An MILP-based approach to tree partitioning with minimal power flow disruption and generator coherency constraints

Tree partitioning has recently been proposed in the power systems literature as a less severe alternative to controlled islanding. In this paper, we formulate an optimization problem to tree partition a network with minimal power flow disruption and generator coherency constraints. We propose a single-stage MILP formulation to compute optimal solutions. Numerical experiments show that our MILP-based approach drastically decreases the power flow disruption when compared to an earlier proposed two-stage approach based on spectral clustering. Moreover, using a search space reduction procedure based on the Steiner Tree problem, the MILP-based approach computes near-optimal tree partitions in sub-second time for instances up to 500 buses.