Recently, the team at X, known for developing breakthrough technologies in self-driving cars, autonomous robots, and carbon-neutral fuels, announced that they are working on a moonshot for the electric grid. This is an exciting development that will likely bring greater attention to an issue we’ve been working on for awhile – specifically, the question of “whether creating a single virtualized view of the grid could make the grid easier to visualize, plan, build and operate with all kinds of clean energy.” 

For years, our team has been championing the need for real-time visibility of the distribution grid. The more data, visibility, and controls a utility has at their disposal, the more precisely they can operate the grid and the easier it is to integrate clean energy tech into grid operations. That’s why we’ve developed the only highly accurate, real-time, data-driven machine learning software that leverages data from the entire distribution grid to improve visibility, security, optimization and control. 

Our Patented Solution 

We recently patented a virtual power flow algorithm and optimal control methodology that processes real-time data from the entire distribution system, uses machine learning models to characterize grid physics, and then makes optimal operational decisions. This is exactly the kind of solution the X team has described. 

We poll a wide range of distribution grid data sources — including substation load tap changers, mid-line voltage regulators, capacitor banks, line voltage monitors and smart meters — as frequently as possible. In some instances, data is available every 5 seconds and we can provide true real-time visibility. For other devices, especially first-generation smart meters, data is only available once a day. Our machine learning model uses all of this available data. What sets it apart is its accuracy and the ability to directly optimize power flow. 

This groundbreaking approach builds upon our team’s decade of experience using real-time data for voltage optimization, where we’ve been able to deliver an industry leading 3% or greater energy savings and 50% increase in DER hosting capacity. 

Key Learnings for the Moonshot 

Through our development and deployment process, we’ve learned a few key things that we think could be incredibly valuable to the X moonshot effort: 

1. The grid needs real-time machine learning technology. Most existing utility operational systems are based on a physics-based model of the grid, i.e., one that uses the physical properties of the electrical conductors/connections. This is a valid approach for many use cases; however, given the dynamism of the grid edge — where power and voltage conditions change much more rapidly, particularly in the presence of DERs — a physics-based model alone will quickly become outdated and inaccurate. 
   
   Utilities’ reliance on static physics-based models which lack grid-edge visibility and control contributes to many grid operational problems today, including slow interconnection times for DERs, overly-conservative DER hosting capacity constraints, a low utilization of flexible demand, and slow storm response and recovery. As states work to achieve their clean energy goals, the complexity of the grid edge will grow exponentially, and these problems will become even more substantial, potentially including moratoriums on solar and electric vehicles on parts of the grid, drastic increases in costs, and growing reliability problems. These are the clean energy challenges that machine learning and computer science are well positioned to help us solve. 
   
   
2. We can’t rely on machine learning or computer science alone. As the X team suggests, machine learning, artificial intelligence, and advanced computing must be part of the solution. But in our team’s decade of experience operating the grid, we’ve learned that computer science approaches alone cannot solve these complex problems.
   
   Given that the electric grid is a massive and complex physical system, even the most advanced operational tools will need to have some grounding in the physics that govern the system. For example, we leverage our team’s electrical engineering expertise to adjust machine learning model structures and apply constraints to better reflect the true nature of the grid.
   
   Our team has a unique combination of expertise in computer science, data science, and power systems engineering that has enabled us to develop ideal methods that bridge the gap between traditional physics-based models and data-driven machine learning models.
   
   Additionally, many utilities simply aren’t equipped to operationalize the amount of distribution system data that is available to them. For example, smart meters are rich data sources that can greatly enhance the accuracy of a machine learning model. However, utilities’ ability to leverage this data is hindered by the limited computational power of existing AMI systems as well as the low bandwidth and high latency of existing communications networks. To transform grid operations we must build the right software solutions, but we must pair those with the right hardware and communications networks and create the right environment for those solutions to be deployed.
   
   
3. Developing the right technology is only the first step in solving the problem. For this moonshot to be successful, utilities need a business model and regulatory environment that incentivizes the adoption of new technology. There is tremendous value in a virtualized view of the grid and huge untapped potential in the data that already exists, especially AMI data. That’s why, in addition to developing new patented technologies, we’ve been working with regulators in several states to push utilities to use AMI data to drive meaningful operational outcomes, including real-time system visibility and DER integrations.
   
   Achieving a more modern, digitized grid isn’t just about getting the technology right — it’s also about creating the right commercial and regulatory environments for that technology to be deployed. We’ve gained solid traction, especially in a few states like Connecticut and New York, where stakeholders understand that grid investments must support a decarbonized future. 

The X team is right that no single organization or industry can solve this problem alone. We’re excited to see what their next steps are as we collectively work on breakthrough grid technologies that will make decarbonization a reality.