Lessons learnt on the EEAC method for fast contingency screening on RTE use-cases

Performing the associated simulations requires a high amount of computational power and may be incompatible with the performance and robustness constraints required for fast and reliable decision-making. A promising strategy to reduce this computational burden consists in contingency screening. Instead of running a detailed time-domain simulation for each fault, a preliminary screening module, approximate but fast, is used to discard “obviously safe” faults.

The remaining – hopefully few – faults are then evaluated using complete and accurate timedomain simulations.

The scientific challenge here is to design a screening method that is both reliable and fast enough. Data-based approaches, in particular machine-learning techniques, have been investigated in the past, notably within the iTesla project. However, their application to RTE operational data yielded limited success. As a result, RTE has shifted its focus toward modelbased alternatives and identified the Extended Equal Area Criterion (EEAC) as the most promising candidate. Originally developed in the late 1980s, EEAC is a direct method that approximates a multi-machine system as a One Machine Infinite Bus (OMIB) model equivalent. The method then applies the classic Equal Area Criterion to estimate the system’s critical clearing time, which is compared against protection action times to assess the fault severity and its potential impact on the system.

An open-source implementation of EEAC has been developed and extensively tested on a wide range of operational cases. Comparative studies with Eurostag, RTE’s reference time-domain simulation tool, have demonstrated that EEAC could help reduce the overall computation time of a dynamic security assessment by a factor of 2, provided that some precautions are taken.

This paper reports on the results of a comparison between EEAC and detailed time-domain simulations performed with Eurostag on real French transmission network cases, involving more than 2,000 buses and 400 synchronous machines. The study reveals several limitations of

EEAC that, to the best of our knowledge, have not been previously documented in scientific literature. In particular, EEAC is shown to provide unreliable results in scenarios where protection actions occur at different times, or where breaker operations lead to the islanding of a significant synchronous generator.