1: \begin{abstract}

2: We consider power networks in which it is not possible to satisfy all

3: loads at the demand nodes, due to some attack or disturbance to the

4: network. We formulate a model, based on AC power flow equations, to

5: restore the network to feasibility by shedding load at demand nodes,

6: but doing so in a way that minimizes a weighted measure of the total

7: load shed, and affects as few demand nodes as possible.  Besides

8: suggesting an optimal response to a given attack, our approach can be

9: used to quantify disruption, thereby enabling ``stress testing'' to be

10: performed and vulnerabilities to be identified.

11: % This solution provides guidance to operators on how to minimize the

12: % impact of a disruptive event on those served by the grid.

13: Optimization techniques

14: including nonsmooth penalty functions, sequential linear programming,

15: and active-set heuristics are used to solve this model. We describe an

16: algorithmic framework and present convergence results, including a

17: quadratic convergence result for the case in which the solution is

18: fully determined by its constraints, a situation that arises

19: frequently in the power systems application.

20: \end{abstract}

21: