Supercomputing to tackle energy challenges

Problems caused by wildfires in California serve as a stark reminder of the fact that power outages are not a phenomenon of the past. And in a future energy system with a high proportion of renewable, fluctuating energy sources like wind and solar, critical loads will appear more frequently. It is therefore good news that a US collaboration has demonstrated how supercomputing can help the energy network adapt quickly to changes in supply and demand.

Since the energy system consists of millions of devices situated at large distances, it is highly challenging to secure coordinated operation. Even small delays in communication – or latencies in the technical terminology – may soon add up to become impossible to manage. In a collaboration with participation from ESnet (the US Department of Energy’s Energy Sciences Network) and partners, supercomputing was applied to achieve fast signal processing and thereby minimize latencies. As a result, it was possible for researchers to simulate an energy system where operators hundreds of kilometers apart may work together just like they were sitting side by side.

“As communities, utilities, and other stakeholders work together to transition to a clean energy future, the research capabilities demonstrated will reduce the risk in this transition and ensure that our future energy system is affordable, reliable, resilient, sustainable, and secure,” says Kevin Lynn, director of grid modernization, Office of Energy Efficiency and Renewable Energy, Department of Energy (DOE).

A tightly coupled experiment

In the project, the National Renewable Energy Laboratory (NREL) and the Pacific Northwest National Laboratory (PNNL) were connected to provide a live demonstration of a Cordova, Alaska, microgrid modeling experiment. The demonstration showed that advanced control systems in the Cordova microgrid could allow it to maintain power to critical resources like the hospital and the airport during an extreme weather event and loss of some of its hydropower resources. During this tightly coupled experiment, NREL simulated the Cordova microgrid while PNNL simulated the advanced control systems.

The successful experiment followed in the footsteps of another demonstration called the Global RealTime SuperLab involving eight laboratories across two continents connected over the internet. This was in 2017. The project uncovered the negative impact of latency. Not only was latency in that experiment sometimes large, but it was also unpredictable. That variability made it difficult for researchers to conduct a meaningful experiment.

To combat the latency issues, ESnet was invited to take part in the new experiment. ESnet, an international network dedicated to support scientific research, has made critical contributions not only for the national laboratories but also internationally, including supporting the Large Hadron Collider, the world’s largest and most powerful particle accelerator.

Significant reduction in latency variance

ESnet provided a reliable, low-latency connection so that research at the two laboratories could exchange frequent command and control information. The roundtrip latency between NREL and PNNL was 24 milliseconds, with a variance of 0.2 milliseconds. In comparison, the average latency in the 2017 experiment was about 27 milliseconds with a variance of about 11.5 milliseconds. The very low latency variance was vital to the research success and made exchange of command-and-control information between the two laboratories nearly deterministic.

Following the successful demonstration, the involved researchers and laboratory partners are already working towards the next step. The team is looking to connect six national laboratories for experiments that use and model millions of interconnected devices. These experiments would be called “SuperLab 2.0”.

This would allow researchers to run simulations like the Cordova microgrid, but at a scale that reflects the complexity of a regional or national energy system. An advanced research network could identify solutions for city- og region-level problems such as outages caused by extreme weather events or cyberattacks.

The text is inspired by the news release “NREL and Pacific Northwest National Laboratory Demonstrate Multi-Lab Capabilities in Live Demonstration” by Alison Reidmohr at www.nrel.gov and by the article “ESnet Plays Key Role in Groundbreaking ARIES “Digital Twin” Demonstration” at www.es.net