🎓 Lesson 15 D5

Sensitivity-Based Voltage Collapse Prediction

Voltage collapse prediction using sensitivity-based methods means spotting how small changes in power system conditions—like load increases or generator outages—can quickly lead to a sudden, uncontrollable drop in voltage.

🎯 Learning Objectives

✓ Calculate voltage sensitivity indices (e.g., dV/dP, dV/dQ) from power flow Jacobian matrices
✓ Analyze system weak buses using participation factors derived from singular value decomposition of the sensitivity matrix
✓ Apply modal sensitivity analysis to rank transmission lines and generators by their impact on voltage stability margin
✓ Explain the physical meaning of the minimum singular value of the reduced Jacobian as a voltage stability indicator

📖 Why This Matters

In mining operations, remote power systems—such as those feeding large AC drive haul trucks, comminution mills, or ventilation fans—are often radial, lightly loaded, and highly sensitive to load fluctuations. A sudden voltage collapse can halt production, damage equipment, and trigger cascading outages. Sensitivity-based prediction provides real-time, computationally efficient early warnings—critical for integrating variable-speed drives and renewable microgrids into mine electrical networks.

📘 Core Principles

Voltage stability hinges on the ability of a power system to maintain steady voltages after disturbances. Sensitivity methods linearize the power flow equations around an operating point: Δx = J⁻¹Δp, where J is the reduced Jacobian (excluding angle variables at PV buses). The most critical sensitivity is the minimum singular value (σ_min) of J — as σ_min → 0, the system approaches the voltage collapse point. Modal analysis decomposes J into left/right singular vectors to identify dominant instability modes and their spatial footprints (e.g., which buses or branches drive instability). Unlike time-domain simulation, this approach offers interpretable metrics without iterative convergence near singularity.

📐 Minimum Singular Value Sensitivity Index

The minimum singular value (σ_min) of the reduced power flow Jacobian quantifies distance to voltage collapse: smaller σ_min implies higher instability risk. It is computed via singular value decomposition (SVD) of the (n−1)×(n−1) reduced Jacobian matrix J_red, formed by eliminating rows/columns corresponding to slack and PV bus angles.

💡 Worked Example

Problem: A 9-bus mine grid (3 generators, 6 PQ buses) yields a reduced Jacobian J_red of size 6×6 after eliminating slack and PV angle variables. Its SVD yields singular values: [8.42, 5.17, 2.93, 1.05, 0.38, 0.072].

1. Step 1: Identify the smallest singular value: σ_min = 0.072

2. Step 2: Compare to threshold — IEEE Std 1204 recommends σ_min < 0.1 indicates high vulnerability

3. Step 3: Compute normalized index: L = 1/σ_min ≈ 13.9; L > 10 confirms proximity to collapse

Answer: The result is σ_min = 0.072, indicating imminent voltage instability — remedial action (e.g., reactive support or load shedding) is required within minutes.

🏗️ Real-World Application

At the BHP Olympic Dam expansion (South Australia), sensitivity-based voltage monitoring was deployed on the 275 kV mine intake substation. During commissioning of a new SAG mill (18 MW, PF=0.85 lagging), real-time σ_min dropped from 0.42 to 0.09 within 45 seconds. The system automatically triggered capacitor bank switching and adjusted STATCOM setpoints — preventing collapse and avoiding a 12-hour production stoppage. Post-event analysis confirmed the sensitivity model predicted the weak bus (Bus OD-07) and dominant reactive power path (Line OD-07–OD-03) with 94% accuracy.