(408e) Development of Start-up Control Concepts for Dividing-Wall Distillation Columns Based on Simulations and Experiments

The separation of multicomponent mixtures in fully thermally coupled distillation columns attracts the chemical engineering world. In that respect, the highest level of process integration is represented by dividing-wall columns. The aspect of getting high value-added products in merely one distillation column with simultaneous savings in both, energy and capital costs, attracts the investors of distillation plants. Furthermore, a growing demand for high-purity products in fine and commodity chemistry within the last years increased the interest of operating dividing-wall columns.

The aim of this contribution is the development of robust and time-optimal process control strategies for the start-up of dividing-wall columns. The strategies allow for significantly shortening the start-up time and for easy application at lab and production-scale columns.

In earlier studies a rigorous process model has been developed, which allows for a detailed analysis of the startup process from ambient conditions. The model considers several features, which are characteristic for a dividing-wall column, such as the self adjusting vapour split and the heat transfer across the dividing wall. An extensive model validation with experimental data of our pilot plant dividing-wall distillation column points out, that the model is capable of predicting the entire startup process. Further comparison with data of a production scale distillation column confirms the significance of the model. This comprehensive validation is indispensable, as a model-based design of startup strategies demands for a reliable process model.

First examinations have revealed, that specific changes in the manipulated variables could significantly reduce the startup time in comparison with conventional strategies. A systematic approach with extensive dynamic optimization studies shows the minimum time requirements and serve as a benchmark for developing the control approach. The various number of manipulated variables and the inherently required inequality constraints for a distillation process result in an elaborate formulation of the dynamic optimization problem. This general formulation of the optimization problem results in an approach which is applicable for a great variety of fully thermally coupled distillation arrangements. Sophisticated time-optimal trajectories of manipulated variables, which are obtained by off-line optimization, are usually not applicable at real plants. Hence, a robust process control strategy is developed, which tracks the time-optimal solution by using a decentralized control scheme and results in a reasonably short startup time.

The presentation starts with a short introduction of the process model and the setup of the pilot plant. Furthermore, simulation results are compared with experiments, whereas the profiles of characteristic process variables are analysed and discussed. The extensive validation addresses the steady state, the dynamic and the startup characteristics of the dividing-wall column. Finally, the results of the dynamic optimization problem and a robust startup strategy are presented, which illustrate the potential for both, a significant reduction in startup time and an easy implementation of the control approach for tracking the time optimal trajectory.