| Popis: |
In this study, response surface methodology (RSM) and artificial neural network (ANN) were used to optimize Terminalia catappa L. kernel oil (TCKO) yield. Solvent extraction method was used for the oil extraction, with n-hexane as the extracting solvent. The highest oil yield was obtained at 55 °C, 150 min, and 0.5 mm. The physicochemical properties of the TCKO were determined using standard methods. Gas chromatographic (GC) analysis and Fourier Transform Infrared (FTIR) were respectively, used to determine the fatty acid composition and prevalent functional groups in TCKO. At optimum conditions of temperature, particle size and extraction time, the RSM predicted oil yield was 62.92%, which was validated as 60.34%, whereas ANN predicted yield was 60.39%, which was validated as 60.40%. The results of the physicochemical characterization of TCKO showed that the dielectric strength (DS), viscosity, flash and pour points values were 30.61 KV, 20.29 mm2 s−1, 260 °C, and 3 °C, respectively. Physicochemical characterization and FTIR results of TCKO indicated its potential industrial application, especially as transformer fluid. Fatty acids compositions result indicated that the oil was highly unsaturated; while XRD results of Terminalia catappa L. kernel (TCK) samples obtained, both before and after extraction, showed difference in their peaks and corresponding intensities, due to the damage effect of solvent. Finally, the obtained optimization results indicated that ANN was a better and more effective tool than RSM, due to its higher R2 and lower RMS values. |
