P2-1. Rooftop Air Conditioning > Background

Rooftop air conditioners are used extensively throughout small commercial and institutional buildings, but compared to larger systems, they tend to be poorly maintained. Application of automated fault detection and diagnosis (FDD), which has been used widely in critical systems, will significantly reduce energy use & peak electrical demand, down time and maintenance costs.

There are three important barriers to automating FDD in packaged HVAC units. FDD for HVAC systems, especially for rooftop air conditioners, is subject to economic constraints, which bring special difficulties and issues not encountered in critical systems. First, since a rooftop AC is itself relatively inexpensive, the cost to realize FDD for HVAC systems must be low. Different faults may have similar symptoms, and a variety of sensors can be helpful in identification, but some useful measurements such as flow rate, pressure or even humidity are simply too expensive. Limited available measurements must be used to extract as much information as possible. Computation requirements must be within the capabilities of a limited microprocessor-based system.

Second, since rooftop units are used in diverse weather conditions and climates, the behavior of the HVAC plant will vary drastically from site to site. In addition, since single-point sensor placement is generally used, measurements tend to be biased and noisy. The FDD must be able to cope with these difficult circumstances. This requires FDD for HVAC systems to have analytical redundancy, meaning the information from system measurements should be preprocessed extensively before it is used to detect and diagnose faults.

Third, unlike critical systems in which no fault can be tolerated, rooftop HVAC requires analysis of the economic impact of the fault: is it important enough to justify service? This requires a fault evaluation and decision step to be added to the software. Finally, unlike a critical FDD system that is engineered for a specific large system, FDD for packaged HVAC systems must be adaptive and generic enough to function on the same type of system, or at least on similar models from the same product family in order to reduce the per-unit costs.

In previous research, Purdue University developed a method that correctly detected and diagnosed single faults before there was about a 5% reduction in cooling capacity and efficiency. During evaluations of the method, faults were introduced in a single unit in the Purdue laboratory at various levels, and the sensitivity of the technique in diagnosing each fault was determined. Five faults were chosen for development and testing of an FDD method. Analyzing the service records from a service company that focuses on small commercial equipment identified the important faults.

• refrigerant leakage,
• condenser fouling,
• evaporator filter fouling,
• a liquid-line restriction, and
• compressor valve leakage.

Back to previous page.