Elsevier

Fuel

Volume 284, 15 January 2021, 119026
Fuel

Full Length Article
Modeling of vegetable oils cloud point, pour point, cetane number and iodine number from their composition using genetic programming

https://doi.org/10.1016/j.fuel.2020.119026Get rights and content

Abstract

Vegetable oils (VOs) are composed of 90–98% of triglycerides, i.e. esters composed of three fatty acids and glycerol, and small amounts of mono- and di-glycerides. Due to their physico-chemical properties, VOs have been considered for uses especially in large ships, in stationary engines and low and medium speed diesel engines, in pure form or in blends with fuel oil, diesel, biodiesel and alcohols. There are about 350 VOs with potential as fuel sources, and for most of them, physico-chemical properties values have not yet been measured. In this context, regression models using only VOs fatty acid composition are very useful. In the present paper, regression analysis of VOs cloud point (CP), pour point (PP), cetane number (CN) and iodine number (IN) as a function of saturated and unsaturated fatty acids is conducted. The study is done by using 4 experimental databases including 88 different data of VOs. Concerning the regression technique, genetic programming (GP) has been chosen. The cost function of GP is defined to minimize the Mean Absolute Error (MAE) between experimental and predicted values of each property. The resulting GP models consisting of terms including saturated and unsaturated fatty acids reproduce correctly the dependencies of all four properties on those acids. And they are validated by showing that their results are in good agreement to the experimental databases. In fact, MAE values of the proposed models with respect to the databases for CP, PP, CN and IN are lower than 4.51 °C, 4.54 °C, 3.64 and 8.01, respectively.

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, xAR: refers to Arachidic fatty acid (wt%), xGA: to Gadoleic’s, xLI: to Linoleic’s, xLN: to Linolenic’s, xMY: to Myristic’s, xOL: to Oleic’s, xPA: to Palmitic’s, xPL: to Palmitoleic’s, and xST: 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)

  • P. Hellier et al.

    The influence of straight vegetable oil fatty acid composition on compression ignition combustion and emissions

    Fuel

    (2015)
  • P. Emberger et al.

    Ignition and combustion behaviour of vegetable oils after injection in a constant volume combustion chamber

    Biomass Bioenergy

    (2015)
  • M. Plank et al.

    Effect of fatty acid composition on ignition behavior of straight vegetable oils measured in a constant volume combustion chamber apparatus

    Fuel

    (2017)
  • M. Plank et al.

    Ignition characteristics of straight vegetable oils in relation to combustion and injection parameters, as well as their fatty acid composition

    Fuel Process Technol

    (2017)
  • D. Capuano et al.

    Direct use of waste vegetable oil in internal combustion engines

    Renew Sustainable Energy Rev

    (2017)
  • H. Sanli et al.

    Predicting the higher heating values of waste frying oils as potential biodiesel feedstock

    Fuel

    (2014)
  • A. Abollé et al.

    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

    (2009)
  • A. Abolle et al.

    The viscosity of diesel oil and mixtures with straight vegetable oils: Palm, cabbage palm, cotton, groundnut, copra and sunflower

    Biomass Bioenergy

    (2009)
  • Z. Franco et al.

    Flow properties of vegetable oil-diesel fuel blends

    Fuel

    (2011)
  • S. Dmytryshyn et al.

    Synthesis and characterization of vegetable oil derived esters: evaluation for their diesel additive properties

    Bioresour Technol

    (2004)
  • B. Esteban et al.

    Temperature dependence of density and viscosity of vegetable oils

    Biomass Bioenergy

    (2012)
  • T. Daho et al.

    Model for predicting evaporation characteristics of vegetable oils droplets based on their fatty acid composition

    Int J Heat Mass Transfer

    (2012)
  • E. Marlina et al.

    The role of pole and molecular geometry of fatty acids in vegetable oils droplet on ignition and boiling characteristics

    Renewable Energy

    (2020)
  • O. Fasina et al.

    Predicting melting characteristics of vegetable oils from fatty acid composition

    LWT-Food Sci Technol

    (2008)
  • L. Redondo-Cuevas et al.

    Revealing the relationship between vegetable oil composition and oxidative stability: a multifactorial approach

    J Food Comp Anal

    (2018)
  • J. Li et al.

    The mathematical prediction model for the oxidative stability of vegetable oils by the main fatty acids composition and thermogravimetric analysis

    LWT

    (2018)
  • J. Corach et al.

    Electrical properties of vegetable oils between 20 Hz and 2 MHz

    Int J Hydrogen Energy

    (2014)
  • J. Corach et al.

    Electrical and ultrasonic properties of vegetable oils and biodiesel

    Fuel

    (2015)
  • J. Kim et al.

    Correlation of fatty acid composition of vegetable oils with rheological behaviour and oil uptake

    Food Chem

    (2010)
  • C.J. Reeves et al.

    The influence of fatty acids on tribological and thermal properties of natural oils as sustainable biolubricants

    Tribology Int

    (2015)
  • S. Pinzi et al.

    Multiple response optimization of vegetable oils fatty acid composition to improve biodiesel physical properties

    Bioresour Technol

    (2011)
  • S. Pinzi et al.

    Influence of vegetable oils fatty-acid composition on biodiesel optimization

    Bioresour Technol

    (2011)
  • F.-F. Ai et al.

    Application of random forests to select premium quality vegetable oils by their fatty acid composition

    Food Chem

    (2014)
  • S. Pinzi et al.

    Influence of vegetable oils fatty acid composition on reaction temperature and glycerides conversion to biodiesel during transesterification

    Bioresour Technol

    (2011)
  • E.G. Giakoumis

    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

    (2018)
  • K. Hosamani et al.

    Renewable energy sources from Michelia champaca and Garcinia indica seed oils: A rich source of oil

    Biomass Bioenergy

    (2009)
  • A. Gopinath et al.

    Theoretical modeling of iodine value and saponification value of biodiesel fuels from their fatty acid composition

    Renew Energy

    (2009)
  • A.O. Barradas Filho et al.

    Application of artificial neural networks to predict viscosity, iodine value and induction period of biodiesel focused on the study of oxidative stability

    Fuel

    (2015)
  • D. Alviso et al.

    Prediction of biodiesel physico-chemical properties from its fatty acid composition using genetic programming

    Fuel

    (2020)
  • L.F. Ramírez-Verduzco et al.

    Predicting cetane number, kinematic viscosity, density and higher heating value of biodiesel from its fatty acid methyl ester composition

    Fuel

    (2012)
  • M. Lapuerta et al.

    Correlation for the estimation of the density of fatty acid esters fuels and its implications. A proposed biodiesel cetane index

    Chem Phys Lipids

    (2010)
  • R. Piloto-Rodríguez et al.

    Prediction of the cetane number of biodiesel using artificial neural networks and multiple linear regression

    Energy Conver Manage

    (2013)
  • E.G. Giakoumis et al.

    Estimation of biodiesel cetane number, density, kinematic viscosity and heating values from its fatty acid weight composition

    Fuel

    (2018)
  • H. Imahara et al.

    Thermodynamic study on cloud point of biodiesel with its fatty acid composition

    Fuel

    (2006)
  • M.J. Ramos et al.

    Influence of fatty acid composition of raw materials on biodiesel properties

    Bioresour Technol

    (2009)
  • Cited by (17)

    • Standard specifications for renewable diesel

      2023, Renewable Diesel: Value Chain, Sustainability, and Challenges
    View all citing articles on Scopus
    View full text