Comparison of Response Surface Methodology (RSM) and Artificial Neural Network (ANN) in Optimization of Aegle marmelos Oil Extraction for Biodiesel Production.

Selvan, S. Sindhanai; Pandian, P. Saravana; Subathira, A.; Saravanan, S.

Non-edible feedstock is attaining importance due to authentic concerns behind the utilization of food crops for fuel production. Aegle marmelos seed is one such feedstock with high oil content portraying a better entrant among other non-edible feedstock. In this study, optimization of oil extraction, and biodiesel production from Aegle marmelos seeds had been reported. Oil extraction performed with n-hexane was optimized by response surface methodology (RSM) and artificial neural network (ANN). The influence of five parameters on oil extraction, namely particle size, acid concentration, solvent-to-seed ratio, extraction time and temperature were investigated. A comparison of performance evaluation between RSM and ANN was executed. The lower value of coefficient of determination (R2=0.998<inline-graphic></inline-graphic>), root mean square error (RMSE=0.2784<inline-graphic></inline-graphic>), standard error of prediction (SEP=0.7068<inline-graphic></inline-graphic>) and absolute average deviation (AAD=0.3425<inline-graphic></inline-graphic>) for ANN compared to those of R2<inline-graphic></inline-graphic> (0.9769), RMSE (0.5349), SEP (1.3326) and AAD (1.1072) for RSM showed the prediction competence of the ANN was much better than RSM. Among the process parameters studied, solvent-to-seed ratio was the most influential variable on oil yield. The maximum oil yield of 45.99 wt% was obtained at optimum conditions, with an acid value of 18.92 mg KOH g-1<inline-graphic></inline-graphic>. Hence, a dual-stage acid-base transesterification was employed to produce biodiesel. It was followed by 1<inline-graphic></inline-graphic>H NMR spectroscopy study and fuel properties analysis. The highest conversion of 98% was ascertained using 1<inline-graphic></inline-graphic>H NMR spectroscopy, and the biodiesel fuel properties were found to comply with ASTM standards.

BIODIESEL fuels; ARTIFICIAL neural networks; PETROLEUM production

Arabian Journal for Science & Engineering (Springer Science & Business Media B.V. ), 2018, Vol 43, Issue 11, p6119

2193-567X

Academic Journal

10.1007/s13369-018-3272-5