Evaluation of Predictive Models for the Viscosity of Biodiesel

Abstract

Viscosity is an important biodiesel parameter, subject to specifications and with an impact on the fuel quality. A model that could predict the value of viscosity of a biodiesel based on the knowledge of its composition would be useful in the optimization of biodiesel production processes and the planning of blending of raw materials and refined products. This work aims at evaluating the predictive capability of several models previously proposed in the literature for the description of the viscosities of biodiesels and their blend with other fuels. The models evaluated here are Ceriani's, Krisnangkura's, and Yuan's models, along with a revised version of Yuan's model proposed here. The results for several biodiesel systems show that revised Yuan's model proposed provides the best description of the experimental data with an average deviation of 4.65%, as compared to 5.34% for Yuan's model, 8.07% for Ceriani's model, and 7.25% for Krisnangkura's model. The same conclusions were obtained when applying these models to predict the viscosity of blends of biodiesel with petrodiesel.

Measurement and prediction of speeds of sound of fatty acid ethyl esters and ethylic biodiesels

“…As biodiesels are composed of fatty esters, the knowledge of the properties of these pure compounds enables the prediction of the properties of those fuels as we have shown elsewhere for other biodiesel properties [10,14,15,22]. Unfortunately there are but a few data available in the literature for fatty acid methyl esters (FAMEs) [13,21,12,[23][24][25][26][27] being the experimental speeds of sound of fatty acid ethyl esters (FAEEs) even more scant [28], although some studies were already done for the shorter methyl and ethyl esters [29].…”

Artificial neural network models to predict density, dynamic viscosity, and cetane number of biodiesel

“…A total of 780 experimental points were obtained from various scientific publications [4,7,8,[15][16][17][18][19][20][21][22][23][24][25][26][27][28] to estimate the density and dynamic viscosity of biodiesel. In this case, the database was randomly divided into three groups with 60% to training, 20% to testing and 20% to validation.…”

“…Freitas et al [7] used different models for predicting the viscosity of biodiesel at various temperatures. Pratas et al [8] proposed a methodology based on the Kay's mixing rule and the group contribution method for predicting the density of ten samples of biodiesel as a function of temperature.…”

Comparison of empirical equations and artificial neural network results in terms of kinematic viscosity prediction of fuels based on hazelnut oil methyl ester

“…The purelin function is used for linear relationship approximation as its name implies and the general definition is expressed as Eqs. (10), (12), and (14) [43].…”