Optimal Controller tuning of a PI controller for a three tank non-interacting process

Abstract

The term process indicates a setup or a plant that is to be controlled or regulated. A set up or a plant to be regulated is indicated by the term process. It may be a chemical plant unit such as a distillation column or a processing station or a pharmaceutical industry etc. In order to automate the process, it is important to monitor certain system output parameters such as tank level, steam pressure, temperature etc. The controlled technique employed may be sequential or continuous in nature and it depends on the application chosen. In this research, a standard three tank - non interacting process is considered. The design of a controller is highly intricate as it involves non linear variations in the parameters. The commonly adopted control techniques in process control are like feedback, feed forward, control of ratios etc. A traditional PI controller is used for making the level of the tank to stick on to the set point value. The transfer model is derived from the basic principles. This paper deals with finding out the best approximation method among Smith's approximation method and Krishnaswamy's approximation method for reducing the higher order transfer function into first order plus delay time model. Three tuning methodologies were adopted such as Skogestad tuning of PI controller, Cohen Coon method and Internal Model Control method. The time domain performance parameters such as settling time, ISE and IAE are measured. From the comparison made, Krishnaswamy's approximation method performs better with all the tuning methodologies.