Time series analysis of field data for soft faults detection and degradation assessment in residential air conditioning systems

Abstract

Residential Air Conditioning units are significant contributors to energy consumption. Soft faults in these units, such as refrigerant leakage and inadequate condenser airflow, can lead to reduced equipment life, decreased cooling capacity, and increased energy consumption. While extensive research has been conducted on Fault Detection and Diagnosis (FDD) in AC systems, most studies rely on laboratory-imposed faults or simulations, which may not reflect real-world conditions. Thus, long-term field data analyses remain scarce. This study develops and validates a time-series analysis-based methodology for detecting and diagnosing these faults in residential air conditioning units. Virtual refrigerant charge is used to detect refrigerant leakage, while the difference between condensing and ambient temperatures is used to detect inadequate condenser airflow. The methodology is tested on a dataset of 81 units across the US and Canada, monitored over a full cooling season (2–7 months). Results show that 2 units exhibited degraded condenser airflow and 5 had refrigerant leakage. Refrigerant leakage resulted in a monthly Coefficient of Performance (COP) reduction of 4–10% and an increase in daily energy consumption by 4–26% over a faulty period of 6.5 to 15 weeks. Similarly, units with degraded condenser airflow experienced a COP reduction of 4–7% per month, and daily electricity consumption increased by 15–17% over a faulty period of 8–8.5 weeks. This study quantifies fault performance degradation under residential conditions by analyzing real-world operational data, offering a field-tested approach for identifying and assessing soft faults. This work highlights the importance of timely fault detection and maintenance in residential Air Conditioning units to ensure efficiency, minimize energy waste, and prevent system damage.