Prediction of Optimum Process Parameters for Karanja Biodiesel Production using Support Vector Machine, Genetic Algorithm and Particle Swarm Optimization

Abstract

The growing energy demand and depletion of the conventional energy resources presented a need for alternative reliable source of energy that can readily replace the conventional fuels like diesel and petrol. In the current work, biodiesel is synthesized from Karanja oil by using transesterification. The yield is obtained at varying KOH concentrations (1 wt %, 1.5 wt %, 2 wt %), varying molar ratios of methanol:oil (3:1, 4.5:1, 6:1) and varying times (15 min, 30 min, 45 min, 60 min). The optimal conditions from experiment are obtained as temperature of 50° C, reaction time of 45 minutes, methanol-oil ratio of 4.5:1 and catalyst concentration of 1.5 %. The viscosity of biodiesel is found to be between 0.036 - 0.038 stokes. Optimum conditions obtained were compared with the statistics available in literature. The produced biodiesel from Karanja oil conform to the ASTM D6751 standards. The produced biodiesel is characterized using Fourier Transform Infra Red (FTIR) Analysis and Gas Chromatography Mass Spectrometry (GC-MS). Further Artificial Intelligence techniques namely Support Vector Machine, Genetic Algorithm and Particle Swarm Optimization have been used for predicting the optimum conditions of the biodiesel production. The predicted yield with Support Vector Machine is compared with yield obtained from experiments. The SVM accurately predicted the experimental results with the R2 = 0.999. PSO and GA can effectively be used as a tool for predicting the optimum parameters for biodiesel production.