Utility companies and regulators should insist that new smart meters be able to monitor voltage and current with at least 60 samples per second frequency and (ideally) 12 bit precision; any less, and valuable energy management features could be crippled. Similarly, direct consumer access to the meter output is an absolute necessity. I know of no smart meters on the market today with this level of monitoring precision; if you do, please leave a comment telling us about it.
High-precision electricity load data can be analyzed using a technique known as non-intrusive load monitoring (NILM). By comparing patterns in aggregate (whole-house or whole-office) load to known appliance profiles, energy analysis software can provide detailed appliance-level energy consumption data without requiring appliance-level electronics. This approach is far more affordable than instrumenting each appliance and could provide an economical way to track consumer responses to price signals, allowing advanced energy management solutions to be rolled out to consumers who aren't willing to pay for more expensive home automation. At the very least, NILM can give a power consumer an extremely detailed real-time view of his electrical loads, allowing him to intelligently target efficiency improvements and behavioral changes (possibly driven by dynamic or TOU pricing).
This is from a recent paper by Carnegie Mellon researchers*:
I'd like to see NIST and EPRI get in touch with these NILM researchers as part of the smart grid standards process. If we don't get high-precision capabilities baked into new meters while it can be done with little, if any, incremental cost, we will be tearing the meters out again in a few years to replace them with ones capable of high-precision NILM. While meter manufacturers may like the idea of built-in obsolescence and rapid product turnover of $300-500 smart meters, rate payers may criticize this waste of their dollars. And from an environmental perspective, we may find that the carbon footprint of the rapidly obsolescent metering and home automation hardware overshadows any energy savings it manages to facilitate. That would be an ironic tragedy, one that ratepayers will not easily stomach.
----
* "Training Load Monitoring Algorithms on Highly Sub-Metered Home Electricity Consumption Data" by Mario Berges, Ethan Goldman, H. Scott Matthews, and Lucio Soibelman published in TSINGHUA SCIENCE AND TECHNOLOGY (ISSN 1007-0214 65/67 pp. 406-411, Volume 13, Number S1, October 2008)
High-precision electricity load data can be analyzed using a technique known as non-intrusive load monitoring (NILM). By comparing patterns in aggregate (whole-house or whole-office) load to known appliance profiles, energy analysis software can provide detailed appliance-level energy consumption data without requiring appliance-level electronics. This approach is far more affordable than instrumenting each appliance and could provide an economical way to track consumer responses to price signals, allowing advanced energy management solutions to be rolled out to consumers who aren't willing to pay for more expensive home automation. At the very least, NILM can give a power consumer an extremely detailed real-time view of his electrical loads, allowing him to intelligently target efficiency improvements and behavioral changes (possibly driven by dynamic or TOU pricing).
This is from a recent paper by Carnegie Mellon researchers*:
While NILM applications require minimal hardware and some instances claim over 90% recognition of some loads, this approach is not without challenges. The requisite hardware must be able to report power readings with at least 1.0 Hz of frequency[13] and ideally calculates at least true power, reactive power, and harmonics. Associating a particular electrical signature with the originating appliance either involves a training period or a large database of known loads. Still, given the continuing decreases in hardware costs and the possibility of distributing software costs and signature categorization, the high quality of data and low labor costs for installation make NILM the most promising technology for detailed end-use electricity consumption data.
...
[13] Cole A, Albicki A. Algorithm for non-intrusive identification of residential appliances. In: Proceedings of the 1998 IEEE International Symposium on Circuits and Systems. Monterey, CA, USA, 1998, 3: 338-341.
I'd like to see NIST and EPRI get in touch with these NILM researchers as part of the smart grid standards process. If we don't get high-precision capabilities baked into new meters while it can be done with little, if any, incremental cost, we will be tearing the meters out again in a few years to replace them with ones capable of high-precision NILM. While meter manufacturers may like the idea of built-in obsolescence and rapid product turnover of $300-500 smart meters, rate payers may criticize this waste of their dollars. And from an environmental perspective, we may find that the carbon footprint of the rapidly obsolescent metering and home automation hardware overshadows any energy savings it manages to facilitate. That would be an ironic tragedy, one that ratepayers will not easily stomach.
----
* "Training Load Monitoring Algorithms on Highly Sub-Metered Home Electricity Consumption Data" by Mario Berges, Ethan Goldman, H. Scott Matthews, and Lucio Soibelman published in TSINGHUA SCIENCE AND TECHNOLOGY (ISSN 1007-0214 65/67 pp. 406-411, Volume 13, Number S1, October 2008)
Leave a comment