Evolutionary algorithm-based model predictive control for a reactive distillation column in biodiesel production

Biodiesel is touted to be an alternative to the fossil fuels as it is conducive to the environment. This investigation proposes a control framework to produce biodiesel in a reactive distillation column via a transesterification process. To extract quality product, the temperature profile must be maintained along the column as per the requirements. However, constant interactions among the products inside the column disturb the temperature profile and consequently the product quality. Therefore, this investigation treats the process as a single input and single output system, where in the process interactions are modelled as disturbances. A model predictive controller (MPC) is designed for the proposed system to ensure product quality. The MPC parameters must be selected appropriately to ensure optimal performance. In this regard, to tune the MPC parameters optimally, we use two evolutionary algorithms namely, the real coded genetic algorithm (RGA) and the bio-geography based optimization algorithm (BBO). The results indicate the proposed control strategy provides offset free set point tracking when compared to the multivariable control strategy employed using the MPC algorithm. Among the two evolutionary controllers used for tuning the MPC parameters, the RGA MPC controller provides a satisfactory performance.