Thermal properties of ethylic biodiesel blends and solid fraction prediction using artificial neural networks

Author(s)

Magalhaes, Ana M. S. ; Brentan, Bruno M. ; Meirelles, Antonio J. A. ; Maximo, Guilherme J.

• Document type: Journal article
• Source: Fluid Phase Equilibria; v. 574, p. 11-pg., 2023-07-03.

Abstract

The cold flow properties are known to be a significant limiting factor for the use of a vegetable oil source for biodiesel production, as well as their use worldwide. However, the use of ethylic biodiesel, new sources and blends have been shown to improve these properties. Looking for vegetable sources alternatives and use of ethanol, this work evaluated the thermal properties (solid fraction content - SFC - behavior and cold flow properties - CFP) of model ethylic biodiesel from Brazilian biodiversity oils such coconut, ucuuba, macauba, and babassu, including their blends with other commercial sources such as palm, sunflower, rapeseed, and jatropha. Artificial Neural Networks (ANN) was used as a tool for prediction of the biodiesels' SFC behavior, looking for the best alternative to produce biofuels with good CFP. Together with the mentioned biodiesels, the database used was composed by SFC data from literature for soybean, palm stearin, palm olein, palm kernel, beef tallow, macauba kernel, including their blends. A Multilayer Perceptron Feed Forward Neural Network with the Sigmoid function as activation function and the Levenberg Marquardt method as the convergence algorithm was used. The search of the best topology was performed by a Central Composite Rotatable Design considering 15 different topologies (3 hidden layers from 2 to 18 neurons each layer), the best of which was 10.18.10 (10 neurons in the first layer, 18 in the second, and 10 in the third) presenting a Mean Square Error value, of 0.0356 and 0.0519 for training and testing. From the results obtained, the present study demonstrates the ability of blends in lowering the CFP (cloud point, pour point and cold filter plugging point). The ANN could simulate the SFC curves behavior of ethylic biodiesel with high accuracy, considering their ethyl ester composition and temperature, expanding the range of possibilities for biodiesel formulation and improvement of biodiesel behavior in cold weather. (AU)

• FAPESP's process: 14/21252-0 - Equilibrium and production processes of biofuels and bioproducts
• Grantee: Antonio José de Almeida Meirelles
• Support Opportunities: Research Projects - Thematic Grants
• FAPESP's process: 16/10636-8 - From the cell factory to the Biodiesel-Bioethanol integrated biorefinery: a systems approach applied to complex problems in micro and macroscales
• Grantee: Roberto de Campos Giordano
• Support Opportunities: Program for Research on Bioenergy (BIOEN) - Thematic Grants
• FAPESP's process: 21/11380-5 - CPTEn - São Paulo Center for the Study of Energy Transition
• Grantee: Luiz Carlos Pereira da Silva
• Support Opportunities: Research Grants - Science Centers for Development