Smith Predictor-based Control of Systems with Time Delay

Smith Predictor is an effective method for control of processes in which the process
time delay is long as it improves the performance of a PI controller. However,
processes are susceptible to delay variation due to uncertainties and this predictive
model does not guarantee plant robustness and stability due to the increase of negative
phase lag. A temperature control experiment is conducted in a shell-tube Heat
Exchanger Process Pilot Plant due to its time delay presence characteristic of interest.
The process model in the form of transfer function is identified using Statistical
Modelling with accuracy of 93.86%. A Smith Predictor-PI controller is developed
using Matlab/Simulink to simulate the control of the temperature loop and results
using performance indicators such as Integral Absolute Error (IAE) and Integral
Square Error (ISE) show that it performs better than a PI controller.