Grid Modernization and Edge Computing
Grid Modernization envisions a system where grid control logic can be distributed from the back office to substations to the edge of the electrical network. This distributed control logic can be enabled by the use of edge devices. These edge devices include switches, controllers, DER, meters, etc. Edge computing extends standard operation of the edge device.
Edge computing is the ability to run arbitrary programs on devices at the edge of the electrical network. These programs can be upgraded over time and used to extend device functionality. With intelligent devices throughout the electrical network, it becomes important to have the ability to run programs on those devices, reducing latency, minimizing network bandwidth needed, and enabling autonomous behavior. Edge computing can be an extended feature in an edge device (e.g. switch, meter, FAN radio) or a dedicated edge compute node whose purpose is to perform analytics and coordinate action by other edge devices.
Edge computing capabilities can be characterized into three classes:
- Basic: Ability to execute operational functionality on the device (e.g. protocol translation).
- Intermediate: Ability to perform complex analytics on high frequency data and either return analytic output and/or perform operations based on output.
- Advanced: Ability to perform complex operations/analytics and coordinate results/actions with local devices in a peer-to-peer fashion
The basic class provides a platform where base functionality can be extended on edge devices. A common application is protocol translation or data filtering performed on the edge device.
Intermediate class edge computing allows for more complex operations/analytics to be performed on edge devices, using much higher frequency data available locally on the device. In many instances, there is a considerable amount of data on edge devices that remains underutilized because of the limited bandwidth of narrowband edge networks. The transmission of large amounts of data over these networks simply overwhelms them. By providing more complex analytics at the edge, devices can analyze this data themselves and return only meaningful data or perform some action, based on the local information.
The advanced class extends edge computing with peer-to-peer communications. Allowing edge devices to share data with neighbors provides for both coordinated analysis and action within the distribution grid. More importantly, it can enable advanced use cases such as autonomous balancing of ad hoc microgrids, dynamic voltage management, or active monitoring of hidden loads by remote intelligent switches.
Although performing protocol translation on devices at the edge of a distribution network is a common occurrence, local analytics and peer-to-peer coordination are only now becoming available in solutions. As this type of functionality becomes more pervasive, it will be increasingly important to understand the state of the distributed infrastructure, i.e. firmware version, configuration, applications running at edge, local communication group, etc. In order to maintain this complex distributed system a device management solution must be implemented. Although there are several management solutions currently within the industry, none have intrinsic capabilities to operate across narrow-band networks, like a Field Area Network (FAN) or inherently operate using UDP. These topics will be covered later when Communication Networks and Device Management are covered in this blog series.