Abstract
Bipolar DC distribution networks (DCDNs) are gaining traction due to their high reliability and transmission capacity. However, the non-convex nature of traditional Optimal Power Flow (OPF) models complicates their efficient operation. This paper introduces a novel convex OPF model using power injection-based equations. By employing Second-Order Cone Programming (SOCP), McCormick relaxations, and a specialized Sequence Bound Tightening Algorithm (STBA), the proposed method ensures computational efficiency and high-precision solutions for voltage balancing and power optimization.
Introduction
Bipolar DC distribution networks, featuring positive, negative, and neutral conductors, provide a robust alternative to unipolar systems. Despite their advantages, managing these networks is mathematically challenging. Most existing OPF models rely on voltage and current as primary variables (current injection-based), leading to non-convex objective functions that are prone to local optima. Furthermore, these models do not intuitively reflect the power exchange between DC networks and AC grids.
This research addresses these limitations by formulating OPF equations where power is the central variable. This shift facilitates easier integration with distributed generation (DG) and provides a more intuitive framework for grid-level planning and economic dispatch.
Methodology
The proposed approach transforms the non-convex OPF problem into a convex one through a multi-stage process:
- Power Injection Formulation: Equations are derived using power as the primary decision variable, streamlining the interface with AC grids.
- Convexification Stack:
- Variable Substitution: Squared terms ($U^2$, $I^2$) are replaced with linear slack variables.
- SOCP Relaxation: Non-convex relationships between power, voltage, and current are converted into convex "cones."
- McCormick Relaxation: Bilinear terms (products of two variables) are handled by creating bounding envelopes.
- Taylor Expansion: Square root terms in static load models are linearized for compatibility.
- Sequence Bound Tightening Algorithm (STBA): To mitigate errors introduced by relaxations, the STBA iteratively shrinks the variable envelopes, ensuring the final solution maintains physical exactness.
Experimental Evaluation
The model was validated using two test cases:
- 5-Node Bipolar Feeder: Used to verify DG capacity configuration and the effectiveness of the STBA in error reduction.
- Modified IEEE 33-Node System: A standard benchmark adapted for bipolar DC operation to test scalability and performance in complex networks with multiple voltage balancers.
Key Results and Discussion
The proposed model demonstrated superior performance in both precision and speed:
- Rapid Convergence: The STBA algorithm proved highly efficient. In the 5-node test case, relaxation errors were reduced by nearly 1000x in the first iteration, reaching a precision of $10^{-6}$ within four iterations.
- Voltage Balancing: The model successfully addressed extreme load unbalance. In scenarios where unbalance reached 10.3%, the optimization of DG outputs brought the voltage unbalance factor (VUF) down to a negligible 0.0564%.
- Computational Speed: The 33-node system was optimized in approximately 4.5 seconds on standard hardware, highlighting the model's suitability for real-time applications.
Conclusion
The power-injection-based convex OPF model offers a computationally efficient and precise solution for the management of bipolar DC distribution networks. By shifting the focus to power variables and utilizing advanced relaxation techniques, the model provides a reliable tool for modern grid operation. Future research will focus on incorporating complex power electronics, such as DC-DC transformers, and extending the framework to multi-period optimization involving energy storage systems.
Summary updated