Optimal Topology Design for Distributed Generation Networks Considering Different Nodal Invulnerability Requirements

Peng Jiao, Shengjun Huang, Bo Jiang, Tao ZhangSymmetry 2022 · DOI

TL;DR

This paper proposes a novel method for designing the optimal topology of distributed generation networks by considering individual node invulnerability requirements, rather than just system-level resilience. By formulating a mixed-integer linear programming model, the authors demonstrate how to minimize connection costs while ensuring that nodes can maintain power supply even when facing multiple failures, offering a more practical approach to network design.

What Problem Does This Paper Address?

The increasing adoption of renewable energy sources has led to a surge in distributed generation (DG) networks and microgrids. While these systems offer advantages like reduced voltage drops and improved energy efficiency, their reliability and resilience, especially in the face of failures, remain critical challenges. Existing research often focuses on system-level invulnerability, treating all nodes equally. However, in real-world scenarios, different nodes within a DG network have varying levels of importance and thus require different degrees of protection against failures. This paper addresses the gap by proposing a method to design network topologies that cater to these differentiated nodal invulnerability requirements, moving beyond a one-size-fits-all approach to resilience.

Methods

The core of this paper is a novel optimal topology design model for distributed generation networks. The authors introduce a mixed-integer linear programming (MILP) formulation to determine the optimal "preset" connection lines between nodes. A key innovation is the classification of nodes into three types (Type I, II, and III), each with a distinct invulnerability requirement. This translates into specific constraints for each node type, dictating how many simultaneous node failures it must withstand.

The N-K failure scenario analysis method is adapted for this purpose, but with a crucial difference: it's applied at the individual node level. The objective function aims to minimize the total length of preset connection lines, thereby reducing configuration costs. The MILP model incorporates these differentiated constraints, ensuring that even when multiple nodes fail, the remaining operational nodes can supply power to the affected ones by closing the pre-defined connection lines. The paper also includes linearization techniques to handle the non-linearities introduced by the problem formulation.

Datasets

The paper uses a simulated distributed generation network based on an aircraft platform as an example for numerical simulations. The "dataset" consists of:

  • Node positions: Provided as 2D coordinates.
  • Demand and Generation Capacity: Specified for each node, with a surplus of generation capacity relative to demand.
  • Node Types: Nodes are assigned to Type I, II, or III based on their invulnerability requirements.

No external datasets were used; the example network was constructed for the purpose of demonstrating the proposed method.

Key Insights and Ideas

  • Differentiated Nodal Invulnerability: The most significant insight is the shift from system-level to node-level invulnerability. Recognizing that not all nodes are created equal in terms of criticality allows for more efficient and cost-effective network design.
  • Topology as a Resilience Mechanism: The paper highlights how the network topology itself, through strategically placed "preset" connection lines, acts as a primary mechanism for ensuring resilience against failures.
  • Trade-off between Complexity and Resilience: As invulnerability requirements increase (from Type I to Type III), the network topology becomes more complex, with more connections and a higher overall cost (total line length). This quantifies the inherent trade-off.
  • Impact of Node Location: The spatial distribution of nodes significantly influences the topology. Nodes in remote locations require more connections to achieve higher invulnerability, increasing costs. Placing critical nodes centrally can leverage existing connections to improve overall network resilience.

Key Results

The paper presents simulation results demonstrating the effectiveness of the proposed model across different scenarios:

  • Varying Node Types:
    • All Type I: Results in a relatively simple topology with independent structures, as constraints are loose.
  • All Type II: Leads to a more complex, mesh-like topology as nodes need to satisfy more stringent requirements.
  • All Type III: Produces the most complex and connected topology, reflecting the strict constraints and higher costs required for maximum invulnerability.
  • Changing Nodal Generation Capacity: Increasing the surplus of power generation capacity generally decreases the total length of preset lines, up to a certain critical point. Beyond this point, further increases in capacity have a diminishing effect on topology optimization.
  • Changing Nodal Type: Assigning higher invulnerability requirements (Type II or III) to specific nodes, especially those in remote locations, significantly increases the total length of preset lines. Conversely, central nodes with high requirements have a less pronounced impact on the overall topology due to their advantageous positions.

Limitations and Future Work

  • High Constraint Complexity: The authors acknowledge that the proposed model suffers from high constraint complexity, particularly due to the linearization operations and the large number of potential failure scenarios. This complexity can make obtaining optimal solutions computationally challenging and time-consuming.
  • Computational Efficiency: Improving the computational efficiency of the model is identified as a key area for future work.
  • Alternative Optimization Algorithms: The paper suggests exploring genetic algorithms or other metaheuristic approaches to find high-quality solutions more efficiently, especially for larger networks.
  • Dynamic Scenarios: The current model focuses on static failure scenarios. Future work could explore dynamic failure scenarios and real-time reconfigurations.
  • Integration with Control Strategies: While topology is optimized, integrating this with optimal control strategies for power sharing and management during failures could further enhance resilience.

Summary updated

BibTeX
@article{jiao2022optimal,
  author = {Peng Jiao and Shengjun Huang and Bo Jiang and Tao Zhang},
  title = {{Optimal Topology Design for Distributed Generation Networks Considering Different Nodal Invulnerability Requirements}},
  journal = {Symmetry},
  year = {2022},
  doi = {10.3390/sym14051014},
}
Figure