Convex optimal power flow based on power injection-based equations and its application in bipolar DC distribution network

Yiyao Zhou, Qianggang Wang, Tao Huang, Jianquan Liao, Yuan Chi, Niancheng Zhou, Xiaolong Xu, Xuefei Zhang

TL;DR

This paper introduces a novel convex Optimal Power Flow (OPF) model for bipolar DC distribution networks (DCDNs) that uses power injection-based equations rather than traditional current-based ones. By combining Second-Order Cone Programming (SOCP), McCormick relaxations, and a specialized Sequence Bound Tightening Algorithm (STBA), the authors provide a computationally efficient method to optimize power exchange and voltage balance with high precision.

What Problem Does This Paper Address?

Bipolar DC distribution networks—which feature positive, negative, and neutral conductors—are increasingly popular because they offer higher reliability and transmission capacity than unipolar systems. However, they are significantly more complex to manage.

Most existing OPF models for these networks use voltage and current as primary decision variables (current injection-based). While this describes the physics well, it makes the objective functions (like minimizing cost or power loss) non-convex because they involve the product of voltage and current. This non-convexity often leads to "local optima" (sub-optimal solutions) or requires massive computational power to solve. Furthermore, current-based models don't easily reflect the power exchange between the DC network and the AC grid. There is a critical need for a model where power is the primary variable, allowing for more intuitive planning and operation.

Methods

The authors propose a multi-stage transformation to turn a "hard" non-convex problem into a "soft" convex one that commercial solvers can handle easily:

  1. Power Injection Formulation: They derive power flow equations where power is the central variable, making it easier to interface with distributed generation (DG) and AC grids.
  2. Convexification Stack:
    • Variable Substitution: Replacing squared terms ($U^2$ and $I^2$) with linear slack variables.
  3. SOCP Relaxation: Converting non-convex power-voltage-current relationships into convex "cones."
    • McCormick Relaxation: Handling bilinear terms (products of two variables, like $U_+ \cdot U_-$) by creating a "box" or envelope around them.
  4. Taylor Expansion: Linearizing the square root terms found in standard static load models.
  5. Sequence Bound Tightening Algorithm (STBA): Because relaxations (like McCormick) can introduce errors, the authors developed STBA. This iterative algorithm shrinks the "envelopes" around variables in each step, ensuring the final solution is "exact"—meaning the mathematical shortcuts taken don't compromise the physical reality of the power flow.

Datasets

The researchers evaluated their model using two primary test cases:

  • 5-Node Bipolar Feeder: A custom-designed small-scale system used to verify the effectiveness of DG capacity configuration and the STBA error reduction.
  • Modified IEEE 33-Node System: A standard power system benchmark adapted for bipolar DC operation to test the model's performance in a larger, more complex network with multiple voltage balancers and DGs.

Key Insights and Ideas

  • Power as a First-Class Citizen: By shifting the decision variables to power, the model becomes much more useful for economic dispatch and grid-level planning where kW and MW are the primary units of interest.
  • The "Exactness" Proof: The paper provides a mathematical proof that their SOCP relaxation is exact under practical conditions (specifically for lossy lines). This gives engineers confidence that the "relaxed" solution isn't just an approximation, but the true global optimum.
  • Neutralizing Asymmetry: Bipolar networks often suffer from voltage unbalance because loads on the positive and negative poles are rarely equal. This model explicitly optimizes "Voltage Balancers" (VBs) to mitigate this, which is crucial for equipment safety.

Key Results

  • Rapid Error Reduction: The STBA algorithm is remarkably efficient. In the 5-node test case, it reduced relaxation errors by nearly 1000x in a single iteration, reaching a high-precision tolerance ($10^{-6}$) in just four iterations.
  • Superior Voltage Balancing: In a scenario with extreme load unbalance (where some nodes saw 10.3% unbalance), the model optimized DG output to bring that unbalance down to a negligible 0.0564%.
  • Computational Efficiency: The entire optimization process for the 33-node system took approximately 4.5 seconds on a standard desktop computer, making it suitable for near-real-time operational use.

Limitations and Future Work

  • Component Scope: The current model focuses on basic network components (lines, loads, DGs, and Voltage Balancers). It does not yet incorporate more complex power electronics like DC-DC transformers or "Electric Springs," which are becoming common in modern microgrids.
  • Lossy Line Assumption: The proof of exactness relies on the assumption that the network lines have resistance (are "lossy"). While true for physical wires, the behavior in superconducting or extremely low-resistance scenarios might require further theoretical refinement.
  • Future Directions: The authors suggest extending this framework to multi-period optimization (considering battery storage over 24 hours) and exploring more complex converter topologies.

Summary updated

Abstract

Optimal power flow (OPF) is a fundamental tool for analyzing the characteristics of bipolar DC distribution network (DCDN). However, existing OPF models face challenges in reflecting the power distribution and exchange of bipolar DCDN directly since its decision variables are voltage and current. This paper addresses this issue by establishing a convex OPF model that can be used for the planning and operation of bipolar DCDN. First, the power flow characteristics of bipolar DCDN are revealed through power injection-based equations, upon which the original OPF model is established. Next, the original OPF model undergoes a transformation into a convex OPF model based on second-order cone programming (SOCP) through variable substitution, secondorder cone relaxation, McCormick relaxation, and first-order Taylor expansion, respectively. Finally, the sequence bound tightening algorithm (STBA) is employed to tighten the boundaries of McCormick envelopes in each iteration to ensure the exactness of the convex OPF model. The effectiveness of this novel OPF model for bipolar DCDN is verified through two case studies, i.e., capacity configuration of distributed generation (DG) and operation optimization of bipolar DCDN.
BibTeX
@inproceedings{zhou2024convex,
  author = {Yiyao Zhou and Qianggang Wang and Tao Huang and Jianquan Liao and Yuan Chi and Niancheng Zhou and Xiaolong Xu and Xuefei Zhang},
  title = {{Convex optimal power flow based on power injection-based equations and its application in bipolar DC distribution network}},
  booktitle = {Electric},
  year = {2024},
  doi = {10.1016/j.epsr.2024.110271},
}
Figure