Development and Application of a Novel Distribution Model Reduction Approach for Utility Feeders

Abstract

With an increasing interconnection of distributed PV into distribution systems, many utilities are facing difficulties to quickly assess grid impact from the intermittent resources. One barrier for PV interconnection study is the high computational complexity resulting from time-series simulation of various PV profiles to capture the uncertainty as well as system operational constraints. Moreover, thousands of nodes may exist in a typical distribution feeder and a utility may have to manage thousands of such feeders. To assist system operators and planners with fast assessment of grid impact of high penetration PV, this study proposes a new analytical method for distribution network reduction. The result is a tool that reduces up to 99% of the feeder nodes while preserving the accuracy of power flow results, such as node voltage errors. To maintain the high accuracy of reduced distribution networks, a modified Ward-equivalent method has been adopted to remove nodes out of interest. An aggregation method based on back sweep has been implemented to aggregate loads and distributed PV in laterals. The developed tool for model reduction has been tested with real-world distribution systems from Duke Energy. The high efficiency and accuracy of the tool is demonstrated by the simulation results.