Computing Optimal Control of Cascading Failure in DC Networks

We consider discrete-time dynamics for cascading failure in DC networks whose map is a composition of failure rule with control actions. Supply-demand at the nodes is monotonically nonincreasing under admissible control. Under the failure rule, a link is removed permanently if its flow exceeds capacity. We consider finite horizon optimal control to steer the network from an arbitrary initial state, defined in terms of active link set and supply-demand at the nodes, to a feasible state, i.e., a state which is invariant under the failure rule. There is no running cost and the reward associated with a feasible terminal state is the associated cumulative supply-demand. We propose two approaches for computing optimal control. The first approach, geared toward tree reducible networks, decomposes the global problem into a system of coupled local problems, which can be solved to optimality in two iterations. When restricted to the class of one-shot control actions, the optimal solutions to the local problems possess a piecewise affine property, which facilitates analytical solutions. The second approach computes optimal control by searching over the reachable set, which is shown to admit an equivalent finite representation by aggregation of control actions leading to the same reachable active link set. An algorithmic procedure to construct this representation is provided by leveraging and extending tools for arrangement of hyperplanes and polytopes. Illustrative simulations, including showing the effectiveness of projection-based approximation algorithms, are also presented.