The efficient bounding method for DC contingency analysis is improved using reciprocity properties. Knowing the consequences of the outage of a branch, these properties provide the consequences on that branch of various kinds of outages. This is used in order to reduce computation times and to get rid of some difficulties, such as those occuring when a branch flow is close to its limit before outage. Compensation, sparse vector, sparse inverse and bounding techniques are also used. A program has been implemented for single branch outages and tested on the actual French EHV 650 bus network. Computation times are 60 % of the Efficient Bounding method. The relevant algorithm is described in detail in the first part of this paper. In the second part, reciprocity properties and bounding formulas are extended f o r multiple branch outages and for multiple generator or load outages. An algorithm is proposed in order to handle all these cases simultaneously.K e v w o r b -Security analysis, DC contingency evaluation, compensation, bounding, sparse inverse, sparse vectors. I "Electricit6 de France (EDF) is renewing its EMS, with a project called CASOAR. As part of this project, two new models are being developed, a new security analysis model for real time use, and a unit commitment model including network security, to be used several times per day. For real time security analysis, AC contingency preselection and evaluation are used, whereas a simplified DC contingency preselection and evaluation method will be included in the unit commitment model. The purpose of this paper is to present the new DC contingency preselection and evaluation method, which speeds up the Efficient Bounding method for linear contingency analysis [l]. This new method can be used, as foreseen at EDF, associated with unit commitment, but it can also be used independently, for any purpose.At the beginning of this study, EDF had some experience of DC contingency evaluation and of lP, 1Q contingency screening; the latter method had been included in security constrained optimal power flows a long time ago [21; but much faster methods were needed. So, the present developments began with an analysis of basic results available in literature.Attention was first paid to matrix sparsity methods, which are a basic tool for speed. Sparsity oriented compensation [ 3 1 and the use of sparse inverse [ 4 1 were compared. For single branch outages, owing to the size of the EDF System ( 6 0 0 to 2000 buses), the use of sparse inverse in order to compute compensations led to shorter computation times and looked attractive as it provided all the compensations at the same time. Partial matrix refactorization [51, useful for AC computations and sustained topological changes, was not found directly 93 WM 234-5 PWRS by the IEEE Power System Engineering Committee of the IEEE Power Engineerinp. Society for mesentation A paper recommended and approved at the IEEE/ January 31 -February 3, December 28, ES 1993 WintLr Meeting, Coiumbus, OH, February 5, 1993. Manuscript submitt...