(671f) Experimental, Anfis Modeling and De Optimization of Pilot Plant Scale Biodiesel Production From Yellow Grease in a Reactive Distillation

Yellow grease was used in this study to produce fatty acid methyl esters (biodiesel) in a continuous reactive distillation column catalyzed by a heteropolyacid, H₃PW₁₂O₄₀ .6H₂O. Biodiesel produced in a batch reactor has some disadvantages such as high alcohol demand, low catalyst neutralization, and high production cost, but reactive distillation (RD) alleviates these problems. This paper considers the application of ANFIS as a model to predict biodiesel production in an RD column. The selected experimental data, obtained from a pilot plant RD column, were trained by a hybrid learning combination of forward and backward pass algorithm. The model agreed with the experimental data, which indicated that ANFIS was an accurate, rapid, and easy way of modeling in such a complex case. Differential evolution (DE) was employed for estimation of maximum yield at different operating conditions. The optimum conditions were determined to be 115.52 (mol/h) total feed flow, 29.32(°C) feed temperature, 1.28 kW reboiler duty, 66.4 methanol/oil ratio and 7.36% (w/w) catalyst concentration to achieve 94.62% yield.