P2-2. Equipment Scheduling and Cycling > Outcomes

Technical Outcomes:

• Detection of devices with step changes in power consumption was more fully automated and a better understanding of the limitations of this method was achieved. Testing in the first phase showed that there may be limits to detectable equipment of/off events based on the magnitude of the change in power relative to the total power measured by the NILM. A number of turn-on and turn-off events for fans and a chiller at the Iowa Energy Center's test building could be detected with two centrally located power meters. As a general rule, for equipment to be monitored from the integrated power profile, it was found that the relative magnitude of the component's power consumption to the system's total power input should not be less than 5% of the total power. Detection limits should determine the number and location of NILM meters required to obtain useful information about a building and its equipment. It may also be difficult for the NILM to discriminate between substantially identical equipment on the same connected load circuit.
  
  
• Automation of load tracking algorithms for constant power devices, such as fans, pumps, and reciprocal chillers was successfully demonstrated, as well as detection of a number of faults that are difficult to detect with non-electrical measurements. Major improvements included a state estimator and automation of steady state and transient detection filters that can detect device events in real time. Detection of variable power, constant-speed loads was started but not complete by the end of the project.
  
• A frequency-analysis FDD technique was developed to track variable-speed drives. Developing this technique was a major accomplishment, since variable speed driven motors change power levels very slowly. The same technique was used to detect power oscillations associated with faults such as a chiller vacuum leak, a mis-aligned fan shaft, and under-damped feedback-control loops. These faults would be very difficult to detect with more traditional thermo-fluid fault-detection methods. Figure 2 shows how the frequency characteristics change over the startup of a VSD system and the relationship between frequency and power.
  
• The project was successful in extending the ability of the NILM technology to automatically detect scheduling and cycling events for constant power equipment. Major improvements included a state estimator and automation of steady state and transient detection filters that can detect device events in real time. The technology was successful in detecting and identifying devices that turn on or off if the device load is more than five percent of the total load at the time the device cycles.

      

          
Figure 2 Illustration of Evolution of Frequency Components over 6-min Motor Startup and Power vs. Frequency

Market Outcomes:

• NILM technology can provide end-use electricity consumption and demand data to building automation systems less expensively than traditional sub-metering. Building automation systems usually do not offer detailed electric load information because sub-metering at the equipment level is expensive. A NILM integrated into a utility meter, or included in a building automation system, could provide trend data to building operators, disaggregated by equipment type. By using statistical analysis of historical data, subtle changes in energy use patterns of individual pieces of equipment could be detected, allowing a more prompt investigation.
  
  
• NILM technology can enable real time pricing for small commercial buildings and allow automated load control when integrated into an energy management system. A barrier to using real time pricing is the lack of information about how electric energy is used at the end-use or individual appliance level. As a less expensive alternative to sub-metering, NILM technology could provide information to decide which loads to curtail.
  
  
• FDD based on NILM offers further insight into equipment operation, similar to vibration analysis, which can provide early clues about equipment malfunctions. A skilled and observant technician can discern details through the high-speed sampling data unobtainable any other way.

Back to Previous Page