Risk-Based Arrester Placement in Substations: A Multi-Objective Probabilistic Approach

Abstract

This study proposes a risk-based arrester placement framework in substations using a multi-objective probabilistic approach that combines electromagnetic transient (EMT) modeling, Latin Hypercube Sampling (LHS) for uncertainty propagation, and NSGA-II to generate a set of cost–risk Pareto solutions. The model incorporates lightning and switching surge sources, equipment characteristics (BIL/LIWV, arrester V–I curves, energy duty limits), and technical–economic consequences (EENS, interruption costs). A case study on a double busbar substation with eight candidate points shows three representative solutions: minimum-cost (3 arresters), knee-point (5 arresters), and minimum-risk (7 arresters). The knee-point solution—arresters at incomers L1–L2, the main bus, and HV & MV transformer terminals—reduces Expected Risk by ≈ 58% and SAIDI by ≈ 57% compared to the deterministic baseline (arresters only at incomers), with improved insulation coordination margins (e.g., p95 of the transformer HV terminals drops from ~712 kV to ~635 kV) and energy reserves of ≥30% over manufacturer specifications. Sensitivity analysis identifies ground grid resistance (Rg), lightning peak current, and strike position as the primary risk drivers, indicating that co-optimization of arresters and grounding has the potential to further improve performance. The results confirm that this approach is robust and economical, and ready to be adopted as a basis for protection investment decisions in modern substations.