissions. The analysis revealed significant effects of biodiesel in decreasing CO₂ emissions, with a size effect of CO₂ emissions quantifiable in −24.1% on average [CI 95% (15.0–32.0%); I² = 0.198; p = 0.000; Q = 20.04 p-value of Q = 0.561]. This size effect indicates that biodiesel significantly reduces CO₂ emissions compared to setups without biodiesel (p < 0.05). The I² = 20.04% parameter indicates small heterogeneity among the data. The results show that most studies are centered, with some skewed to the left. The overall risk ratio in the last row is 0.759, indicating that CO₂ emissions in engines using biodiesel are on average 75.9% of that in engines not using biodiesel. In other words, using biodiesel decreases CO₂ emissions. The plot’s shift towards the left indicates that adding biodiesel results in a decrease in CO₂ emissions. The CO₂ released during combustion is partially offset by the CO₂ absorbed by plants used to produce biodiesel, though the extent of this offset depends on the efficiency and sustainability of the entire biodiesel production and utilization process [70].

Figure 10 presents three scatter plots illustrating the relations, respectively, between the blending ratios (top), the engine types (middle), the biodiesel sources (bottom), and the changes in CO₂ emissions. There is a significant linear relation (p = 0.0001) between increasing blending ratios and CO₂ emissions, described by the following equation:

Y [%] = + 2.16 − 0.63 × BR [%],

(4)

where Y stands for the (percentage) variation of emitted CO₂ and BR for the (percentage) blending ratio. The results indicate that the blending ratios account for 45.0% (R² = 0.45) of the experimented variation in CO₂ emissions, highlighting the strong effect of blending ratios on CO₂ emissions.

In contrast, the biodiesel sources and the engine types do not significantly predict CO₂ emissions variance, with their effects being negligible. These findings show that increasing blending ratios can decrease CO₂ emissions, while the engine type and the biodiesel sources do not have significant effects.

3.6. The Effects of Biodiesel on CO Emissions

Figure 11 illustrates the forest plot results for the effects of biodiesel on CO emissions. The analysis revealed significant effects of biodiesel in decreasing CO emissions, with an average size effect of −17.5% [CI 95% (16.0–18.0%); I² = 98.62; p = 0.000; Q = 97.69, p-value of Q = 0.001].

The analysis revealed significant effects of biodiesel, showing a decrease in CO emissions. The results show that most studies are centered, with some skewed to the left. The overall risk ratio in the last row is 0.825, indicating that CO emissions in engines using biodiesel are about 82.5% of that in engines not using biodiesel. In other words, using biodiesel decreases CO emissions by 17.5%. The plot’s shift towards the left indicates that adding biodiesel results in a decrease in CO emissions. However, all the studies were on the left side of the plot, confirming their significant effects in reducing CO emissions. The I² value of 97.69 indicates high heterogeneity among the data. Biodiesel combustion typically results in lower CO emissions if compared to conventional diesel [71]. This is because biodiesel contains more oxygen, which enhances the combustion process and reduces the formation of CO, a product of incomplete combustion [72].

Figure 12 presents three scatter plots illustrating the relations, respectively, between the blending ratios (top), the engine types (middle), the biodiesel sources (bottom), and the changes in CO emissions. There is a significant linear relation (p = 0.0001) between increasing blending ratios and CO emissions, described by the following equation:

Y [%] = + 6.46 − 0.71 × BR [%],

(5)

where Y is the (percentage) variation of CO emissions and BR is the (percentage) blending ratio. The results indicate that blending ratios account for 54.0% (R² = 0.54) of the variance in CO emissions, highlighting the effect of blending ratios on CO emissions. The results of the meta-regression revealed that biodiesels at ratios spanning from 50% to 100% can decrease CO emissions by 29% to 64%, respectively. As noted, blending ratios are robust predictors for the reduction in CO emissions and can account for a significant portion of these emissions. The biodiesel sources and the engine types do not significantly affect CO emissions variations, with their effects being negligible. These findings show that increasing blending ratios can decrease CO emissions, while the engine type and the biodiesel sources do not have significant effects.

3.7. The Effects of Biodiesel on NO Emissions

Figure 13 illustrates the forest plot results for the effects of biodiesel on NO emissions. The analysis revealed significant effects of biodiesel in increasing NO emissions, with an average size effect of +11.0% [CI 95% (6.0–15.0%); I² = 98.51; p = 0.000; Q = 157.56, p-value of Q = 0.001]. This relatively large-size effect indicates that biodiesel significantly increases NO emissions. Most studies were on the right side of the plot, demonstrating the increased NO emissions associated with biodiesel compared to controls (without biodiesel). The overall risk ratio in the last row is 1.11, indicating that the NO emissions in engines using biodiesel are 111% of those in engines not using biodiesel. In other words, using biodiesel increases NO emissions by +11%. The overall tendency of the forest plot to the right confirms the higher NO emissions with biodiesel. High heterogeneity is indicated by the elevated I² and Q values. NO emissions are primarily formed at high temperatures during combustion, and biodiesel’s higher oxygen content can increase combustion temperatures, thus leading to higher NO production [73]. It is essential to mention that the increase in NO is a consequence of very complex phenomena, of which the fuel type is only a factor. Using oxygenated blends can enhance the combustion process and contribute to controlling NO emissions by promoting more efficient and cleaner burning at lower temperatures. [74]. It has been reported that the blending ratio of biodiesel can significantly impact the emissions of NO [75].

Figure 14 presents three scatter plots illustrating the relations, respectively, between the blending ratios (top), the engine types (middle), the biodiesel sources (bottom), and the changes in NO emissions. There is a significant linear relation (p = 0.0001) between increasing blending ratios and NO emissions, described by the following equation:

Y [%] = −5.69 + 0.35 × BR [%],

(6)

where Y is the (percentage) variation related to the NO emissions and BR is the (percentage) blending ratio. The results indicate that the blending ratios account for 66.00% (R² = 0.66) of the NO emissions, highlighting the strong effects of the blending ratio on the NO emissions. The findings regarding NO emissions also indicated that in biodiesel ratios ranging from 60% to 100%, NO emissions increased by 15 to 29%. At lower blending ratios, emissions were comparatively lower. Higher ratios of biodiesel contain more oxygen, which can decrease CO and CO₂ emissions but increase NO emissions. Various factors contribute to NO, CO, and CO₂ emissions, but the blending ratio of biodiesels plays a crucial role within the set of investigated factors. The results also suggested that the engine type could not significantly influence gas emissions. Indeed, decreases in CO and CO₂ and increases in NO were observed in both engines but at similar levels. The biodiesel sources and the engine types do not significantly predict NO emissions variation, with their effects being negligible. The results show that blending ratios can decrease NO emissions, while the engine type and the biodiesel sources do not have a significant effect. The feedstock utilized in biodiesel production did not impact emissions characteristics. Although various feedstocks can lead to varying levels of emissions reduction or increase, as depicted in the figures, the observed variability is not significant, so they cannot be considered predictive factors in meta-regression. Therefore, it is not conclusive that a specific feedstock will enhance gas emissions.

3.8. Resume of Observed Results

Table 3. Summary of the effects of biodiesels, their ratios, sources, and engine types on the performance and gas emissions of farm tractor engines (BR: blending ratio).

Parameter	BR	Engine Type	Sources	Equation	Min. BR of Influence	% Variat. at 50% BR	% Variat. at 100% BR
Torque	Linear rel.	No effect	No effect	Y [%] = + 6.89 − 0.53 × BR [%]	13.0	−19.61	−46.11
Power	Linear rel.	No effect	No effect	Y [%] = + 2.94 − 0.40 × BR [%]	7.4	−17.06	−37.06
SFC	Linear rel.	No effect	No effect	Y [%] = − 3.55 + 0.18 × BR [%]	19.7	+5.45	+14.45
CO2	Linear rel.	No effect	No effect	Y [%] = + 2.16 − 0.63 × BR [%]	3.4	−29.34	−60.84
CO	Linear rel.	No effect	No effect	Y [%] = + 6.46 − 0.71 × BR [%]	9.1	−29.04	−64.50
NO	Linear rel.	No effect	No effect	Y [%] = − 5.69 + 0.35 × BR [%]	16.3	+11.81	+29.30

depicts a summary of the effects of biodiesel and their ratios, sources, and engine types on the performance and gas emissions of farm tractor engines.

By only looking at the mechanical performance of farm tractors’ engines fueled with biodiesel blends, it is possible to state that any blending ratio below 7.4% of biodiesel in the blend (i.e., the minimum value between 13.0%, 7.4%, and 19.7%) can be considered “safe” as far as concerns a possible engine performance derating.

Similarly, by only looking at the environmental performance of the same engines fueled with biodiesel blends, the range 9.1–16.3% of blending ratios can be considered optimal to trigger the beneficial effects of biodiesel on the CO (evident for blending ratios higher than 3.4%) and CO₂ emissions (for blending ratios higher than 9.1%) without raising sensibly the NO emissions (if blending ratios are lower than 16.3%).

Unfortunately, as evidenced by these considerations, these two blending ratio intervals do not intersect. This means that a range of blending ratios able, at the same time, to minimize the emissions of CO and CO₂, not increase the NO emissions, not decrease the torque and the power, not increase the SFC, does not exist. It is necessary, therefore, to find a tradeoff solution that, for example, privileges the environmental performance without decreasing the engine mechanical performance too much. The range of blends having a share of biodiesel from 9.1% to 13.0% offers a good compromise: within this range of percentages of biodiesel, it is possible to provide optimal environmental benefits (the indicated range is fully included in the above-individuated range 9.1–16.3%, optimal for environmental performance), with only a minor impact on power (−2.3% at a 13% blending ratio) and no significant effects on torque or SFC.

3.9. Bias

A bias in meta-analysis refers to systematic errors that can compromise the validity and reliability of the results. Addressing and minimizing these biases is crucial for providing a more accurate and reliable summary of the existing evidence. The absence of publication bias among the included studies was assessed using funnel plots, as shown in Figure 15. Funnel plots help to visually detect publication bias by visualizing the distribution of study results. Each publication bias is indicated by an asymmetrical distribution of studies (represented as white circles) outside of the inclined lines forming the “funnel”. An asymmetrical plot suggests the presence of bias. In the reported figures, the distribution of white circles appears symmetrical around the funnel’s center, indicating a lack of publication bias. Additionally, when two white and black diamonds are aligned below the horizontal axis, this indicates that there is no bias. These observations confirm that there are no significant biases in the studies included in this meta-analysis.

Egger’s test is a statistical method used in meta-analyses to detect publication bias, an alternative to funnel plots. This test assesses the asymmetry of the funnel plot by examining the intercept of the regression line. If the p-value associated with this intercept is less than the chosen significance level (typically 0.05), it indicates significant asymmetry and suggests potential publication bias. The values obtained from the Egger’s test for the studied parameters are reported in

Table 4. Statistical results for bias control.

Variable	Eggers (p)	Fail-Safe (n)
Torque	0.864	4402
Power	0.615	5217
SFC	0.650	2350
CO2	0.950	3503
CO	0.660	3726
NO	0.984	3218

.

Another parameter is fail-safe N, which is a statistical method used in meta-analysis to assess the robustness of the results to potential publication bias. It estimates the number of additional, hypothetical studies with null results that would need to be added to the meta-analysis to reduce the observed overall effect to a non-significant level. The fail-safe N values were 4402, 5217, 2350, 3503, 3726, and 3218 for torque, engine power, specific fuel consumption, and emissions of CO₂, CO, and NO, respectively (see Table 4). These values, very high with respect to the number of considered studies (24), indicate that an unrealistically large number of “null” studies would need to be located and included in order for the combined two-tailed p-value to exceed 0.050 for each of the variables, meaning that the 24 considered studies have a statistical significance that is very high.

4. Conclusions

In conclusion, biodiesels have different effects on engine performance and emissions, with the dominant factor being the blending ratio. There is no single feedstock that can be definitively deemed optimal, as the impacts on engine parameters and pollutant emissions do not consistently align. Consequently, identifying an ideal biodiesel blending ratio, i.e., able to maintain engine performance while reducing gas emissions, is not feasible because the response of parameters to different biodiesel blending ratios is highly different. The range of 9.1% to 13.0% biodiesel appears to be a safe and practical option for substituting fossil fuels. This range effectively maintains engine performance with only a modest decrease in power (up to 4%) while achieving reductions in CO and CO₂ emissions (respectively by 0/−2.8% and −3.6%/−6.0%). These improvements can be obtained without the need for additional pollutant abatement systems.

The current study’s strength lies in the novelty of its approach, able to explore and summarize the results on engine parameters and gas emissions of many experiments involving biodiesels. However, some limitations exist, such as the focus solely on farm tractor engines, meaning that the results may not be generalized to other engine types unless including also other types of engines in the study since the initial creation of the set of articles constituting the base of the study. Future studies should delve into more detailed considerations, such as operational conditions and engine speed, to provide a comprehensive understanding of biodiesel effects with the illustrated holistic approach using the literature results. It is also suggested to use machine learning data analysis tools to cluster the results from the literature, given some inputs (e.g., blending ratio, type of engine, type of fuel injection system, type of emissions abatement system, etc.), so as to highlight the results in terms of performance and emissions for each clustered category.