DistFlow Extensions for AC Transmission Systems

Carleton Coffrin, Hassan L. Hijazi, Pascal Van HentenryckarXiv 2015 · OpenReview

Abstract

This paper addresses the gap between power distribution and transmission modeling by extending the Convex DistFlow (CDF) relaxation to incorporate complex transmission features, including transformers, bus shunts, and line charging. We provide a formal mathematical proof establishing that this extended CDF model is equivalent to the widely utilized Second-Order Cone (SOC) relaxation. This equivalence ensures that researchers and practitioners can select the formulation best suited to their computational requirements without compromising modeling accuracy.

Introduction

The Power Flow equations, which govern electricity movement across a grid, are inherently non-linear and non-convex, making them computationally challenging for large-scale optimization. To facilitate grid management tasks like Optimal Power Flow (OPF), engineers employ convex relaxations—simplified mathematical models that provide efficient solutions and bounds on the true optimal.

Historically, three primary relaxations have dominated the field: Semi-Definite Programming (SDP), Second-Order Cone (SOC), and Convex DistFlow (CDF). While the equivalence of SOC and CDF has been well-established for simple radial distribution networks, the standard CDF model lacked the necessary components to represent high-voltage transmission systems accurately. Specifically, it omitted bus shunts, line charging, and transformers. This work modernizes the CDF framework to bridge this gap, enabling its application across the entire power system.

Methodology and Notation

The derivation of the extended power flow relaxations follows a first-principles approach, identifying fundamental algebraic properties of Ohm’s Law and Kirchhoff’s Current Law (KCL) that remain invariant across network types.

Table 1: Nomenclature and symbols used in the extended DistFlow and SOC formulations.

The model extension incorporates three critical transmission parameters:

  1. Bus Shunts ($Y^s$): Components representing power leakage to the ground.
  2. Line Charging ($Y^c$): Capacitive effects inherent in long-distance transmission lines.
  3. Transformers ($T$): Devices facilitating voltage level transitions across the grid.

The methodology involves "lifting" these complex equations into higher-dimensional spaces—the $W$-space for SOC and the $L$-space for CDF—and subsequently relaxing the constraints into convex forms.

Theoretical Equivalence

The central contribution of this work is the proof of Theorem 5.1, which establishes a mathematical bijection between the solutions of the Extended SOC and Extended CDF models. This result demonstrates that the equivalence between these two relaxations is not merely a property of simple radial networks but a robust characteristic that persists even with the inclusion of complex transmission physics.

A key insight is the "L-Space" advantage. While SOC relaxations focus on voltage products at disparate nodes, the CDF (L-space) approach emphasizes line losses and current magnitudes. This perspective is often more intuitive for power engineers and provides an alternative numerical path for optimization solvers. Furthermore, the paper demonstrates that even with transformers and line charging, voltage products can be factored in terms of power and voltage magnitude variables, which is essential for the validity of these relaxations.

Applications and Benchmarks

The theoretical framework developed here is designed for application to standard power system benchmarks, such as:

  • NESTA (The NICTA Energy System Test Case Archive): A library of power system cases for testing optimization algorithms.
  • PGLib-OPF: IEEE PES Task Force benchmarks for Optimal Power Flow.

These datasets contain the complex parameters (shunts, charging, transformers) that the original CDF model was previously unable to accommodate.

Conclusion and Future Work

By proving that Extended CDF $\iff$ Extended SOC, this research ensures that theoretical guarantees or "tightness" results found in one model are directly applicable to the other. While this work provides the formal proof and a real-number implementation, it remains primarily theoretical.

Future research should focus on extensive numerical benchmarking. Although the models are mathematically equivalent, different commercial solvers (e.g., Gurobi, MOSEK) may exhibit varying performance when handling the $L$-space equations versus the $W$-space equations. Determining which formulation offers superior computational efficiency for continent-scale power grids remains a critical next step.

Summary updated

BibTeX
@misc{coffrin2015distflow,
  author = {Carleton Coffrin and Hassan L. Hijazi and Pascal Van Hentenryck},
  title = {{DistFlow Extensions for AC Transmission Systems}},
  year = {2015},
  url = {https://openreview.net/forum?id=iKIxkdxI2X},
}
Figure