Abo Bibliothek: Guest
Proceedings of the 24th National and 2nd International ISHMT-ASTFE Heat and Mass Transfer Conference (IHMTC-2017)

ISSN Online: 2688-7231

ISBN Online: 978-1-56700-478-6

OPTIMISATION OF ENERGY AND MATERIALS IN ACID ESTERFICATION OF RUBBER SEED OIL THROUGH RSM AND ANN

pages 1425-1431
DOI: 10.1615/IHMTC-2017.1970
Get accessGet access

ABSTRAKT

Biodiesel is a potential renewable fuel that is produced from either edible or non-edible oils which can substitute petroleum derived diesel fuel. In this study, rubber seed oil with high free fatty acid content (FFA: 47% or acid value: 94 mg KOH/g) was used for the production of biodiesel. For high FFA content oils, acid esterification pre-treatment process is essential to reduce it to the anticipated lower limit. Since acid value of the present oil was too high, two step acid esterification method was employed to reduce it to the expected value of less than 6 mg KOH/g. A four-factor-five-level Central Composite Design (CCD) was employed in Response Surface Method (RSM), which generated 21 experimental runs with different sets of parameters keeping acid value as response variable. Parameter optimisation have been carried out by considering minimisation of energy requirements for the chemical reaction. By applying RSM, significant models were developed for reduction of acid value to 15.26 mg KOH/g under optimised conditions in the first-step and around 5.5 mg KOH/g in the second-step. The independent variables in the RSM models were methanol/oil molar ratio (w/w), catalyst concentration (wt.%), reaction temperature (°C) and reaction duration (minute). ANOVA (Analysis of Variance) table was also created to investigate the individual and interactive effects of the variables on acid value reduction in both the steps. These experimental designs of RSM were used for training, testing and validation of Artificial Neural Network (ANN). Acid values were also predicted by ANN method for both steps. Coefficient of Determination (R2) and Average Absolute Deviations (AAD) were used to compare the predictive capability of both models. The results showed that ANN model is a better fitting model than RSM.

Digitales Portal Digitale Bibliothek eBooks Zeitschriften Referenzen und Berichte Forschungssammlungen Preise und Aborichtlinien Begell House Kontakt Language English 中文 Русский Português German French Spain