Desenvolvimento e avaliação do desempenho de um sistema de controle em cascata através de modelo empírico relacionando temperatura de topo e vazão de vapor em uma torre de stripper

Abstract

The chlor-alkali industry plays a significant role in society, with its production process involving the electrolysis reaction of saline solution (raw material) converted into three main products: gaseous chlorine, caustic soda, and gaseous hydrogen. Regarding the first component, various purification processes exist, with cooling being the initial step. In this stage, gaseous chlorine undergoes heat exchange in heat exchangers to cool down after the electrolytic reaction. During thermal exchange with the refrigerant liquid, water impurities in the gaseous chlorine are condensed, but portions of chlorine remain entrained in this liquid in a gaseous form. This solution cannot be discarded into the effluent system and must be treated in a chlorine desorption tower or stripper, utilizing saturated vapor, for instance. Maintaining process conditions within specified ranges is essential to ensure project operability and chemical process safety. Therefore, the objective was to propose a cascade control strategy for the column, focusing on the relationship, transfer function, between the column's top temperature (process variable) and vapor flow rate (manipulated variable). To determine the process transfer function, the system was initially modeled and simulated using the commercial Aspen Plus simulator. With the converged system, disturbances were applied to the vapor flow rate in open loop in Aspen Dynamics to observe and collect data on the stripper's top temperature. The data generated during the transient regime were used to identify the transfer function through the MATLAB system identification tool. Subsequently, a cascade control was proposed in the MATLAB Simulink environment, along with controller tuning using the trial-and-error method, Ziegler-Nichols, and Cohen-Coon. Regarding the results, the behavior of the top temperature exhibited dead time and a significant time to stabilize at a new steady-state value. The transfer function obtained for this method was a second-order model with poles in the complex axis, a gain of X, a time constant of Y, and a dead time. When tuning the flow controller, the best method for PID controller parameters was trial and error with IAE 114.1 and ITAE 10310. It was observed that for the flow controller, the Ziegler-Nichols and Cohen-Coon methodologies did not eliminate the steady- state error (offset), given the high proportional gain obtained by these methods from the reaction curve. For cascade-configured controllers, quick responses were achieved compared to a purely feedback loop, effectively handling disturbances in both the inner and outer loops, returning to the steady-state value set by the reference. Therefore, cascade control proved advantageous in quickly addressing disturbances compared to a simple PID control.