P2-2. Equipment Scheduling and Cycling > Conclusions

• The NILM technology can make a significant contribution to improving control of energy use in building systems. The information generated by application of NILM technology will be less expensive than that created using traditional power submetering and acoustic/vibration monitoring. Electric power use information will be more widely available, allowing enhanced control and diagnostics features to be incorporated into energy management systems.
  
  
• More work will be required to fully automate the technology, as described in the Recommendations section below.

• The lead researchers investigated several alternatives for developing NILM based products or services. They had discussions with a major utility meter manufacturer as well as several energy management service providers. Although there is interest in the potential to provide additional information to end-users at a lower cost than power sub-metering, there are issues regarding identification of multiple units of devices that are of the same make and model within a facility or on a branch circuit. Additional field-testing will be required to address the accuracy of device identification and power estimation.
  
  
• Testing of the NILM to determine its commercial value has not been started. This work requires that NILM load-tracking output be compared with submeters and that a potential commercializer determine whether NILM output, either load tracking or power signatures associated with faults, has value to its customers.

• Development efforts should be geared toward a product for the small and medium sized commercial building market. Most small to medium size commercial buildings do not have a building automation system with energy management capability. Automating energy management is needed because the building owners and managers do not have the time or expertise to manually control energy use. The NILM technology could be an important contribution to reduce monitoring costs and to provide key electrical load information that is currently missing from most building automation systems.
  
  
• Several important additional research steps need to be taken.
  
  • Ability to track variable-power, constant-speed loads. This last remaining load class includes most chillers and those fans still using inlet vanes. Relations between real and reactive power for these loads should be explored.
    
  • Integration of the disaggregator for constant-power loads and the VSD tracker. The code was designed to be integrated. However, most constant-power loads are for pumps and tower fans associated with operation of a chiller. Until the chiller can be detected or is assigned remaining HVAC power, the integration has reduced practical value.
    
  • Automated training of the NILM to the extent possible. The state estimator is a good start, because it makes a best guess of the loads in operation at the time the NILM is turned on. However, the state estimator and the load disaggregator rely on knowledge of the characteristics of individual loads. To date, this information has been gathered manually.
    
  • Automated detection of faults. As a starting point, detection of faults associated with oscillatory power signals should be automated, based on frequency analysis. Automatic fault detection will aid those potential users who have limited time or ability to analyze NILM data.