Full Length ArticleModeling of vegetable oils cloud point, pour point, cetane number and iodine number from their composition using genetic programming
Introduction
Vegetable oils (VOs) are a complex mixture of 90–98% of triglycerides and small amounts of mono- and di-glycerides, with 1–5% of free fatty acids. Triglycerides are esters of three fatty acids and glycerol. The fatty acids present in VOs often vary in carbon chain length and unsaturation [1].
In compression ignition (CI) engines, several alternative fuels have been tested in order to reduce the dependence on fossil fuels, such as VOs, bio-oil [2], biodiesel [3], [4], [5], [6] and alcohols [7], [8]. Among them, VOs are a promising candidate to replace a fraction of fossil fuels consumption.
Concerning VOs applications in diesel engines, due to their physico-chemical properties differences, VOs have led to possible engines problems, if the engines are not properly modified [9]. Consequently, several alternatives have been proposed, such as the viability of using pure VO in large ships [10], the VOs characteristics for uses in stationary diesel engines [11], and in low and medium speed diesel engines for power plants [12], the use of crude filtered VO in diesel engines [13], its application from triglyceride based biomass to CI engines [14], and the airflow characteristics improvement inside the combustion chamber of CI engines [15]. Whereas concerning VOs fundamental combustion studies, several works have been recently carried out in order to understand the differences in the combustion behavior. For instance, the influence of straight VO fatty acid composition on CI combustion and emissions [16], on ignition behavior [17], [18], and on injection parameters [19].
Waste VOs have been considered for uses as a fuel or to produce biodiesel, and several works have been performed on this subject, such as the direct use of waste VO in CI engines [20], the prediction of the higher heating values of waste frying oils [21], the variables affecting the yields and characteristics of the biodiesel from used frying oil [22], and the influence of high temperature and duration of heating on the properties for sunflower seed oil [23] and olive oil [24]. In addition, a comparison of extraction systems and solvents and their influence on tomato seed oil properties was presented by [25].
Blends are a viable alternative to utilize a fraction of VOs along with diesel and other renewable fuels. In this sense, several studies have been carried out, including those of density, viscosity and cloud point of diesel oil blends with straight VOs [26], [27], [28], [29], the effects of the blends of VOs and animal fat on biodiesel product composition and quality [30], the use of VO derived esters as diesel additive [31] and the study of quaternary blends of diesel–biodiesel-VO-pentanol in a diesel engine generator, with the objective to increase the performance and combustion behaviors and to reduce exhaust emissions [9].
Fatty acid composition of VOs has been the focus of intense research over the last decade. It has been used to characterize its physico-chemical properties, such as viscosity [32], [33], [34], [35], [36], density [34], evaporation of droplets [37], ignition and boiling of droplets [38], melting characteristics [39], oxidative stability [40], [41], electrical and ultrasonic properties [42], [43], rheological behaviour [44], [45], tribological and thermal properties [46], and acid and peroxide values [47]. In addition, it has been used to improve biodiesel physical properties [48], [49], to select premium quality vegetable oils [50], and for glycerides conversion to biodiesel [51].
There are about 350 VOs with potential as fuel sources. And for most of these VOs, the properties values have not yet been measured. In this context, regression models can be very useful. In the literature, for some properties, models based on VOs unsaturation are available [52], [53], [54]. For example, the analysis of 22 VOs properties and fatty acid composition has been recently presented by Giakoumis [54]. In this work [54], the authors have concluded that, a high correlation between the iodine number and the degree of unsaturation has been found, whereas the cloud point, pour point, cetane number and oxidation stability seem to have a good correlation with the degree of unsaturation. On the other hand, the degree of correlation for properties such as density, low and high heating values, kinematic viscosity and flash point are between moderate to very weak.
In addition, regression studies of VOs properties from their fatty acid composition have also been presented. However, unlike models of biodiesel from its composition, a theme well explored in the literature, which have been developed for properties such as iodine number [55], [56], [57], [58], cetane number [58], [59], [60], [61], [62], [63], cloud point [58], [64], [65], [66], pour point [58], [64], [65], cold filter plugging point [58], [64], [65], [67], [68], density [59], [60], [62], heating values [59], [62], kinematic viscosity [57], [58], [59], [62], [64], and flash point [69], [64], [58]; for VOs there are only few available models. They include the prediction of properties such as density [70], and high heating value [21], [71], [72]. Moreover, correlations between VOs physico-chemical properties have also been found in the literature [73], [74], [75], [76], [77], [78].
Machine learning is a tool allowing to obtain non-parametric regressions, mainly through artificial neural networks (ANN) and genetic programming (GP). ANN has been used in studies related to VOs [79], [80], [81]. However, one of the main drawbacks of this technique is related to the form of a neural network, from which the resulting relationship expression can not be obtained easily. On the other hand, GP has not yet been used in VOs studies. In addition, in this case the models expressions are obtained without any problem.
In a previous work [58], the authors have presented prediction models of biodiesel properties from its fatty acid composition using genetic programming (GP). In that work, the authors showed that for several properties (including cloud point, pour point, cetane number and iodine number), the agreement of the proposed equations was better for GP models, in comparison to multiple linear regressions available in the literature. For the present work, which can be considered a continuation of [58], the focus is on the prediction models of VOs physico-chemical properties using also genetic programming.
Following the findings of Giakoumis [54] on the correlation of VOs properties and the degree of unsaturation, and taking into account that the degree of unsaturation is obtained directly from the fatty acid composition of each VO, in the present work four physico-chemical properties were chosen to carry out the regression study: cloud point, pour point, cetane number and iodine number. Oxidation stability was not considered due to lack of adequate amount of information in the selected databases. To the best of our knowledge, regression models of VOs for these properties from their fatty acid composition are not yet available in literature. Therefore, this paper proposes these regression models using 4 different experimental databases (fatty acids and selected properties) of VOs, which are presented in the next section.
Section snippets
Experimental databases
Concerning the experimental databases of VOs, there are several works reporting values of their fatty acid composition and corresponding physico-chemical properties, and the exhaustive analysis of these data is not within the scope of the present paper. Thus, we have extracted the values from recent reviews and papers on the subject.
In this sense, several studies and works have been recently published. For instance, prediction models for fats, oils, and biodiesel properties [82], the analysis
Regression using genetic programming
The implementation of GP in the present work will be briefly described next. More information can be found in [58], [85].
Regression models
In this section, : refers to Arachidic fatty acid (wt%), : to Gadoleic’s, : to Linoleic’s, : to Linolenic’s, : to Myristic’s, : to Oleic’s, : to Palmitic’s, : to Palmitoleic’s, and : to Stearic’s.
Conclusions
Vegetable oils (VOs) are mainly composed of triglycerides (90–98%), i.e. esters of 3 fatty acids and glycerol. VOs have been considered for uses in large ships, in stationary engines and in low and medium speed diesel engines for power plants. 350 VOs are considered as possible fuel sources, and for most of them, their physico-chemical properties values have not yet been measured. Regression models using only VOs fatty acid composition are then very useful.
This article has presented regressions
CRediT authorship contribution statement
Dario Alviso: Conceptualization, Investigation, Formal analysis, Writing - original draft. Cristhian Zárate: Software, Validation, Writing - review & editing. Thomas Duriez: Software, Supervision, Writing - review & editing.
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgement
Dario Alviso and Cristhian Zárate would like to thank Conicet, Argentina for the postdoctoral and doctoral scholarships, respectively.
References (91)
- et al.
Vegetable oils as fuels for internal combustion engines: a review
J Agric Eng Res
(1994) - et al.
Experimental and numerical studies of biodiesel combustion mechanisms using a laminar counterflow spray premixed flame
Fuel
(2015) - et al.
Development of a diesel-biodiesel-ethanol combined chemical scheme and analysis of reactions pathways
Fuel
(2017) - et al.
Study of biodiesel production from animal fats with high free fatty acid content
Bioresour Technol
(2011) - et al.
Light refraction effects in counterflow non-premixed flames
Fuel
(2019) - et al.
Investigation on the structural effects of the addition of alcohols having various chain lengths into the vegetable oil-biodiesel-diesel fuel blends: An attempt for improving the performance, combustion, and exhaust emission characteristics of a compression ignition engine
Fuel
(2020) - et al.
Characteristics of vegetable oils for use as fuel in stationary diesel engines-Towards specifications for a standard in West Africa
Renew Sustainable Energy Rev
(2013) - et al.
Use of crude filtered vegetable oil as a fuel in diesel engines state of the art: Literature review
Renew Sustainable Energy Rev
(2010) Application of straight vegetable oil from triglyceride based biomass to IC engines–A review
Renew Sustainable Energy Rev
(2017)- et al.
A review on improving airflow characteristics inside the combustion chamber of CI engines to improve the performance with higher viscous biofuels
Fuel
(2020)
The influence of straight vegetable oil fatty acid composition on compression ignition combustion and emissions
Fuel
Ignition and combustion behaviour of vegetable oils after injection in a constant volume combustion chamber
Biomass Bioenergy
Effect of fatty acid composition on ignition behavior of straight vegetable oils measured in a constant volume combustion chamber apparatus
Fuel
Ignition characteristics of straight vegetable oils in relation to combustion and injection parameters, as well as their fatty acid composition
Fuel Process Technol
Direct use of waste vegetable oil in internal combustion engines
Renew Sustainable Energy Rev
Predicting the higher heating values of waste frying oils as potential biodiesel feedstock
Fuel
The density and cloud point of diesel oil mixtures with the straight vegetable oils (SVO): palm, cabbage palm, cotton, groundnut, copra and sunflower
Biomass Bioenergy
The viscosity of diesel oil and mixtures with straight vegetable oils: Palm, cabbage palm, cotton, groundnut, copra and sunflower
Biomass Bioenergy
Flow properties of vegetable oil-diesel fuel blends
Fuel
Synthesis and characterization of vegetable oil derived esters: evaluation for their diesel additive properties
Bioresour Technol
Temperature dependence of density and viscosity of vegetable oils
Biomass Bioenergy
Model for predicting evaporation characteristics of vegetable oils droplets based on their fatty acid composition
Int J Heat Mass Transfer
The role of pole and molecular geometry of fatty acids in vegetable oils droplet on ignition and boiling characteristics
Renewable Energy
Predicting melting characteristics of vegetable oils from fatty acid composition
LWT-Food Sci Technol
Revealing the relationship between vegetable oil composition and oxidative stability: a multifactorial approach
J Food Comp Anal
The mathematical prediction model for the oxidative stability of vegetable oils by the main fatty acids composition and thermogravimetric analysis
LWT
Electrical properties of vegetable oils between 20 Hz and 2 MHz
Int J Hydrogen Energy
Electrical and ultrasonic properties of vegetable oils and biodiesel
Fuel
Correlation of fatty acid composition of vegetable oils with rheological behaviour and oil uptake
Food Chem
The influence of fatty acids on tribological and thermal properties of natural oils as sustainable biolubricants
Tribology Int
Multiple response optimization of vegetable oils fatty acid composition to improve biodiesel physical properties
Bioresour Technol
Influence of vegetable oils fatty-acid composition on biodiesel optimization
Bioresour Technol
Application of random forests to select premium quality vegetable oils by their fatty acid composition
Food Chem
Influence of vegetable oils fatty acid composition on reaction temperature and glycerides conversion to biodiesel during transesterification
Bioresour Technol
Analysis of 22 vegetable oils’ physico-chemical properties and fatty acid composition on a statistical basis, and correlation with the degree of unsaturation
Renew Energy
Renewable energy sources from Michelia champaca and Garcinia indica seed oils: A rich source of oil
Biomass Bioenergy
Theoretical modeling of iodine value and saponification value of biodiesel fuels from their fatty acid composition
Renew Energy
Application of artificial neural networks to predict viscosity, iodine value and induction period of biodiesel focused on the study of oxidative stability
Fuel
Prediction of biodiesel physico-chemical properties from its fatty acid composition using genetic programming
Fuel
Predicting cetane number, kinematic viscosity, density and higher heating value of biodiesel from its fatty acid methyl ester composition
Fuel
Correlation for the estimation of the density of fatty acid esters fuels and its implications. A proposed biodiesel cetane index
Chem Phys Lipids
Prediction of the cetane number of biodiesel using artificial neural networks and multiple linear regression
Energy Conver Manage
Estimation of biodiesel cetane number, density, kinematic viscosity and heating values from its fatty acid weight composition
Fuel
Thermodynamic study on cloud point of biodiesel with its fatty acid composition
Fuel
Influence of fatty acid composition of raw materials on biodiesel properties
Bioresour Technol
Cited by (17)
Standard specifications for renewable diesel
2023, Renewable Diesel: Value Chain, Sustainability, and Challenges