Analysis of the effects of recycling on process control

Abstract

Abstract The union of different devices in order to obtain a specific response for a process is commonly called a control system. For a control system, it is necessary to have one or more controllers. Among the most used in the industrial sector are the PID and PI controllers. Next to these controllers is the control software. Scilab is a good example of control software. It is characterized as free code software, with no cost for its acquisition, in addition to having a large computational power and integrated tools, such as Xcos, intended for modeling and simulation. For the union with Scilab, there is Arduino. Such a mixture can be used, for example, to control liquid levels in tanks. In this context, the present work aims to study the tank-level control system based on PID and PI controllers through the union between Scilab and Arduino. Phenomenological models were developed based on closed-loop control (feedback control system) of the process with two tanks not coupled with recycle. Furthermore, for comparison purposes, two approaches were used for each process: one considering the saturation of the manipulated variable and the other without the presence of such saturation. At first, there was a need to implement an anti-windup system. For tuning the controller parameters, the ISE method was used, executed through a programming code developed in Scilab. The parameters found for the two systems were tested on a made-up experimental bench. Therefore, using the block diagrams and the method here called “ISE method”, satisfactory values were obtained for the control parameters. These were ratified in the tests carried out in the experimental module. Level control was achieved with greater prominence for the PI controller since there is one less parameter to be tuned and processed by the system. This controller provided results close to the PID controller for cycles up to 50%. In general, the PI controller showed maximum response deviations smaller than the PID, such as deviations of 1.55 cm and 2.40 cm, respectively, for the case with 75% recycle. It was also clear the influence of the saturation of the manipulated variable on the system response, but not on the tuning of the controller parameters.