Automated Performance Tracking for Heat Exchangers in HVAC

Abstract

Performance tracking, consisting of system state identification, fault detection, and propagation prediction, is critical to system operation optimization and maintenance scheduling. It is typical that dynamic systems degrade in a nonlinear manner, often times accompanied by abrupt fault occurrences, before and after which the degradation rate varies. Commonly used state estimation techniques for uncertainty evaluation, such as extended or unscented Kalman filter and particle filter (PF), are suited for nonlinear system state tracking without abrupt transient phenomena. To address this challenge, an integrated performance tracking technique is developed. It consists of a local search PF, which can dynamically adjust its resampling operation to avoid sample impoverishment as is commonly seen in conventional PFs, and a total variation filter, which can detect transient changes in the system states due to abrupt fault occurrence. The developed technique is evaluated for fault detection and performance tracking of a heat exchanger, which is a key component in the heating, ventilation, and air conditioning system. The performance of the developed technique is first evaluated using a finite element model of a heat exchanger, under both normal and faulty conditions. Subsequently, evaluation using experimental data provided by the American Society of Heating, Refrigeration, and AC Engineering was performed. The results have confirmed the accuracy and reliability of the developed technique in fault detection and performance degradation tracking involving complex fault propagation scenarios.