Biodiesel production from waste oils by enzymatic transesterification: process optimization with hybrid neural model

In this paper an experimental and theoretical study of the reaction of enzymatic trans-esterification of glycerides with ethanol have been reported The enzyme was Lipase from Mucor Miehei, immobilized on ionic exchange resin, aimed at achieving high catalytic specific surface and recovering, regenerating and reusing the biocatalyst. As glycerides low quality and waste vegetable oil have been used in order to reduce pollution problems related to their treatment. A mathematical hybrid model based on the Artificial Neural Networks (ANNs) has been formulated in order to determine the mutual relationships existing between the inputs (feed composition and operating conditions) and the outputs (biodiesel composition and yield). Artificial Neural Networks, infact, are a very interesting alternative to model biotechnological processes that are inherently non-linear and rather difficult to be modeled in a fundamental way. Nevertheless, a “pure” neural model does not make use of any equation that could help to determine, on the basis of fundamental principles, the mutual relationships existing between the inputs and the outputs. Hybrid model predictions are given as a combination of both a theoretical and a “pure” neural network approach, together concurring to the obtainment of system responses. By this way it has been possible to describe some well-assessed phenomena by means of a fundamental theoretical approach, leaving the analysis of others, very difficult to interpret and describe in a traditional way, to rather simple “cause-effect” models, based on ANN. The effect of enzyme/substrate and glyceride/alcohol feed mass ratio, enzyme reuse, mixing rate and bioreactor configurations have been evaluated in order to find the optimal process performances.