Experimental Benchmarking of PID Empirical and Heuristic Tuning for Networked Control of Double-tank System

Controlling tank levels is one of the essential processes in different and several industrial applications. In this paper, a family of PID controllers are designed with different configurations to control a tank level of a large lab-scale Double-Tank System (DTS) for benchmarking purposes. These two cascaded tanks represent a Double Integrating Plus Dead Time (DIPDT) system. Different empirical methods are used for tuning PID controllers such as Ziegler-Nichols (Z-N), modified Ziegler-Nichols: Chien, Hrones and Reswick (C-H-R), Cohen-Coon (C-C) and minimum error criteria methods that include Integral of Absolute Error (IAE), Integral of Squared Error (ISE) and Integral Time Absolute Error (ITAE). Moreover, the Genetic Algorithm (GA) is used in tuning the PID parameters by finding their optimal values using different performance indices. Simulation results of the system investigate the performance of the P, PI and PID controllers of each technique. Finally, experimental validation is performed in real-time to the proposed DTS with the node.js platform.