Quantum Hardware-in-the-Loop for Optimal Power Flow in Renewable-Integrated Power Systems

TL;DR

This work develops a Quantum Hardware-in-the-Loop framework that integrates RTDS with quantum and quantum-inspired hardware (D-Wave Advantage and Fujitsu Digital Annealer) to perform adiabatic quantum PF (AQPF) and OPF (AQOPF) for renewable-integrated power systems. By mapping PF/OPF to Hamiltonians and QUBO formulations with penalty terms, the approach enables real-time optimization in a proof-of-concept on the IEEE 9-bus system with solar and wind resources, showing close agreement with classical Newton-Raphson benchmarks. The study demonstrates that both QA and QIIO can yield accurate PF/OPF results, with QA often delivering better net power accuracy and QIIO offering substantially faster compilation and iteration in certain cases, while highlighting the ongoing challenge of scalability to larger grids. Overall, the results indicate a viable path toward quantum-enhanced real-time grid optimization, particularly as hardware advances improve qubit counts and connectivity for large-scale OPF problems.

Abstract

This paper presents a proof-of-concept for integrating quantum hardware with real-time digital simulator (RTDS) to model and control modern power systems, including renewable energy resources. Power flow (PF) analysis and optimal power flow (OPF) studies are conducted using RTDS coupled with Fujitsu's CMOS Digital Annealer and D-Wave's Advantage quantum processors. The adiabatic quantum power flow (AQPF) and adiabatic quantum optimal power flow (AQOPF) algorithms are used to perform PF and OPF, respectively, on quantum and quantum-inspired hardware. The experiments are performed on the IEEE 9-bus test system and a modified version that includes solar and wind farms. The findings demonstrate that the AQPF and AQOPF algorithms can accurately perform PF and OPF, yielding results that closely match those of classical Newton-Raphson (NR) solvers while also exhibiting robust convergence. Furthermore, the integration of renewable energy sources (RES) within the AQOPF framework proves effective in maintaining system stability and performance, even under variable generation conditions. These findings highlight the potential of quantum computing to significantly enhance the modeling and control of future power grids, particularly in systems with high renewable energy penetration.

Figures (6)

• Figure 1: Computational and information flow within the QHIL framework for OPF calculations. The OPF formulation receives system parameters, real-time data (e.g., $P_i^D$, $Q_i^D$, $P_i^{PVF}$, $P_i^{WF}$), and operational constraints. The AQOPF algorithm solves the OPF using a problem Hamiltonian on quantum/digital annealers, and the results are compared to the classical NR method using Pandapower solver. The computed generator set points ($V_{i,ref}^G$, $P_{i,ref}^G$) are then employed in the real-time power system simulation.
• Figure 2: Hardware configuration of the proposed QHIL framework for real-time power system simulations using RTDS® and RSCAD. A GTNETx2 card with SKT protocol enables TCP communication to exchange system parameters and RES generation data with the OPF algorithm running on the quantum hardware or local device, which then returns generator and converter set points to the simulator in real time.
• Figure 3: IEEE 9-bus test system with integrated RES. The system includes nine buses, three generators at buses 1, 2, and 3 ($PV$ buses), and loads at buses 5, 7, and 9 ($PQ$ buses). A 1 MW solar farm and a 2.5 MW wind farm are connected to bus 7 which are scaled to 20 MW each. Both RES units are connected using step-up transformers and $RL$ branches, maintaining a constant short circuit ratio. Voltage controllers at the RES buses allow them to function as $PV$ buses in PF and OPF formulations.
• Figure 4: PF results for the IEEE 9-bus system using the AQPF algorithm on QA and QIIO, compared to the classical NR method. The top row shows the computed values for (a) bus voltage magnitude ($V_i$), (b) phase angle ($\delta_i$), (c) active power ($P_i$), and (d) reactive power ($Q_i$). The bottom row illustrates the absolute errors between the quantum solvers (QA and QIIO) and the NR benchmark for each corresponding parameter.
• Figure 5: OPF results for the IEEE 9-bus system using the AQOPF algorithm on QIIO, compared to the classical NR method. The top row shows the computed values for (a) bus voltage magnitude ($V_i$), (b) phase angle ($\delta_i$), (c) active power ($P_i$), and (d) reactive power ($Q_i$). The bottom row illustrates the absolute errors between the quantum solver (QIIO) and the NR benchmark for each corresponding parameter.