Using Dynsimr to study the implementation of advanced control in a Propylene/Propane Splitter

Abstract

In the process industry, advanced control is usually implemented so that they ensure stability and constraints satisfaction. Moreover, a competitive global market and environmental regulations results in the necessity for the economic optimization of the process operation. Real Time Optimization (RTO), which is based on an economic criterion, is usually performed in an upper level of the control structure and sends optimizing targets to the lower dynamic control layer where the advanced control drives the system to optimum targets. In this structure, the RTO employs a complex stationary non-linear model of the process for the optimization and the advanced control is usually implemented through a MPC based on a linear model. In this paper, the application of such optimization structure to an industrial Propylene/Propane (PP) splitter is tested in a simulation platform based on the integration of the commercial dynamic simulator Dynsimr and the rigorous steady-state optimizer ROMeor with real-time facilities of Matlab. The advanced control is represented by an Infinite Horizon Model Predictive Controller (IHMPC), based on a space-state model in the incremental form that reproduces the step response model and considers the existence of zone control, optimizing targets for the inputs and can accommodate time delays. In this simulation platform the optimization and advanced control of a Propylene/Propane splitter of an oil refinery is studied. The simulation results show that proposed RTO/advanced control structure is stable and can be implemented in the real system.