The Case for Autonomous Network Management

There is a near-global drive to replace carbon-based, polluting, resource-constrained energy sources with non-carbon renewables as the primary fuel for the electricity industry. Even traditional sources of bulk power production that do not emit carbon, such as nuclear and large-scale hydroelectric plants, are not the answer, as nuclear plants are increasingly costly to build, and most of the best locations for large-scale hydro projects are already in operation. The final nail in the coffin of carbon and nuclear is the fact that the levelized cost of energy (LCOE, or cost of energy produced over the lifespan of the asset) of intermittent renewable resources is now competitive with traditional generation, without subsidies.

The replacement of gas-powered cars with electric vehicles now appears unstoppable, as more vehicle manufacturers announce electric models. There is pressure on electrical utilities to adopt distributed energy resources (DERs) and adapt to this changing world.

On the legislative front many countries and states have set target dates and percentage goals (up to 100%) for a carbon-free grid. There are multiple state and national mandates, climate accords, and individual cities hoping to become “smart,” resulting in the accelerating uptake of renewables, and in the early retirement of nuclear and coal facilities.

Then there is the consumer. Who wouldn’t want to power their home and vehicles with free, non-polluting energy from the sun on their rooftop, if it was economic to do so? It is fast becoming both possible and economically feasible. After a century of near-stasis in the industry, these global changes in the way electricity is generated and consumed are accelerating, and this requires a whole new control systems approach: enter autonomous network management or ANM.

The Need for ANM

When looking at the implementation of ANM, we must consider: How can we use the existing distribution infrastructure to interconnect DERs without incurring huge new costs to upgrade the grid? That’s Problem 1. Second, how is it possible to operate and maintain the stability and reliability of a new grid like this, if we do connect all these intermittent and uncontrollable DERs out at the grid edge? That’s Problem 2.

1. Autonomous network management (ANM) systems are designed to enable more distributed energy resources, such as wind power, to be connected to existing utility infrastructure. Courtesy: Smarter Grid Solutions

Problem 1. The development of ANM systems came about precisely as an attempt to help mitigate this problem. The cost of upgrading the grid to allow a new wind (Figure 1) or solar farm to be connected can destroy the economics of the project, making it financially infeasible. A key feature of ANM systems is that they are designed to enable far more DERs to be connected to existing utility infrastructure than traditional utility models would allow, and this was the original use case for the technology.

The reason is that traditional utility planning models take an incredibly conservative approach to the addition of new generation sources on distribution feeders. They consider the worst-case scenario for potential issues like power back feed and voltage instabilities that could result in annoying things such as “flicker” on customers’ power supply. These situations might typically be encountered on only a few days every few years, but nevertheless the project developer has to pay for the system upgrade to ensure this doesn’t happen.

2. The dashboard of an ANM system provides a real-time look at the performance of various distributed generation components across the grid. Courtesy: Smarter Grid Solutions

An ANM system (Figure 2) manages the interconnection point by looking for any voltage or thermal issues and backing down the generation just enough to mitigate the problem, or by sending the energy into a local battery or hot water system instead of the grid. ANM operates within the limits of the protection equipment, so that the wind or solar farm can keep producing, but at a reduced level. As soon as conditions return to normal, the project is released back to full capacity.

The deployment of ANM allows project developers to connect to existing infrastructure, but at a price—they get what is called a “managed interconnect,” which means the solar or wind generation may be curtailed when the situations described are encountered. This is different from the usual approach of just generating as much as possible whenever the sun is shining or the wind is blowing. But, given that curtailment may be rare, if at all, and the cost of the ANM interconnect is far less than the wires upgrade and can be done in a few months rather than waiting possibly years for the grid upgrade, this very often allows a project to proceed when it otherwise could not have.

In fact, studies at the National Renewable Energy Laboratory in Colorado have shown that up to 300% more DERs may be connected using ANM than utilities would otherwise have allowed. Interestingly, the deployment of ANM to manage constrained grid capacity then opens up a whole host of other benefits.

Problem 2. Energy management, and supervisory control and data acquisition (SCADA) systems use a pre-defined, centralized model of the whole power system (what is connected to what), and then every few seconds collect whatever telemetered measurements are available—voltage and current readings, switch settings, breaker positions, etc., from various points on the grid. From the original model and these readings, they then try to create a picture of what the grid looks like at a given moment in time. By using a forecast of the pending demand and running various power flow analysis tools on the grid model, they decide if anything needs to be done, such as an increase or decrease in generation.

This works well for bulk, high-inertia, centralized power production at a high voltage. There is no good “original” distribution system model. There is often little visibility from the limited telemetry that exists, and things often happen much too quickly for an operator to intervene.

The way to solve this is to place some of the centralized control system intelligence out to the point of interconnection of the DER, enabling it to swiftly and autonomously act when needed. In the new architecture, smart ANM software is deployed at the distribution substation or right at the point of interconnection of the DER (note that it may already be there to solve Problem 1, but the two uses can be independent).

This approach does not replace the traditional control approach—it supplements it. The ANM system works along with the utility’s centralized systems to handle the intermittency of visibility and sheer number of DERs through fast and autonomous action at each location.

This enables local optimization of each new DER connected to the distribution system without requiring a complete and accurate model of the entire distribution network. This “local maxima” optimization system works well at the distribution level, because the impact of changes on a distribution feeder does not affect the remote system in the same way as changes at the transmission level. The ANM system uses whatever it can see and act on in a limited location, both to keep the system operating within required limits and also to optimize the use of the connected assets.

ANM systems are improving and developing to solve many more problems. The use of artificial intelligence (AI) is here: DER connectivity, data collection, and advanced analytics provide the opportunity to deliver AI-based solutions for predictive analytics for generation/load forecasting. AI is a key aspect of future developments.

There have been attempts to solve the DER interconnection problem with hardware solutions, including intelligent transformers and similar embedded-in-the-power-flow devices. For almost all industries, pure software approaches have always proven more capable, flexible, and open to rapid advances and development than hybrid hardware-software digitalization approaches. Software always wins in the end.

Market Opportunities and Limitations

The implementation of an ANM system benefits both renewable project developers and utilities. Another feature of this approach is that utilities can deploy ANM systems wherever they are needed most without requiring a wholesale change-out of the distribution management system and without the need for a solved network model.

However, there are two major impediments to wide deployment. The first is that utility regulators are not keeping up with this technology. The utility revenue model still relies on providing a return to investors based on the development of additional infrastructure, in direct opposition to the capabilities of ANM. This has to be fixed, but changing these ancient methods, as we all know, takes a long time at a regulatory level. The second is that utility staff take a very conservative approach to the deployment of new technology (with good reason, as job one is to keep the lights on).

It is difficult to envision a world where the network model is not king. The good news is that change is starting now, as a new generation of young engineers takes control of our electricity supply, a generation highly concerned with the challenges of climate change and highly adaptable to implementing the latest technology to help solve it. ■

—Pete Maltbaek is general manager North America at Smarter Grid Solutions.