Supply Air Temperature Control in an Industrial HVAC Pilot Plant Based on Smith Predictor and Inward Approach

This paper focuses on the supply air temperature control in a complex heating ventilating and air conditioning system. In industrial applications it is a common practice to use standard PI-controllers for temperature control. Typically the controller parameters are either found empirically or by employing standard tuning rules. The present work proposes a systematic approach to tackle the design problem. It is based on the so-called inward approach in combination with a Smith predictor. Given performance criteria are taken into account with the presented method. The plant model used for controller synthesis is a second order system with dead time. Its parameters were identified by experiments. The performance of the suggested (non-PI) strategy is compared to a conventional PI-controller which is tuned via the well known T-sum-rule. Finally, experimental results of a real world pilot plant are presented.