Real-time model based process monitoring of enzymatic biodiesel production.

ABSTRACT

In this contribution we extend our modelling work on the enzymatic production of biodiesel where we demonstrate the application of a Continuous-Discrete Extended Kalman Filter (a state estimator). The state estimator is used to correct for mismatch between the process data and the process model for Fed-batch production of biodiesel. For the three process runs investigated, using a single tuning parameter, qx = 2 × 10(-2) which represents the uncertainty in the process model, it was possible over the entire course of the reaction to reduce the overall mean and standard deviation of the error between the model and the process data for all of the five measured components (triglycerides, diglycerides, monoglycerides, fatty acid methyl esters, and free fatty acid). The most significant reduction for the three process runs, were for the monoglyceride and free fatty acid concentration. For those components, there was over a ten-fold decrease in the overall mean error for the state estimator prediction compared with the predictions from the pure model simulations. It is also shown that the state estimator can be used as a tool for detection of outliers in the measurement data. For the enzymatic biodiesel process, given the infrequent and sometimes uncertain measurements obtained we see the use of the Continuous-Discrete Extended Kalman Filter as a viable tool for real time process monitoring.