The decision tree transient stability method is revisited via a case study carried out on the French EHV power system. In short, the method consists of building off-line decision trees, able to subsequently assess the system transient behavior in terms of precontingency parameters (or ''attributes'') of it, likely to drive the stability phenomena. This case study aims at investigating practical feasibility aspects and features of the trees, at enhancing their reliability to the extent possible, and at generalizing them. Feasibility aspects encompass data base generation, candidate attributes, stability classes; tree features concern in particular complexity in terms of their size and interpretability capabilitities, robustness with respect to both their building and use. Reliability is enhanced by defining and exploiting pragmatic quality measures. Generalization concerns multicontingency, instead of single-contingency trees. The results obtained show real promise for the method to meet practical needs of electric power utilities.

A study was conducted on the extra high voltage (EHV) French power system in order to explore the extended equal-area criterion and test its suitability as a fast transient stability indicator. The assumptions underlying the method are reexamined, causes liable to invalidate them are identified, and indices are devised to automatically circumvent them. The selection of candidate critical machines is also reconsidered, and an augmented criterion is proposed. The various improvements were developed and tested on about 1000 stability scenarios, covering the entire 400 kV system. The severity of the scenarios, resulting from the combination of weakened pre and post-fault configurations, subjected the method to particularly stringent conditions. The simulation results are summarized.< >