Degree Type


Date of Award


Degree Name

Doctor of Philosophy


Electrical and Computer Engineering

First Advisor

Chen-ching Liu


The intentional area partitioning and automated distribution system restoration are two important "Smart Grid" technologies to enhance the robustness of a power network and improve the system reliability. In this dissertation, the research work is focused on deriving and implementing efficient graph-theoretic algorithms to analyze and solve such two real-world problems in power systems as follows.

In response to disturbances, a self-healing system reconfiguration that splits a power network into self-sufficient islands can stop the propagation of disturbances and avoid cascading events. An area partitioning algorithm that minimizes both real and reactive power imbalance between generation and load within islands is proposed. The proposed algorithm is a smart grid technology that applies a highly efficient multilevel multi-objective graph partitioning technique. The simulation results obtained on a 200- and a 22,000- bus test systems indicate that the proposed algorithm improves the voltage profile of an island after the system reconfiguration compared with the algorithm that only considers real power balance. In doing so, the algorithm maintains the computational efficiency.

The distribution system restoration is aimed at restoring loads after a fault by altering the topological structure of the distribution network by changing open/closed states of some tie switches and sectionalizing switches in the distribution system. A graph-theoretic distribution system restoration strategy that maximizes the amount of load to be restored and minimizes the number of switching operations is developed. Spanning tree based algorithms are applied to find the candidate restoration strategies. Unbalanced three-phase power flow calculation is performed to guarantee that the proposed system topology meets all electrical and operational constraints. Simulation results obtained from realistic feeder models demonstrate the effectiveness of the proposed approach.


Copyright Owner

Juan Li



Date Available


File Format


File Size

92 pages