Nonlinear MPC based on multi-model for distillation columns

A multi-model structure for model predictive control has been established for a petroleum distillation column through first principle analysis, and its parameters have been estimated from data generated by a rigorous model. The nonlinear multi-model has been used for state-estimation and prediction in a MPC scheme. The controller has been applied to quality control of a FCCU fractionator. Realistic simulations show that multi-model MPC (MMPC) is an attractive option for nonlinear MPC since the scheme implies a straightforward extension of linear MPC, and the results are promising.