The technology world recently reached a tipping point. In 2014, for the first time, consumers more often used ‘apps’ on their mobile and tablet devices than mobile browsers or PCs to access the internet and get things done. In other words, apps running at the edge of the network, utilizing the intelligence and computing power of a lower-cost and unified mobile computing platform, are dominating our increasingly connected world.

While this may not seem like much more than an interesting technology and marketing statistic, there’s important insight to be gained here as we plan and build the next iteration of the smart grid. Just as distributed applications and edge intelligence offer specific value to consumers, these capabilities will play a key role in the interconnected power grid of tomorrow. The move to leverage distributed intelligence to improve the reliability and efficiency of the grid is well underway as the Internet of Things (IoT) comes into sharper focus.

In fact, distributed intelligence opens up many new possibilities in smart grid use cases by greatly extending the capabilities and business value a smart grid  communications network can deliver. In fact investing a technology platform that supports initially even just a small number of distributed applications related to grid reliability, efficiency and safety will increase financial benefits to the utility by up to 50% compared to a current generation smart metering network. While ancillary and consumer benefits unleashed by distributed analytics have the potential to double that benefit stream. This definitely changes the game.

These benefits are made possible by enabling high-speed data processing and analysis at every level of the network, by making much more efficient use of network bandwidth and capacity, and by enabling a diverse ecosystem of grid devices to communicate with each other and interoperate on a localized level. Not all problems are bestsolved at the edge. But the key to enabling these capabilities is deploying intelligence and analytic capability in the right place.

The applications of the application of distributed intelligence to smart grid use cases creates optimized solutions to some specific utility business problems and challenges that, until now, were neither practical nor affordable to solve. We already see tremendous potential value in several core applications that significantly improve the return on investment for smart metering technology. Here are four examples:

Real-time diversion detection

Theft of electricity has a material financial impact on utilities and their customers throughout the world. While worldwide electricity theft is estimated to be in the range of 8% of revenue, in some regions, non-technical loss resulting from diversion represents 20 to 30% of revenue. That’s a huge number, but it also represents a significant opportunity to improve financial performance of the utility.

Even with current generation smart metering technology, detecting energy theft can still be an inefficient and laborious exercise of analyzing historical data from disparate systems and drawing inferences about where diversion may be taking place. By applying distributed intelligence to the problem, diversion detection is now based on realtime, continuous, and localized analysis of changes in electricity current flows and voltage levels in the distribution network, rather than sifting through circumstantial data delivered after the fact. This approach increases the accuracy of energy diversion and theft detection by as much as 300% over current smart metering systems.

It is a basic fact of the physics of electricity that when current is drawn through a conductor, voltage drops in a measureable way on the network. Through their ability to communicate directly with other meters at different levels of the network, and to know exactly where they are on the distribution system, meters monitor voltage levels on the network continuously, ubiquitously and precisely. They use that information to identify when current is drawn on the secondary of a transformer that did not go through a meter (i.e. theft) – and without requiring dedicated metering at the distribution transformer.

Detection of unsafe grid conditions

High impedance connections (HIC) on the low-voltage distribution system represent a safety risk, while also causing customer voltage problems and utility energy losses. A high impedance connection is simply a poor electrical connection that can be created when splicing, tapping or connecting wires, when foliage touches a line, or when a cable or connection fails.

When current is drawn through the high impedance connection, heating occurs and there is a voltage drop across the connection. As heating continues, the connection is further degraded, and this causes the HIC to worsen over time. Symptoms start as voltage problems, and can deteriorate to power outages or fires. Unfortunately, until now, there has been no practical way to identify and resolve these issues until they lead to severe voltage problems or failure/fire.

Distributed intelligence changes the game in HIC detection, and provides a practical and cost-effective solution for utilities to identify these losses, voltage anomalies and potential safety issues before they become a safety hazard or a costly liability. This enables meters to continuously monitor impedance and to notify the utility of the presence and location of high impedance connections.

In the event of a sudden change in impedance, as caused by failing connections or cables, this solution can send a priority message over the network to the utility informing them of the event, the relevant data and the location of the suspected fault so that field service resources can be dispatched quickly and precisely to correct the problem.

Outage detection and analysis

Like energy theft detection, the current state of outage detection and analysis via the smart metering network is still an inferential exercise based on how many affected meters can successfully transmit ‘last gasp’ outage messages over the network, how many of those reach the utility, and the filtering and analysis continues from there. This process is still hampered by the lack of an accurate and continually updated connectivity model that associates meters and distribution system assets. Peer to peer communications, combined with data analysis at the edge of the network, delivers a better approach and can reduce outage analysis time by up to 50% compared to current AMI systems.

By combining locational awareness on the grid with peer-to-peer communications, the network systematically and continuously evaluates the status of nearby meters and devices to quickly model and localize outage events, and report reliable and actionable information back to the utility in near real time.

When an outage event takes place, regardless of the scale or location, the system automatically initiates a ‘heuristic’ analysis among smart meters and devices on the network. The meters that still have power quickly go into a progressive outage analysis mode via peer-to-peer communications with other devices – in effect, asking their ‘buddies,’ do you have power? By initially analyzing the data at the edge of the network, many of the difficulties current smart metering networks encounter in outage detection–including largescale volumes of unfiltered outage data congesting the network – are eliminated.

Instead, the utility receives accurate and actionable information – including scale and location of the outage, affected meters, and affected transformers, and more – in a compressed timeframe. The outage is detected (usually before the first customer calls), and then modeled so that the extent and probable cause of the outage is rapidly understood and the appropriate resources can be dispatched efficiently to exactly where they are needed to begin restoration efforts.

Transformer load management

Overloading of distribution transformers is an increasingly common problem. Putting intelligence at the edge allows the load on individual distribution transformers to be analyzed continuously and managed locally in real time. Distributed analytics determine which meters are on the same transformer, and are able to detect changes and maintain correct relationships as restoration and maintenance work and the addition of new meters occurs over time.

Once meters determine that they are on the same transformer, they can communicate with each other locally and calculate the total load on the transformer in either direction. At the same time, a back office component of the analytic application interfaces with utility asset management and GIS systems to identify the transformer and its rated capacity, and delivers that information to each meter it serves.

Based on the total load and the capacity of the transformer, the meters can identify locally when the transformer is approaching overload conditions, whether from the line side or customer side. When this occurs, a distributed analytic running on the meters determines whether to shut off controlled loads behind the transformer, turn on or increase local distributed generation behind the transformer, turn on loads behind the transformer, or decrease or shut off distributed generation behind the transformer – depending upon which direction power is flowing through the transformer – to automatically reduce demand below allowed levels. In this way, safe loads are automatically and continually maintained on each transformer simply by the smart meters working together locally.

This represents a key area of opportunity for utilities to leverage the investment in distributed intelligence and AMI to significantly improve reliability, while optimizing capital and operational expenditures associated with transformer procurement and maintenance. Best of all, these results can be achieved with low incremental investment beyond the smart metering infrastructure.

There is absolutely no doubt that the convergence of information technology and operational technology in the global utility industry will continue and accelerate in an IoT world. The convergence of smart grid with the emerging smart cities and Internet of Things markets is accelerating this trend toward a more distributed model. The result will be both a stronger business case for smart metering and new, highly innovative solutions to longstanding grid operations challenges. MI