Dartmouth Events

Large cascading failures in electrical grids are rare but catastrophic, sometimes affecting millions of people and incurring substantial social and economic costs. Quantifying the risk of cascading failure is thus of critical importance for grid planning and operation, but is extremely challenging due to the rarity of blackout-causing events amongst the vast search space of possible combinations of component outages. In this talk, we present new approaches for estimating this risk in a computationally tractable way and show that it is orders of magnitude faster than current approaches.  We demonstrate the method on two realistic test cases; a model of the Polish grid with 2896 transmission lines under varying load conditions and a larger model based on the geography of the Western US with 12,706 transmission lines. Examining the sensitivity of overall blackout risk to individual transmission line failures facilitates the identification of low-cost strategies for reducing risk.  For example, we show that the risk of catastrophically large blackouts in the Polish grid model can be reduced by 83% for only a modest 1.9% increase in operational costs. Blackout risk estimation is often simplified by assuming that initiating outages are independent events, so that only pairwise initiating outages need be considered. In reality, common exogenous causes (such as severe storms) often induce spatial correlation in transmission line outages. We develop methods to incorporate such spatial correlation and show that the contribution to risk of 3-component initiating outages, relative to 2-component initiating outages, increases as a function of spatial correlation. We are currently exploring tractable methods for estimating the risk due to even higher-order contingencies.

RSVP Here

Read about Margaret Eppstein here.

Hosted by the Dartmouth Energy Collaborative