Joint Distributed Generation Maximization and Radial Distribution Network Reconfiguration

Abstract

This paper studies an optimization problem for joint radial distribution system network reconfiguration and power dispatch for distributed generation (DG) maximization. We provide counterexamples to show that for DG maximization, standard techniques such as interior point method (as in Matpower), linear approximation and second order cone relaxation (e.g., by Jabr et.~al.~and by Farivar and Low) do not deliver the desired control. Instead, we propose a control decision model based on exact DistFlow equations bypassing relaxation and a solution approach based on spatial branch-and-bound algorithm. We justify our work with comparative studies and numerical demonstrations with benchmarks and a 533-bus real example, performing reconfiguration and power dispatch in a time scale relevant for control center applications.

Figures (12)

• Figure 1: 33-bus distribution system from baran1989network with five tie-switches (dashed lines). Line width is proportional to current rating labeled next to the line.
• Figure 2: Reconfiguration with $K = 2$ (line {2, 22} opened and line {24, 28} closed). Bus color shows the voltage level and the percentage of current rating of bottleneck lines are shown. Total DG output is 4.33 MW.
• Figure 3: Reconfiguration with $K = 4$ (lines {2, 22}, {28, 29} opened, and lines {24, 28}, {17, 32} closed). Bus color shows the voltage level and the percentage of current rating of bottleneck lines are shown. Total DG output is 5.67 MW.
• Figure 4: 3-bus distribution system example demonstrating failure of the relaxation approach. The voltage at bus 0 is 1 pu.
• Figure 5: Pareto curves of 33bw benchmark (600A line rating) due to the proposed model in \ref{['opt:reconfig']} and conic relaxation in \ref{['opt:SOCP_relax']}