Nonlinear structure-preserving network reduction using holomorphic embedding

Network reduction is an effective tool for reducing the complexity of many analysis, design and optimization problems. However, many of the conventional reduction methods, like Ward, and REI (Radial, Equivalent, Independent) are only accurate at the base case. When the operating condition changes, the reduced model does not match the full model performance because linearization is used somewhere in the process. In this paper, a new reduction method that preserves the model's nonlinear structure using the holomorphic embedding (HE) technique is proposed, to generate network reductions which are accurate over a broader range of operating conditions. When applied to the power-flow problem, simulation results show that the proposed method can significantly improve bus-voltage and branch-flow accuracy, matching the full-model power-flow solution exactly when moving along the so-called α line. In addition, the HE reduction is more efficient than traditional methods when calculating nonlinear network solutions under many operating conditions.