Jump to Main Content
Disturbance-Based Alternate Feedback Control Scheme To Enhance Economic Performance of Batch Processes
- Lu, Peng-Cheng, Chen, Junghui, Xie, Lei
- Industrial & engineering chemistry process design and development 2019 v.58 no.10 pp. 4143-4153
- batch systems, economic performance, financial economics, process design
- In the past few decades, several robust designs dealing with disturbances in batch processes have been developed. The previous methods hold the viewpoint that all disturbances are harmful because they may drive the trajectory off the nominal optimal trajectory and lead to economic loss and safety issues. Are disturbances definitely detrimental? Actually, from the economic viewpoint, not all the disturbances are unfavorable to operating batch processes. To utilize disturbances fully for the improvement of the economic performance, disturbances are first divided into favorable and unfavorable types in this paper. An alternate feedback control (AFC) method is proposed to switch the feedback gains according to the types of disturbances, holding favorable disturbances and rejecting unfavorable disturbances. As a result, AFC will lead to an asymmetrical distribution of the trajectory. It is quite different from conventional feedback control (CFC), which rejects both favorable and unfavorable disturbances and leads to a symmetrical distribution. Second, because disturbances are the source of the variation and state errors are the combination of the previous state errors, control inputs, and disturbances, a disturbance-based feedback scheme is developed for AFC in this paper. Unlike traditional error-based feedback whose feedback gain acts on the state errors, the feedback gains of AFC act on the disturbances directly. Finally, a design combining robust economic optimization and AFC regulation is proposed for economic performance improvement and quick implementation of the regulation. The parameters of the online AFC are given by the robust economic optimization. This design can avoid the mismatch between the optimization layer and the regulating layer. Additionally, there will not be any computational burden for the novel design since the difficult and complex problem can be done offline and then AFC is implemented online using the gains generated offline. The optimality and the potential applications of the proposed integrated design are demonstrated through a typical dynamic batch reactor process.