In building HVAC systems, chilled water flowmeter is an important sensor whose reading could be used to measure the real time cooling load, a critical variable for automated control of building HVAC systems. To maintain the reading data accurate, the fault detection and diagnosis (FDD) of flowmeters is necessary. Existing FD/FDD methods for chilled water flowmeters have several common shortcomings: (1) High requirements on sensor integrity: multiple sensors are usually involved to build energy balance models; (2) Complex methodology, the fault of any monitored sensor could trigger the detection hit, thus diagnosis procedure is unavoidable to isolate the faulty sensor; (3) the more sensors involved, the harder to collect fault-free historical data to build a fault-free benchmark. To tackle these existing problems, a user-friendly fault detection (FD) method for building chilled water flowmeters is proposed in this study. The proposed method requires three types of variables to function: pump frequency, pump power, and measured chilled water flowrate on the header pipe. The field data of a real HVAC system is used in the case study to validate the performance of the proposed method. Results of the validation case study suggest that the proposed method could reach high hit rates confronting different faults (bias, noise and drift) at different levels. Compared to existing FD methods, the simple workflow and low sensor requirements make the proposed method more feasible and user-friendly for engineering practice.
Keywords Fault detection and diagnosis, Flowmeter, Chilled water pump, Affinity law, Random forest