A Real Time Self-Tuning Algorithm for PI Control of the Heating and Cooling Coils in Buildings

Abstract

Proportional and Integral (PI) controllers are used widely in HVAC applications. It is necessary to choose suitable values for PI gains for PI control of the heating and cooling coils in buildings. Consequently, suitable indoor comfort can be maintained; the energy efficiency of the heating and cooling systems will be maximized; and the lifetime of heating and cooling valves will be prolonged. A self-tuning algorithm is developed in this dissertation to find the good P and I gains. The proposed algorithm evaluates the controller performance by the integral square error (ISE) of the coil supply air temperature. The controller operation will be considered as not suitable when the ISE is too big. The P and I gains are required to change at this condition. The algorithm uses the •iteration' concept to find better values for P and I gains. It first changes the P and I gains to certain values to decrease the ISE. The iteration then will stop if the ISE is small enough; otherwise the change of the P and I gains will continue until optimal values are found. Further, the new algorithm determines the iteration direction by observing the spectrum distribution of the ISE. P and I gains will be increased if the low frequency content is the dominant part in the coil supply air temperature; P and I gains will be decreased if high frequency content is the dominant part.

The new algorithm can be realized on-line without interrupting the coil normal operation. It functions as a watchdog to be sure the PI controller maintains smooth and rapid tracking properties. The new algorithm only requires about forty lines of program code in the current Direct Digital Control (DOC) systems used for HVAC control. The basic concept of the algorithm will be developed by theoretically analyzing the dynamic behavior and frequency characteristics of the cooling and heating coils in buildings. The whole algorithm is then demonstrated by controlling a coil in a building, thus proving its feasibility.