Introduction

Transmission System Operators (TSOs) often assess the impact of new power plant connections by simulating n-1 and even n-k contingencies to determine whether the inclusion of the new plant poses a threat to either static or dynamic security standards. However, in large grids, the number of credible contingencies can quickly reach several thousand, making time-domain (TD) simulation of the full set impractical during early planning—only the most severe cases require detailed analysis to reveal the instability mechanisms and lead timely mitigation strategies. Static screening approaches are often used to reduce this burden but commonly rely on SCR-based indices, which poorly capture Inverter-based resources (IBR) dynamics. This paper addresses this gap with a contingency severity screening methodology based on static calculations as proxies for dynamic behavior, allowing early-stage filtering without requiring full TD simulations.

Methodology

The screening relies on three static indicators, computed from load-flow and short-circuit solutions: the ratio of pre-fault to post-fault point of connection (POC) impedance ΔZth, the ratio of pre-fault to faulted POC voltage ΔVth, and power transfer shift across POC branches. These indicators serve as proxies for the underlying power-angle characteristic, aiming to quantify the disturbance magnitude by capturing the extent of imbalance experienced at the POC by each power plant. These are subsequently used as input features in a supervised learning framework, where their relationship to a dynamic performance indicator (target variable, derived from TD simulations) is modeled using regression methods. Training and validation of the regression model are performed using a dataset of operating points generated by a Monte Carlo method, each secured under static operating conditions. The proposed screening methodology is evaluated on a modified IEEE 39-bus power system using a phasor-domain model. IBRs account for 50% of the installed capacity.

Results

The proposed screening method is evaluated across several operating points, each with 68 simulated n-1 contingencies. The estimated severity showed a strong correlation with the dynamic performance indicator. Key performance insights from the static screening are as follows:

• 80% of the contingencies leading to instability in TD simulations are ranked within the top 10 most severe by the static screening method, highlighting the effectiveness in capturing the most critical events.
• Over 80% of all contingencies ranked with the proposed method (sorted from most to least severe) fall within three positions of their dynamic ranking. This reinforces the reliability of static screening to prioritize severe contingencies.
• By reserving dynamic simulations only for the top-ranked contingencies identified through static screening, the number of required simulations can be reduced by approximately 75%, while still capturing most critical cases and maintaining a conservative safety margin.

Conclusions

This work presents a large-signal stability screening method that relates each contingency to an expected system performance outcome, consistent with TSOs’ DSA practices. The approach achieves high alignment with dynamic results and enables early-stage filtering of contingencies based on static indicators. Thereby reducing the need for time-domain simulations and supporting more scalable contingency assessment in planning-stage studies.