Application of quantum computing to temporal aggregation for efficient capacity expansion in power systems

Capacity expansion planning (CEP) requires full-year, hourly time series to capture variability in load and renewable generation, but this temporal resolution makes optimization problems computationally challenging. a quantum–classical temporal aggregation framework has been proposed that formulates representative-day selection as a Quadratic Unconstrained Binary Optimization (QUBO) problem and solves it using the Quantum Approximate Optimization Algorithm (QAOA). To keep each QUBO within practical simulator or near-term hardware limits (about 15 qubits), the year is partitioned into 24 half-month windows; each window is reduced to k representative days via QAOA-based clustering and assigned non-uniform weights proportional to the sizes of their clusters. The weighted days are then integrated into a CEP model built on PyPSA (Python for Power System Analysis) with linearized optimal power flow. Evaluations on IEEE 9, 30, and 118 bus systems show that small sets of QAOA-selected, non-uniformly weighted representative days preserve planning-cost fidelity within 4%–6% of the full-resolution benchmark while reducing the temporal dimension by over 75%. This yields substantially faster CEP solves after a one-time clustering step. Our contribution is a structured, optimization-based approach to temporal aggregation that embeds a QUBO–QAOA subroutine within a conventional CEP workflow. By bridging QUBO-based temporal aggregation with conventional CEP modeling, this work presents a proof-of-concept framework that can be adapted as quantum hardware and hybrid solvers mature. Novelty: Unlike previous representative-day methods that rely solely on heuristic clustering, QUBO–QAOA has been integrated into a structured, optimization-based temporal reduction within a CEP workflow. Similar quantum techniques have been explored for related power-system problems (e.g., unit commitment, market optimization), but their direct use for temporal aggregation in CEP has been limited. • QUBO formulates the choice of k representative-day selection for temporal aggregation in CEP. • PCA reduces daily feature dimension before QUBO construction. • Half-month segmentation keeps each QUBO within current qubit hardware limits. • Using k=3 days yields 4–6% cost deviation on IEEE 9, 30, and 118 systems. • Compared against classical MCMC k-medoids representative-day selection.