REDEFINING SUCCESS FOR A DISTRIBUTED ENERGY GRID: THE THREE TENETS
When it comes to effectively leveraging distributed energy resources (DERs), there are three critical success factors that any DER management system or Virtual Power Plant (VPP) must embody. In a previous blog we focused on Tenet #1: the importance of speed.
In today’s blog, we address another of the top three criteria: accuracy. Just as the question “how fast is fast enough” was answered with “it depends,” so too does the question “how accurate is accurate enough” have the same response. The criticality of accuracy depends on what the distributed energy resources are being dispatched to do.
If the goal is simply to shed load or mitigate demand peaks, a demand response program focused on curtailment doesn’t need pinpoint accuracy. But if your goal is a grid service like renewable firming or regulation service, accuracy levels of 99.9 percent could be required.
Achieving this level of accuracy mandates a precise and detailed understanding of the behavior of each participating asset. In other words, when using distributed energy resources to provide services that support the immediate and critical stability of the grid, you can’t merely control bulk energy behavior. You need to control precise power behavior, which, in turn, requires an in-depth understanding the assets being controlled.
Based on our experience with many of the projects we’re working on around the world, we have observed that today’s distributed energy resource management systems (DERMS) and virtual power plants (VPPs) require speed of response, coupled with system accuracy levels approaching 99.9%. And this must be provided 24x7, 365 days a year. Furthermore, today’s DERMS and VPPs should allow for continually updated modeling of distributed assets, leveraging continuously streaming sub-metered sensor data to model DER assets that exist in the physical world and to precisely represent the real-time status and working conditions of the DER assets under control.
Why? Because the optimization nodes in these systems use real-time feedback from each DER to calculate optimal real and reactive power set points assets within the network, and it’s this network which acts as a single, dispatchable resource that is continuously optimizing the dispatch of the DER portfolio. As the number of DERs grows to hundreds of thousands – or even millions – the cumulative impact of even the slightest inaccuracies can be dramatic.
It’s worth noting that modern machine learning capabilities have been a tremendous boon to our ability to model, measure and understand distributed asset portfolios. Without it, considerable time, effort and money must be dedicated to configuring and experimenting with each device – an approach that is error-prone and which fails to consider that the behavior of assets is continually changing.
With machine learning algorithms, the DERMS or VPP can predict peaks, flexibility and behavior based on what is continually being learned about each asset type. A particular HVAC unit, for example, can be observed under different climatic conditions and times of day to measure load flexibility models and infer behavior under diverse situations. The algorithm continually learns more and more about each unit to determine what load flexibility exists and, at the same time, becomes increasingly more accurate. It’s a self-learning, self-training model that’s purely data driven with virtually no human intervention needed to train new distributed energy devices or to continually align the load calculation models based on changes in asset behavior over time.
It’s the high degree of accuracy enabled by machine learning that also enables us to leverage more and more, smaller and smaller distributed devices because it’s much easier to model them automatically with no costly human intervention needed. If we can keep the costs of asset utilization down, we can leverage progressively smaller asset loads to move power to the right places at the right time.
There is no question that as the deployment of DERs continues to expand at a breakneck speed, accuracy is a major concern as utilities strive for successful management of a diverse and distributed energy mix.
If this topic interests you, I invite you to read our white paper about the new normal in leveraging DERs for optimal grid balance.
Stay tuned for our blog on the third of the Three Tenets for DER Success: Scalability.