Comparative Analysis of Predictive and Adaptive Sliding-Mode Controllers for Temperature Stabilization of a Laboratory Bell-Type Furnace

Temperature control of industrial furnaces is challenging because of large thermal inertia, actuator constraints, and asymmetric heating and cooling dynamics. This paper compares four temperature control strategies for a laboratory bell-type furnace: an analytically tuned internal model control PID (IMC-PID) controller, dynamic matrix control (DMC), model algorithmic control (MAC), and an adaptive sliding-mode controller (ASMC). A control-oriented furnace model obtained through experimental system identification was used for controller design and comparative simulation at a sampling period of 10 s. Under nominal simulation conditions, MAC achieved the lowest tracking errors with a mean absolute error (MAE) of 5.70 ∘C and root mean squared error (RMSE) of 20.32 ∘C. DMC achieved similar performance (MAE 6.13 ∘C; RMSE 20.56 ∘C) while reducing control-signal total variation by approximately 29% relative to MAC. ASMC achieved an MAE of 8.96 ∘C and RMSE of 23.82 ∘C, improving nominal tracking relative to IMC-PID (MAE 14.68 ∘C; RMSE 31.04 ∘C). A ±20% process-gain perturbation study showed comparatively moderate changes in MAC and DMC performance. In contrast, ASMC exhibited greater sensitivity to gain mismatch, particularly through increased tracking error at reduced gain and increased control activity at elevated gain. All four controllers were subsequently implemented on an industrial B&R PLC using Structured Text and PWM-based binary heater actuation. Real-time furnace tests provided qualitative experimental verification of their practical executability on the physical thermal system. The simulation results demonstrate the tracking advantages of predictive control under the considered conditions. In contrast, the qualitative PLC verification confirms the real-time executability of all four controllers and the practical feasibility of their PWM-based implementation on the physical furnace.