Control of a reactive distillation column for biodiesel production

The control of the bottom product temperature in a reactive distillation column, which is used for the production of biodiesel via calcium oxide (CaO)-catalyzed transesterification of edible sunflower oil, is investigated in this study using the Model Predictive Control (MPC) method. The laboratory-scale packed column, which had a height of 1.5 meters, a diameter of 5 mm, and a reboiler volume of 4 L, was divided into stripping and reaction zones. Raschig rings were filled into the 80 cm long reaction zone and the 40 cm long stripping zone. The reaction zone was also filled with CaO catalyst. The solution in the reboiler, located at the bottom of the column, was boiled using a 700 W heating mantle. Four thermocouples were used to measure the real-time temperatures of the boiler, rectification, reaction, and stripping sections in the column. The process was first simulated using Aspen HYSYS v12.1 software before the process control studies were started. The operating parameters investigated through simulation studies of the system were optimized with Design Expert 13.0. The optimization results of the steady-state showed that the required edible sunflower oil feed flow rate for a maximum conversion of 99.49%, with a feed molar ratio of 6.03, and a reboiler temperature of 80.3 ℃. In the experimental study conducted using these parameters, a %97.2 conversion was achieved for a 7-hour steady state. Finally, the control of the system operated at optimal operating parameters was experimentally performed using decoupled and undecoupled MPC algorithms. The reboiler duty was selected as the process manipulated variable, and the reboiler temperature was determined as the controlled variable. The developed decoupled and undecoupled control algorithms and models were created using Matlab and Simulink software. The control algorithms were tested with various set-point trajectories, and their control performances were evaluated. The calculated Integral Absolute Error (IAE) and Integral Square Error (ISE) values indicated that the DMPC algorithm had a better control performance."