Indoor unit fault detector for a multi-split VRF system in heating mode

• Fault detection algorithm is proposed for a multi-split variable refrigerant flow system.
• Raw data can be directly fed to the algorithm compared to conventional algorithm that needs data preprocessing.
• Air side indoor unit fouling is consistently detected with a state observer.
• Stuck expansion valve can be detected by monitoring uncontrolled refrigerant temperature.
• A new FDD idea using raw data can be derived from consideration of system dynamics and control.

A multi-split VRF system operates unsteadily most of time due to the constantly varying refrigerant flow rates of associated indoor units. VRF systems require a different approach from conventional techniques to detect faults, which have developed based on steady-state operations. In this paper, two fault detection techniques are proposed. Their advantage is that they do not require the test data to be preprocessed to obtain steady-state data. The first technique is applied to detect heat exchanger fouling by a state observer, and the other technique is used to detect valve sticking by temperature variance. These techniques were not chosen haphazardly but were derived from physical reasoning. Their validity was confirmed by test data. The methodology developed in this study can be applied similarly to other HVAC equipment that operates mostly in transient states.