Assessing the correlation between fatty acid composition of biodiesel with the fuel property using artificial intelligence and optimization

This work explores the underlying correlation between fatty acid composition and biodiesel fuel properties using artificial neural network (ANN) and establishes the complex non-linear relationship between the fuel parameters and the significant fatty acids present in the composition of biodiesel. Six physico-chemical properties of biodiesel, that is, specific gravity, (o)API, aniline point, cetane number, flash point, and calorific value have been assessed and a modeling framework has been developed. High value of R-2 and low value of RMSE signify that ANN model captures the inherent relationship. Results demonstrate that with 1% change in arachidic acid, an increment of 21.74 MJ/Kg in heating value and a decrease of 7.17 in cetane number of biodiesel fuel take place, respectively. This suggests that the presence of a regulated amount of arachidic acid in biodiesel composition would assist in obtaining desired heat value and diesel index of biodiesel that would ultimately aid in improving fuel quality and hence, would help in obtaining higher grade biodiesel fuel. Furthermore, ANN-based stochastic optimization techniques, genetic algorithm, and particle swarm optimization have been used to predict the maximum possible biodiesel blend that can be used in diesel engines retaining the fuel qualities.

Automated summary

This study utilized artificial neural network to explore the correlation between fatty acid composition and biodiesel fuel properties, establishing a complex non-linear relationship that can help optimize fuel quality by adjusting the content of specific fatty acids.