2025-11-25T00:07:17.537013

Geometry of Distance Protection

Taylor, Domínguez-García
Distance relays detect faults on transmission lines. They face uncertainty from the fault's location and resistance, as well as the current from the line's remote terminal. In this paper, we aggregate this uncertainty with the Minkowski sum. This allows us to explicitly model the power grid surrounding the relay's line, and in turn accommodate any mix of synchronous machines and inverter-based resources. To make the relay's task easier, inverters can inject perturbations, or auxiliary signals, such as negative-sequence current. We use Farkas' lemma to construct an optimization for designing inverter auxiliary signals.
academic

Geometry of Distance Protection

Basic Information

  • Paper ID: 2510.04379
  • Title: Geometry of Distance Protection
  • Authors: Joshua A. Taylor (New Jersey Institute of Technology), Alejandro D. Domínguez-García (University of Illinois at Urbana-Champaign)
  • Classification: math.OC cs.IT cs.SY eess.SY math.IT
  • Publication Date: October 2025 (arXiv preprint)
  • Paper Link: https://arxiv.org/abs/2510.04379

Abstract

Distance relays are used to detect transmission line faults but face uncertainties regarding fault location, fault resistance, and remote-end current. This paper employs Minkowski sums to aggregate these uncertainties, enabling explicit modeling of the grid surrounding the relay line and accommodating arbitrary combinations of synchronous machines and inverter resources. To simplify the relay's task, inverters can inject perturbations or auxiliary signals (such as negative-sequence current). The authors use Farkas' lemma to construct optimization problems for designing inverter auxiliary signals.

Research Background and Motivation

Problem Description

  1. Challenges Facing Traditional Distance Protection: Distance relays detect faults by measuring local voltage and current to calculate impedance to the fault point, but face uncertainties in fault location, fault resistance, and remote-end current.
  2. New Issues Introduced by Inverter Resources: As inverter-based resources (IBR) increase their share in the grid, traditional distance protection characteristics based on synchronous generator assumptions may experience misoperations.
  3. Limitations of Existing Methods:
    • Traditional quadrilateral characteristics rely on physically-inspired ad-hoc adjustments to handle remote-end current
    • These adjustments assume synchronous generators as power sources and may fail in IBR-dominated grids
    • Lack of systematic methods for designing IBR auxiliary signals

Research Motivation

The authors aim to formalize uncertainty aggregation in distance protection through geometric methods (Minkowski sums) and develop optimization methods to design IBR auxiliary signals, thereby improving the reliability of distance protection in modern grids.

Core Contributions

  1. Theoretical Contribution: First application of Minkowski sums to formalize uncertainty aggregation in distance relay characteristics, providing mathematical foundations for traditional quadrilateral characteristics.
  2. Modeling Innovation: Proposes a framework capable of explicitly modeling the grid and power sources, accommodating arbitrary combinations of synchronous machines and IBR.
  3. Optimization Method: Constructs optimization problems for designing IBR auxiliary signals based on Farkas' lemma, ensuring the relay can clearly distinguish different fault scenarios.
  4. Computational Method: Develops efficient computational methods, including zonotope projection algorithms and zonogon relaxations, enabling online computation of the new characteristics.

Methodology Details

Task Definition

Input: Grid topology, power source uncertainties, fault parameter ranges Output: Relay characteristic region, optimal auxiliary signals Objective: Minimize auxiliary signal magnitude while ensuring all fault scenarios are separable

Physical Model

Network Modeling

  • Synchronous machines modeled as voltage sources: vS=vS∘v_S = v_S^{\circ}
  • IBR modeled as current sources: iC=iC∘+Δi_C = i_C^{\circ} + \Delta (including auxiliary signals)
  • Loads modeled using impedance or RLC equivalent circuits

Fault Loop Analysis

For different fault types, Kirchhoff's voltage law (KVL) is established:

Ground Fault (ag): vLa=mzz(iLa+kiL0)+mrrF(iLa+iRa)v_L^a = m_z z(i_L^a + ki_L^0) + m_r r_F(i_L^a + i_R^a)

Phase-to-Phase Fault (ab): vLa−vLb=mzz(iLa−iLb)+mrrF2(iLa−iLb+iRa−iRb)v_L^a - v_L^b = m_z z(i_L^a - i_L^b) + \frac{m_r r_F}{2}(i_L^a - i_L^b + i_R^a - i_R^b)

where mz∈[0,1]m_z \in [0,1] is the normalized fault distance and mr∈[0,1]m_r \in [0,1] is the normalized fault resistance.

Uncertainty Modeling

Power Source Uncertainty

Define the uncertainty set: U={[iC∘vS∘]+Σλ∣λ∈Λ}U = \left\{\begin{bmatrix} i_C^{\circ} \\ v_S^{\circ} \end{bmatrix} + \Sigma\lambda \mid \lambda \in \Lambda\right\}

where Λ\Lambda can be:

  • Norm-bounded: Λ={λ∣∥λ∥2≤1}\Lambda = \{\lambda \mid \|\lambda\|_2 \leq 1\}
  • Polyhedral (zonotope): Λ={λ∣Pλ≤1}\Lambda = \{\lambda \mid P\lambda \leq 1\}

Minkowski Sum Aggregation

For fault scenario η\eta, the apparent impedance uncertainty set is: Z~Aη=ZT⊕Z~Fη\tilde{Z}_A^{\eta} = Z_T \oplus \tilde{Z}_F^{\eta}

where ZT={mzz∣mz∈[mz‾,1]}Z_T = \{m_z z \mid m_z \in [\underline{m_z}, 1]\} is the line impedance set and Z~Fη\tilde{Z}_F^{\eta} is the fault impedance set.

Pre-Test and Post-Test Uncertainties

Post-Test Uncertainty (Relay Characteristic)

After testing, local current iLi_L is known, and the uncertainty set becomes convex, enabling efficient projection onto the complex plane.

Pre-Test Uncertainty (Auxiliary Signal Design)

Before testing, iLi_L is unknown, containing bilinear terms mzλm_z\lambda and mrλm_r\lambda, requiring convex relaxation:

Relaxation Method 1: UmRel={m[iC∘vS∘]+Σλ∣λ∈Λ,m∈[m‾,1]}U_m^{\text{Rel}} = \left\{m\begin{bmatrix} i_C^{\circ} \\ v_S^{\circ} \end{bmatrix} + \Sigma\lambda \mid \lambda \in \Lambda, m \in [\underline{m}, 1]\right\}

Relaxation Method 2: Separate variables λmz\lambda_{m_z} and λmr\lambda_{m_r}, removing coupling constraints.

Auxiliary Signal Optimization

Separation Condition

Define fault pair (η1,η2)(η_1, η_2) as separated by auxiliary signal Δ\Delta if and only if: P˘η1,η2(Δ)=∅\breve{P}^{η_1,η_2}(\Delta) = \emptyset

i.e., the intersection of two pre-test uncertainty sets is empty.

Optimization Problem

\min_{\Delta} &\quad \Delta^{\dagger}Q\Delta \\ \text{s.t.} &\quad \breve{P}^{η_1,η_2}(\Delta) = \emptyset, \quad \forall (η_1,η_2) \in \mathcal{F}^2 \end{align}$$ #### Farkas' Lemma Application Use Farkas' lemma to convert the empty set constraint into feasibility of a dual system: $$\breve{P}^{η_1,η_2}(\Delta) = \emptyset \Leftrightarrow \mathcal{S}^{η_1,η_2}(\Delta) \neq \emptyset$$ The final optimization problem becomes a biconvex program, solved using the ADMM algorithm. ## Experimental Setup ### Test System - **Base Network**: IEEE 14-bus system - **Power Source Configuration**: - Synchronous machine nodes: $\mathcal{S} = \{1,3,5,14\}$ - IBR nodes: $\mathcal{C} = \{2,4,7,8,12\}$ - Load nodes: $\{6,10,11,13\}$ ### Parameter Settings - Synchronous machines: Unit magnitude voltage sources with phase angles uniformly distributed in $[0, 2π/10]$ - IBR: Unit magnitude current sources with phase angles uniformly distributed in $[0, 2π/3]$ - Noise: Each source has $\lambda_k[1, α, α^2]^T$ added, $λ_k \in \Lambda_k$ (regular polygon) - Fault parameters: $k=2$ (zero-sequence compensation coefficient), $r_F=1$, $\underline{m_z}=0.15$ ### Computational Implementation - **Language**: Python - **Visualization**: Matplotlib - **Optimizer**: CVXPy + Clarabel solver - **Computation Time**: - All 11 fault types' $\tilde{Z}_A^{\eta}$: approximately 0.9 seconds - Zonogon relaxation $\hat{Z}_A^{\eta}$: approximately 1 millisecond ## Experimental Results ### Post-Test Uncertainty Sets Figure 4 presents the exact uncertainty sets $\tilde{Z}_A^{\eta}$ and their zonogon relaxations $\hat{Z}_A^{\eta}$ for ag and ab faults. Results show: - The relaxation approximates the exact set well - Adding cutting constraints $\text{Im}[z]\text{Re}[z_A^{\eta}] \geq \text{Re}[z]\text{Im}[z_A^{\eta}]$ tightens the relaxation ### Auxiliary Signal Design Results #### Visualization Analysis Figure 5 presents auxiliary signal sets for separating different fault pairs: - Minimum auxiliary signal separating (N,ag), (N,ab), and (ag,ab): $δ = j0.2$, $\|\Delta\|_2 ≈ 0.53$ - Minimum auxiliary signal separating all LG and LL faults: $δ = 1.6+j0.6$, $\|\Delta\|_2 ≈ 4.52$ #### Optimization Results Using ADMM algorithm optimization yields: $$\Delta ≈ [1.55+j0.57, 1.56+j0.56, 1.59+j0.50, 1.50+j0.64, 1.50+j0.63, 1.52+j0.60, 1.48+j0.66]^T$$ Final magnitude $\|\Delta\|_2 ≈ 4.35$, representing a 4% improvement over grid search results. ### Comparison with Traditional Methods - **Quadrilateral Characteristics**: Based on ad-hoc adjustments, assumes synchronous machine dominance - **Proposed Method**: Exact Minkowski sums, accommodates arbitrary power source combinations - **Computational Efficiency**: Zonogon relaxation completes in 1 millisecond, supporting online computation ## Related Work ### Distance Protection Fundamentals - Traditional distance protection based on impedance measurement and quadrilateral characteristics - Faces misoperation issues in IBR-dominated grids ### Impact of IBR on Protection - IBR control characteristics differ from synchronous machines, causing changes in fault current characteristics - Existing research primarily focuses on control strategy adjustments ### Auxiliary Signal Methods - Harmonic injection, negative-sequence current injection, and other methods - IEEE Standard 2800 requires IBR to inject 90° leading negative-sequence current ### Geometric Methods in Protection - First application of Minkowski sums to distance protection - Provides a systematic framework for uncertainty aggregation ## Conclusions and Discussion ### Main Conclusions 1. **Theoretical Contribution**: Minkowski sums provide a rigorous mathematical framework for uncertainty aggregation in distance protection 2. **Practical Value**: The new method accommodates complex power source combinations in modern grids 3. **Computational Feasibility**: Zonogon relaxation enables online computation 4. **Optimization Effectiveness**: The auxiliary signal design method based on Farkas' lemma ensures fault separability ### Limitations 1. **Model Simplifications**: - Relatively coarse power source modeling (voltage/current sources) - Network matrix calculated using nominal fault parameters - Ignores dynamic processes 2. **Computational Complexity**: - Bilinear characteristics of pre-test uncertainty sets require relaxation - ADMM algorithm convergence depends on initialization 3. **Practical Considerations**: - Requires accurate network models and power source parameters - Practical implementation of auxiliary signals may face constraints ### Future Directions 1. **Electromagnetic Transient Simulation Validation**: Verify method effectiveness in detailed power system simulations 2. **Network Matrix Improvements**: Consider multiple nominal fault parameters or adaptive updates 3. **Multi-Relay Coordination**: Extend to coordinated protection of multiple transmission lines 4. **Other Protection Schemes**: Apply core ideas to overcurrent protection, time-domain protection, etc. ## In-Depth Evaluation ### Strengths 1. **Strong Theoretical Innovation**: - First application of Minkowski sums to distance protection - Provides rigorous mathematical framework replacing ad-hoc adjustments - Clever and effective application of Farkas' lemma 2. **High Practical Value**: - Addresses real-world problems in IBR-dominated grids - Efficient computational methods support online applications - Systematic auxiliary signal design 3. **Complete Methodology**: - Comprehensive analysis covering 11 fault types - Multiple convex relaxation methods provided - Detailed algorithm implementation and open-source availability ### Shortcomings 1. **Model Assumption Limitations**: - Oversimplified power source models may impact practical applications - Ignores grid dynamic characteristics - Insufficient treatment of intermediate feeders 2. **Insufficient Validation**: - Lacks verification with actual grid data - No comparison with commercial relays - Electromagnetic transient simulation validation pending 3. **Computational Complexity**: - Global optimality difficult to guarantee for bilinear optimization problems - Scalability to large-scale networks not fully verified ### Impact Assessment 1. **Academic Impact**: - Introduces new mathematical tools to power system protection - May inspire more geometric methods in protection applications - Important reference value for IBR protection research 2. **Engineering Value**: - Provides theoretical foundation for modern grid protection equipment upgrades - Auxiliary signal design method has practical application potential - Contributes to relevant standards and regulations development ### Applicable Scenarios 1. **High IBR Penetration Grids**: Particularly suitable for systems with high renewable energy penetration 2. **New Transmission Lines**: Can consider new protection strategies during design phase 3. **Protection Equipment Upgrades**: Provides solutions for intelligent upgrades of existing relays 4. **Grid Planning**: Considers protection performance optimization in grid planning ## References The paper cites 26 important references, primarily including: - Classical distance protection textbooks: Ziegler (2011), Horowitz et al. (2022) - Recent research on IBR impact on protection: Kasztenny (2022), Baeckeland et al. (2022) - Auxiliary signal-related work: Saleh et al. (2020), Banaiemoqadam et al. (2019) - Mathematical tools: Minkowski sum theory, Farkas' lemma, zonotope theory --- **Overall Assessment**: This is a high-quality academic paper with significant contributions in both theoretical innovation and practical value. The authors skillfully introduce geometric methods into the traditional power system protection field, providing new perspectives for addressing protection challenges in modern grids. Although there is room for improvement in model assumptions and validation, the core ideas and methodological framework possess important academic value and engineering application prospects.