Two-Stage Energy Storage Allocation Considering Voltage Management and Loss Reduction Requirements in Unbalanced Distribution Networks

Hu Cao, Lingling Ma, Guoying Liu, Zhijian Liu, Hang Dong

TL;DR

This paper introduces a novel two-stage approach for optimally placing and sizing energy storage systems (ESS) in complex, unbalanced urban distribution networks. By developing a Comprehensive Power Flow Sensitivity Variance (CPFSV) metric and employing an enhanced optimization algorithm, the method effectively mitigates voltage issues and network losses caused by the increasing integration of electric vehicles (EVs) and solar (PV) power.

What Problem Does This Paper Address?

Modern urban distribution networks are facing significant challenges due to the rapid influx of distributed energy resources (DERs) like EV charging stations and rooftop solar panels. These DERs, especially in older, unbalanced networks, lead to issues such as heavy loads, voltage sags, and increased network losses. Traditional network upgrades are often costly and inefficient. Energy storage systems (ESS) offer a promising solution, but their effective deployment requires careful consideration of location and capacity to maximize benefits and minimize costs. This paper addresses the critical need for a systematic and efficient method to determine the optimal placement and sizing of ESS in these complex, real-world scenarios.

Methods

The paper proposes a two-stage methodology:

  1. Energy Storage Location Selection: A new metric, the Comprehensive Power Flow Sensitivity Variance (CPFSV), is introduced. This metric combines voltage sensitivity and network loss sensitivity to identify optimal bus locations for ESS. It's calculated using a time-series analysis to capture dynamic network behavior.
  2. Energy Storage Capacity Planning: A multi-objective optimization model is developed. This model considers the benefits of peak shaving and valley filling, the costs associated with ESS (investment, operation, maintenance), and the impact on distribution network voltage deviations.

To solve this complex optimization problem, the authors employ an improved Beluga Whale Optimization Algorithm (TWBWO). This algorithm is enhanced with:

  • Chaos Reverse Learning: To improve the initial population's quality and the algorithm's ability to escape local optima.
  • Water Wave Algorithm Refraction: To further boost the algorithm's global search capability and prevent premature convergence.

Datasets

The proposed method is validated using two datasets:

  • A modified IEEE-33 node system: This benchmark system is adapted with actual operational data from Kunming, China, incorporating PV sources, EV charging stations, and urban village loads.
  • A 199-node actual distribution network system: This real-world system, also from Kunming, with measured line parameters and load data, is used to demonstrate the practical engineering applicability of the method.

Key Insights and Ideas

  • CPFSV as a Unified Metric: The CPFSV metric is a significant contribution, offering a single, time-series-based indicator that effectively balances the competing needs of voltage management and loss reduction for ESS placement.
  • Two-Stage Optimization is Crucial: Separating the location selection from capacity planning simplifies the problem and allows for more targeted optimization, avoiding the "solution dimension disaster" of integrated approaches.
  • Enhanced Metaheuristics Matter: The TWBWO algorithm, with its chaos reverse learning and water wave refraction, demonstrates superior performance in finding diverse and optimal solutions for complex multi-objective problems compared to standard BWO.
  • Real-World Validation is Key: Testing on both a modified benchmark and a large-scale actual distribution network provides strong evidence of the method's robustness and practical utility.

Key Results

  • Peak-to-Valley Difference Reduction: The proposed method reduced the peak-to-valley load difference by 43.7% compared to the original system and 51.1% compared to a system with only DERs (EVs and PV).
  • CPFSV Reduction: The maximum CPFSV was reduced by 52% compared to the original system and 75.1% compared to the system with DERs.
  • Voltage Management: In the 199-node system, the integration of ESS significantly improved voltage profiles, bringing low voltages back into the normal range during peak hours and effectively managing high voltages during low load periods.
  • Cost Savings: While not explicitly quantified as a single percentage, the analysis indicates that Case 5 (using the proposed CPFSV strategy) achieved a lower total cost compared to other scenarios, particularly Case 4, due to higher ESS utilization efficiency and reduced network losses.

Limitations and Future Work

  • Computational Complexity: While the TWBWO algorithm is improved, solving complex multi-objective optimization problems for large networks can still be computationally intensive.
  • Stochastic Nature of DERs: The paper acknowledges the stochastic nature of DERs but focuses on a typical day's data. Future work could explore more advanced probabilistic or robust optimization techniques to handle greater uncertainties.
  • Dynamic Network Changes: The current model uses time-series data for a typical day. Investigating the impact of more dynamic network changes or different seasonal variations could further enhance the method's applicability.
  • Scalability: While tested on a 199-node system, further studies on even larger and more complex networks would be beneficial to confirm scalability.

Summary updated

Abstract

The authors propose a two-stage sequential configuration method for energy storage systems to solve the problems of the heavy load, low voltage, and increased network loss caused by the large number of electric vehicle (EV) charging piles and distributed photovoltaic (PV) access in urban, old and unbalanced distribution networks. At the stage of selecting the location of energy storage, a comprehensive power flow sensitivity variance (CPFSV) is defined to determine the location of the energy storage. At the energy storage capacity configuration stage, the energy storage capacity is optimized by considering the benefits of peak shaving and valley filling, energy storage costs, and distribution network voltage deviations. Finally, simulations are conducted using a modified IEEE-33-node system, and the results obtained using the improved beluga whale optimization algorithm show that the peak-to-valley difference of the system after the addition of energy storage decreased by 43.7% and 51.1% compared to the original system and the system with EV and PV resources added, respectively. The maximum CPFSV of the system decreased by 52% and 75.1%, respectively. In addition, the engineering value of this method is verified through a real-machine system with 199 nodes in a district of Kunming. Therefore, the energy storage configuration method proposed in this article can provide a reference for solving the outstanding problems caused by the large-scale access of EVs and PVs to the distribution network.
BibTeX
@inproceedings{cao2024two,
  author = {Hu Cao and Lingling Ma and Guoying Liu and Zhijian Liu and Hang Dong},
  title = {{Two-Stage Energy Storage Allocation Considering Voltage Management and Loss Reduction Requirements in Unbalanced Distribution Networks}},
  booktitle = {Energies},
  year = {2024},
  doi = {10.3390/en17246325},
}
Figure