Skip to Main Content U.S. Department of Energy
Building Systems

Automated Fault Detection and Diagnostics Agent (AFDD Agent)

The automated fault detection and diagnostic (AFDD) process is a two-step process where a fault with equipment operation is detected and then the actual cause of the fault is isolated (see the schematic below). The process generally relies on analytical or physical redundancies to isolate faults during the diagnostic step. Most rooftop units (RTUs) in commercial buildings lack physical redundancy because heating ventilation and air condition (HVAC) systems in commercial buildings are considered non-critical. An AFDD process can use proactive diagnostic processes to create analytical redundancy to help isolate the cause of a fault. The proactive diagnostic process is similar to functional testing that is performed during manual commissioning of systems.

AFDD Diagram
AFDD Diagram

Proactive AFDD is a process that involves automatically initiating changes to cause or to simulate operating conditions that may not occur for some time, thus producing results that might not be available for months otherwise. Such tests could be automated to cover a more complete range of conditions or to deepen diagnosis beyond what might be possible without this capability. The proactive diagnostic process can help diagnose and isolate faulty operations to a much greater extent than passive diagnostics, but it is intrusive. Some building owners and operators may consider this to be disruptive to the normal operation of their RTU systems. They may not, however, if such proactive tests can be conducted quickly enough to maintain acceptable control of the RTU systems. Proactive diagnostic procedures could provide continuous persistence of performance if they are frequently triggered (e.g., once per day, once per week, once per month or perhaps seasonal only). These procedures might be scheduled to occur during building startup hours or at the end of day to further reduce their intrusiveness or could be scheduled on demand.

Seven automated AFDD algorithms were developed and deployed on the TN platform for detecting and diagnosing faults with RTU economizer and ventilation operations using sensors that are commonly installed for control purposes. The algorithms utilize rules derived from engineering principles of proper and improper RTU operations. The seven algorithms include:

  • Compare discharge-air temperature with mixed-air temperature for consistency (AFDD0)
  • Check if outdoor-air damper is modulating (AFDD1)
  • Detect sensor faults (outside-air, mixed-air and return-air temperature sensors) (AFDD2)
  • Detect if the RTU is not economizing when it should (AFDD3)
  • Detect if the RTU is economizing when it should not (AFDD4)
  • Detect if the RTU is using excess outdoor air (AFDD5)
  • Detect if the RTU is bringing in insufficient ventilation air (AFDD6).

The intent of these algorithms is to provide actionable information to building owners and operations staff. In today’s market environment, no one has time for nuisance interruptions, so the algorithms have been designed to minimize false alarms. On the other hand, if HVAC systems and their controls are starting to fail, having an indicator (a.k.a. "check engine light") of a real problem is always helpful – especially if it allows operations and maintenance staff to be proactive, rather than reactive.

AFDD agents are currently running on two demonstration sites in western Washington and northern California. See the current results.

Research Areas

Additional Information