IonQ and Oak Ridge National Laboratory Develop Hybrid Quantum-Classical Algorithm to Optimize Power Grid Operations

IonQ and Oak Ridge Researchers Utilize Quantum-Classical Algorithm to Address Power Grid Challenges

A new study by IonQ and the Oak Ridge National Laboratory has showcased how a hybrid quantum-classical algorithm can effectively tackle one of the power industry’s most challenging issues: determining which power generators to activate and how much electricity each should produce. This research, published on arXiv, introduces a method that combines quantum computing with classical optimization to solve the complex “unit commitment” (UC) problem, which is critical for daily power grid operations.

The hybrid approach employs a quantum variational algorithm to identify promising configurations of power generators and classical optimization techniques to refine power output, resulting in efficient and near-optimal solutions across large-scale simulations and on IonQ’s quantum hardware. By sampling a limited set of candidates instead of evaluating every possibility, this method presents a potential solution to other large decision-making challenges, such as airline crew scheduling and drug discovery.

### Understanding the UC Problem

At its essence, the UC problem revolves around cost management, which can significantly impact consumers. Power grid operators must decide, on an hourly basis, which generators to turn on and how much power to allocate to each to meet demand. Although this seems straightforward, the number of potential combinations increases exponentially with the number of generators and time intervals. For instance, with 15 generators over a 24-hour period, there are over 10^108 possible configurations. Moreover, real-world constraints complicate the situation, as generators have specific operational limits, turning them on and off incurs costs, and sudden fluctuations in power demand can disrupt plans. Traditionally, solving this problem optimally requires extensive computational resources and time, making quantum computers an ideal tool for finding solutions.

### The Proposed Solution: A Three-Step Hybrid Approach

To tackle the UC problem, the researchers developed a three-step hybrid solution:

1. **Quantum Scanning**: A quantum algorithm scans for promising combinations of generator states, effectively filtering the vast decision space to identify feasible candidates.

2. **Classical Optimization**: For each identified candidate configuration, classical optimization techniques are employed to fine-tune the power output of each generator.

3. **Selection of Optimal Solution**: Finally, the best solution is selected from the refined set.

The quantum component of the algorithm uses a variational quantum approach, where a quantum circuit is iteratively adjusted to achieve the lowest energy state of a mathematical expression, specifically a cost function representing generator operation. The researchers transformed the problem into a QUBO (Quadratic Unconstrained Binary Optimization) format, allowing it to be framed as finding the lowest-energy configuration of a network of binary variables, which aligns well with quantum computing capabilities.

### Performance Results

The researchers tested their method using both simulated quantum circuits and actual runs on IonQ’s Forte quantum processor, which supports up to 36 algorithmic qubits. The algorithm was applied to power grids with 3, 10, and 26 generators across 24 hourly time steps. Results from simulations showed that the hybrid algorithm produced solutions that were within 0.55% to 2.7% of the optimal cost, depending on the problem’s scale and the quantum circuit’s depth. For smaller systems, the algorithm achieved exact solutions, while for the largest configurations, the average error remained around 2.5%. The results on IonQ’s real quantum hardware closely mirrored the simulated outcomes, demonstrating the effectiveness of the hardware even amidst noise.

### Exploring Further Enhancements

The study also examined how the complexity of the quantum circuit influenced performance. Increasing the number of circuit layers improved accuracy, but required additional time for training and execution. The researchers noted that the number of optimization steps grew roughly linearly with the number of circuit parameters, highlighting the importance of smart circuit design to balance expressiveness and computational cost.

Future iterations of the algorithm may incorporate more advanced quantum techniques, such as variational quantum imaginary time evolution (varQITE), which could enhance performance on constrained optimization tasks like unit commitment. Custom circuit designs tailored to the specific characteristics of power grid problems could further improve the algorithm’s scalability to real-world applications involving hundreds of generators and various operational constraints.

This research is part of a broader initiative to leverage quantum computing for energy system optimization, which has significant implications for cost efficiency, grid stability, and carbon emissions. By demonstrating that quantum computers can play a meaningful role in solving practical power planning challenges—even in a hybrid framework—this study strengthens the case for early quantum advantages in industrial optimization beyond areas like cryptography and chemistry.

The study was conducted by a collaborative team from IonQ, including Willie Aboumrad, Martin Roetteler, and Evgeny Epifanovsky, along with Phani R. V. Marthi and Suman Debnath from Oak Ridge National Laboratory. For those interested in a deeper understanding of this work, the detailed research paper is accessible on arXiv.