Dynamic Optimisation of Batch Distillation with a Middle Vessel using Neural Network Techniques

A rigorous model (validated against experimental pilot plant data) of a Middle Vessel Batch Distillation Column (MVC) is used to generate a set of data, which is then used to develop a neural network (NN) based model of the MVC column. A very good match between the “plant” data and the data generated by the NN based model is eventually achieved. A dynamic optimisation problem incorporating the NN based model is then formulated to maximise the total amount of specified products while optimising the reflux and reboil ratios. The problem is solved using an efficient algorithm at the expense of few CPU seconds.