How distributed intelligence demonstrates value
More than ever, operating a modern grid with increasing penetration of variable decentralised generation and complex power flows requires visibility from the edge up to the control room. As the number of connected devices increases, from smart appliances and distributed generation in the home to sensors and gateways in substations and the network, distributed intelligence opens the way for more rapid awareness and insights.
This feature article was originally published in The Global Power & Energy Elites 2022
When central Florida utility Tampa Electric Company embarked on the upgrade of its 810 000 meters with smart meters, the opportunity presented itself to investigate the potential for distributed intelligence (DI) in the meters to provide customer and grid operation benefits.
Distributed intelligence applications
To explore this innovation and validate the decision to implement distributed intelligence applications, Tampa Electric, in partnership with Itron, opted for a leading analytics company to test the performance against back-office cloud analytics to determine which option would deliver the maximum value in terms of detecting conditions more effectively.
Tampa Electric expected that moving the analytics to the meter with access to one-second data and peer-to-peer communications would deliver greater accuracy in finding conditions and result in a higher yield and fewer inferences and wasted resources.
With faster decision-making based on more valuable information – assuming the value of data degrades with latency – a significant drop in the total cost of ownership could result through fewer data backhaul, storage, and analysis in the back-office.
Tampa Electric selected three applications (apps) for testing in the lab over one month: meter bypass theft detection; residential neutral fault detection; and high impedance detection.
Among the results, the meter bypass theft detection DI app identified all ten use cases; how whereas, while the back office analytics identified all the use cases as well, it also produced seven false positives.
The residential neutral fault detection DI app identified six use cases, but the back-office analytics identified zero use cases as the attributes required to identify broken neutrals are not present in the data available in the back-office.
Read the full project feature here
