Identification and elimination of hunting behavior in HVAC systems

Undesired oscillations (hunting) in process variables such as fan speeds, and valve and damper openings. in heating, ventilation and air-conditioning (HVAC) systems results not only in suboptimal performance, but also component wear and increased energy costs. This paper first presents a simple algorithm to detect these undesired oscillations in building HVAC systems. The paper then provides the results of a survey done to gauge the magnitude of the problem of hunting in building HVAC systems by implementing the algorithm on data obtained from air-handling units (AHUs) of 10 different buildings on the Texas A&M University campus. Through the survey, it was found that 70% of the chilled-water (CHW) valves hunt for 6% to 78% of the time, and 22% of supply fans exhibited hunting for 6% to 26% of the time. The results of the survey indicate that hunting in building HVAC systems may be a significant problem. To illustrate the cause of this hunting behavior, the paper also presents a simulation model of an AHU that shows that the system gains vary considerably with changing operating conditions. Pi-type control is used to govern most of the operations in building HVAC systems which are designed for a particular operating condition. Due to the nonlinear nature of the HVAC systems, changing operating conditions lead to changing system gains and poor performance of the Pi-type controllers. A cascaded control loop architecture is proposed in this paper as a solution to the problem of hunting, and its effectiveness is demonstrated through a simulation example.