Skip to main content

Modeling Power Grid Dynamics using Quantum Computing

Computational tools that enable both fast and accurate description of the grid state and its of evolution, and provide predictive capabilities, are needed to operate at near-optimal efficiency and foresee/avoid possible contingency scenarios. However, achieving this modeling capability for realistic, high-resolution grid dynamic models is still far from achieving real-time performances, even with modern computer architectures, while the ability to continuously explore a variety of near-future scenarios (faster than real time) is even more challenging.

Quantum Processing Units (QPUs) are rapidly progressing towards the demonstration of a paradigm-change in the acceleration of numerical calculations in physics and engineering with the perspective to become a new reference for the post-exascale computational era. ORNL has an established a leading role in developing capabilities and applications of quantum-enhanced computational methods.

Along these lines, this study intends to investigate a novel Quantum Computing application, where a QPU is the computational engine for large-scale electric power grid modeling to enable real-time, large interconnect-sized dynamic modeling. The potential payoff is to reach the ability to simulate in real-time, hybrid (electromagnetic and electromechanical) models of complex grid layouts, with complete characterizations of the state of the system at any instant in time, and predictive analytics capabilities.

Technical Approach.
This research will investigate the use of quantum computing to implement an electric power grid model that is mapped to a variant of quantum-enhanced optimization algorithm (already tested and available). Development and testing will be performed using available, state-of-the-art QPUs that are accessible at ORNL.

As a first step, the research plan considers the development of a weighted graph representation for a reference grid interconnect bus (such as IEEE bus test systems) in a suitable format for the QC hardware. Then, a few constrained optimization problems will be defined with choices focused on power dispatch, transient analysis, predictive analytics and grid planning for resilience enhancements. These examples, chosen for the critical relevance, will be developed along with the of the algorithm to take advantage of the inherent features of QC.

Finally, comparative cases will be also run on conventional hardware (cluster/supercomputer) to investigate present limitations, issues related to scalability to larger problems, and to infer needs/directions required for further developments in QC technology (both h/w and s/w). For the purpose of the proposed research, the size of QC hardware (e.g. terms of number of qubits) presently accessible at ORNL is adequate, since the primary focus is to look at algorithm development for reduced size, but scalable, applications.