Recovering Distribution Grid with Phaseless Measurements

Inspiration

Utilities face significant challenges in modeling their distribution grids, with outdated or incomplete models impacting grid reliability and planning. Advanced sensor solutions exist but are prohibitively expensive. Inspired by the opportunity to leverage existing infrastructure, we set out to create a cost-effective, scalable solution that enhances grid resilience and supports reliable power delivery.

What it does

Our project enables utilities to recover accurate grid models using only phaseless voltage and current measurements from existing infrastructure, as opposed to the current state-of-the-art. By leveraging advanced algorithms, it guarantees exact recovery of the grid structure in noiseless scenarios, providing utilities with a robust and scalable tool for improving operations and planning.

How we built it

We applied concepts from compressed sensing and sparse linear modeling to develop a specialized Gauss-Newton algorithm that ensures exact recovery of grid parameters. When faced with scalability challenges, we incorporated the Randomized Kaczmarz algorithm, which offered similar guarantees while being computationally efficient for grids with thousands of nodes. We tested these algorithms on IEEE 14-bus and IEEE 33-bus systems, achieving promising results.

Challenges we ran into

One major challenge was scaling our algorithm to handle larger grid systems. The Armijo descent condition in the Gauss-Newton algorithm slowed down computation for grids beyond 33 nodes. Identifying and implementing a more efficient algorithm while maintaining recovery guarantees was critical. We also navigated the complexities of working with sparse data and ensuring numerical stability during testing.

Accomplishments that we're proud of

We successfully developed an approach that enables utilities to use existing infrastructure for accurate grid modeling, potentially saving tens to hundreds of millions of dollars in sensor costs and enabling accurate models for future planning of the electric grid. Our integration of the Randomized Kaczmarz algorithm addressed scalability challenges, enabling efficient recovery of grid structures in larger systems. These achievements bring practical and impactful innovation to the energy sector.

What we learned

We learned the importance of balancing theoretical guarantees with practical implementation. Exploring compressed sensing and iterative methods deepened our understanding of sparse linear models and their real-world applications. Understanding the challenges the key stakeholders actually encounter to inform the development of our novel method allowed us to tackle real-world problems. Additionally, collaborating on a multi-disciplinary problem broadened our perspective on the intersection of power systems and high dimensional statistics.

What's next for Recovering Distribution Grid with Phaseless Measurements

The next steps for this project involve extending our testing to larger grid models, such as the IEEE 118-bus system, to validate the scalability and robustness of our approach. Additionally, we aim to formalize the theoretical guarantees provided by the tools of compressed sensing, solidifying the mathematical foundation of our methodology. Lastly, we plan to engage with key stakeholders, including utility companies, to adopt and test our solution on real-world distribution grids, ensuring its practical applicability and impact in enhancing grid resilience and reliability.

Built With

• julia