Robust Volt-Var control of a smart distribution system under uncertain voltage-dependent load and renewable production

Abstract

As empirical reports validate dependency between power load and voltage at the distribution level, the distribution system operators are concerned with how to benefit from this relationship to achieve a more efficient Volt-Var control (VVC). Specifically, the potential impact of this dependency is a critical issue under uncertain operational parameters. For the first time, this paper develops a robust optimization (RO) model to handle the uncertain voltage-dependent load (VDL) and renewable power production involved in the VVC of a smart distribution system. The uncertain data are presented as uncertainty sets in a finite rolling horizon. The smart grid technology allows the VVC unit to schedule the system devices near real-time. This way, the VVC unit seeks to maintain voltages within their admissible limits through the schedules while minimizing the total cost of power losses, switching operations, and renewable energy spillage plus the penalty of voltage violations. A number of comparative studies are performed through a 69-bus test system to analyze the impact of load models, robustness parameters, renewable penetration level, and alternative uncertainty models on the VVC performance. The results verify that by ignoring the voltage-dependent terms of the ZIP model, the VVC unit estimates total cost and voltage violation with meaningful errors. Additionally, a conservative RO strategy provides a reduced voltage violation compared to a less conservative RO strategy while resulting in a higher cost. Compared to the deterministic and stochastic programming approaches, the RO model has the medium priority based on the metrics of the total cost, voltage violation, and computation time under a medium renewable penetration level, and even the top priority from the aspect of voltage violation under a high renewable penetration level.