Speaker
Description
The integration of wind and solar generation leaves grid operators with input data known only within ranges. Interval power flow (IPF) seeks guaranteed bounds on bus voltages and line flows given interval-valued injections and loads. Classical approaches based on interval and affine arithmetic propagate these intervals through the power flow equations and suffer from the wrapping effect: dependencies between variables are lost, the computed hull is systematically wider than the true solution set, and the resulting operational margins are more restrictive than physics requires.
This work proposes two alternative routes exploiting quantum computers. The first obtains each bound by directly searching for the extremal operating point over discretized input intervals, encoded as a well-understood QUBO with the power flow equations as penalty terms and solved with QAOA on a 24-qubit IQM Star processor. The second casts IPF as a non-diagonal Hamiltonian, opening the problem to sample-based quantum diagonalization methods.
The poster presents both encodings, the qubit cost as a function of bus count and discretization resolution, and the benchmarking protocol for 3-bus to 5-bus test systems against established classical algorithms.