Análise de Sistemas de Distribuição com REDs Utilizando Python e DSS: Comparação entre Armazenamento Centralizado e Distribuído

Abstract

This work analyzes the impact of battery energy storage allocation in distribution networks with distributed generation, comparing centralized and distributed strategies in terms of voltage violation mitigation. The study models a low-voltage distribution feeder with distributed photovoltaic generation, single-phase loads, and representative time-varying demand and generation profiles. To do so, it integrates quasi-static time-series simulations developed in Python with OpenDSS and formulates an optimization problem with discrete and continuous decision variables, solved through a genetic algorithm. The objective function seeks to minimize steady-state voltage violation indicators while considering technical constraints related to the size and placement of the storage systems. The results show that the insertion of distributed generation without storage worsens the electrical performance of the network, increasing the occurrence of overvoltages and expanding the violation indexes. The centralized strategy significantly reduces these problems, eliminating critical violations and substantially decreasing precarious violations, although residual violations remain. In contrast, the distributed strategy presents superior performance, fully eliminating the violations in the evaluated arrangements and providing more homogeneous voltage support along the feeder. However, the results also indicate that progressively increasing the number of batteries does not lead to proportional gains, which suggests diminishing marginal returns. It is concluded that distributed storage is more effective in ensuring voltage compliance, whereas the centralized solution provides relevant improvement with lower installed capacity.