Saraca asoca seeds – A novel candidature for biodiesel production: Studies on yield optimization using ANN coupled GA and properties of biodiesel blends

ABSTRACT

Biodiesel, an alternative green energy for fossil diesel, is very much inevitable and valuable in two ways when derived from non-edible oil resources: one, significantly reducing the production costs and the other, eliminating the food vs. fuel dispute. One such novel feedstock, namely, Saraca asoca seeds have been investigated to produce biodiesel. This work uncovers the potential of Saraca asoca seeds as novel feedstock for biodiesel and its modeling-optimization done with ANN-GA, and evaluation of fuel properties of biodiesel and its blend with diesel. The acid value of S. asoca seed oil (8.43 mg KOH/g) was >2 mg KOH/g. Hence, a dual stage acid-base transesterification was employed. Transesterification parameters, namely, methanol to oil mole ratio, catalyst concentration, reaction time, and temperature were optimized using ANN-GA. The optimal ANN topology (4–7-1) was obtained by the heuristic analysis of R and MSE values. After optimization, the maximum yield predicted by ANN-GA was 97.04% corresponding to the experimental yield of 96% achieved at 69.99°C with methanol to oil ratio of 10.24:1 and catalyst concentration of 1.18 wt% in 50.11 min. Hypothesis tests were accomplished to prove the validation of the developed ANN-GA model. The significant fuel properties of S.asoca biodiesel and its blends with diesel were investigated and compared with ASTM D6751 and EN 14214 standards. The new S. asoca biodiesel and its blends of diesel have satisfactorily complied with the above standards and could be a reliable alternative to diesel in diesel engine applications.

KEYWORDS:

• Saraca asoca seed oil
• biodiesel
• artificial neural network
• genetic algorithm
• fuel-blend properties

Nomenclature

List of abbreviations	=
ANN	=	Artificial Neural Network
AOCS	=	American Oil Chemists’ Society
ASTM	=	American Society for Testing Materials
ATSO	=	Aleurites trisperma seed oil
BY	=	Biodiesel Yield
CCD	=	Central Composite Design
EN	=	European Norms
FFA	=	Free Fatty Acid
GA	=	Genetic Algorithm
LM	=	Lavenberg Marquardt
MSE	=	Mean Square Error
R	=	Coefficient of determination
RSO	=	Rubber seed oil
SAB	=	S. asoca biodiesel
SASO	=	S. asoca seed oil
SMSO	=	Silybum marianum L. seed oil
List of symbols	=
	=	methoxy protons integrated peak area,
	=	methylene protons integrated peak area
	=	bias of hidden layer
	=	bias of output layer
	=	weight from input layer to hidden layer
	=	weight from hidden layer to output layer
	=	Number of inputs in input layer

Supplementary Material

Supplementary data for this article can be accessed here