AVT - Research - Research project

As shown in practice, distillation control design is successful when chemical engineering know-how of column behaviour is included in the control concept as a priori knowledge. This project aims at incorporation of such knowledge in the model predictive control framework developed in our group. Several control strategies will be implemented and complex distillation model structures will be created to test the quality of control.

There are manifold examples for distillation sequences in industry and academic world. As an example for modeling and control of distillation sequences, we consider a sequence for separation of Acetone-Chloroform-Benzene-Toluene mixture (Krämer 2009). This is a non-ideal mixture and the process contains coupled and less coupled submodels. Thus, both modeling and control of the ACBT process are challenging. Modeling of physical properties was already done, so having incorporated the knowledge available at our department we will focus on plant-wide control of the process.

The setup of a plant-wide control is a difficult task by its nature and this process is not an exception. Many operating objectives must be satisfied simultaneously and some of them contradict each other. Moreover, even control of an individual column requires a more detailed insight into operating conditions. Not only restrictions having their origins in overall plant operating objectives but also specific objectives must be fulfilled.

Summing up, in the project, we aim at creating the process, observer and control models, incorporating the models in our control framework and study different plant-wide control strategies using the example of the ACBT process.