Yul Yunazwin
The design of control system for natural gas processing in typical chemical plants is often done sequentially and in a distributed manner in which the overall plant is first decomposed into several control loops and then linear controller is implemented and tuned in each control loop. Such a design method often results in a suboptimal performance of the plant’s operation, not only due to the heuristic tuning methods that are used for each loop’s control parameters but also because it usually only considers the steady state behaviors of the plant’s dynamics. In the presence of disturbance propagation across different loops or hard nonlinear constraints on the process’ transient dynamics, this design method often results in severe instabilities of the plant’s performance This research proposal is aimed at addressing the limitations of the aforementioned design approach by proposing the use of Mixed Integer Nonlinear Programming (MINLP) framework to design optimal plant’s control structure and controller parameters. The proposed MINLP framework not only takes into account the hard nonlinear constraints on the process dynamics but also considers possible disturbances that may occur in the plant. This research will examine the effectiveness of the framework in a nonlinear model of interconnecting distillation column. The outcomes of this research will include (i) the development of MINLP framework for optimal design of control structure and controller parameters and (ii) a characterization of computational complexity and tractability of the framework when implemented in realistic models of natural gas processing plants. The findings from the proposed research will be examined using benchmark models of purification stage in natural gas processing plant. We envision that the results of this research will positively contribute to further development of advanced methods for simultaneous model estimation and control systems design in natural gas processing plant. We believe that this vision is of great importance in order to improve and optimize currently developed methods in throughput improvements of natural gas processing plants.