🎓 Lesson 6 D4

Gauss-Seidel Method: Algorithm Walkthrough & Convergence Behavior

The Gauss-Seidel method is a step-by-step way to solve a set of equations by repeatedly improving guesses until the answers stop changing much.

🎯 Learning Objectives

  • Calculate successive iterations of voltage magnitudes and angles for a 3-bus power system using Gauss-Seidel
  • Analyze convergence behavior by monitoring relative error and iteration count
  • Explain how initial guess selection and system conditioning affect convergence rate and stability
  • Apply relaxation factors to improve convergence in ill-conditioned load flow cases
  • Compare Gauss-Seidel performance against Newton-Raphson for small-scale distribution networks

📖 Why This Matters

In power system operation, engineers must know voltages and power flows at every bus—this is the 'load flow' problem. While modern EMS use Newton-Raphson, Gauss-Seidel remains foundational: it’s intuitive, memory-efficient, and teaches core concepts like iteration, convergence, and sensitivity to network topology. Understanding it helps diagnose why real-world solvers diverge—and how to fix them.

📘 Core Principles

Gauss-Seidel treats each bus equation as a function of all other bus voltages. Starting from an initial estimate (e.g., flat start: V_i = 1.0∠0°), it solves for V₁ using known V₂, V₃,… then immediately uses that updated V₁ to solve for V₂, and so on. This sequential update accelerates convergence over Jacobi—but introduces dependency on ordering and matrix properties. Convergence hinges on diagonal dominance: the magnitude of each diagonal element must exceed the sum of off-diagonal magnitudes in its row—a condition often met in well-designed transmission networks but violated in weakly meshed or high-R/X ratio distribution systems.

📐 Voltage Update Formula

For PQ and PV buses in per-unit, the Gauss-Seidel voltage update isolates V_i using nodal admittance matrix elements. The formula leverages known injected power and neighboring voltages to iteratively refine V_i.

Gauss-Seidel Bus Voltage Update (PQ bus)

V_i^{(k+1)} = \frac{1}{Y_{ii}} \left[ \frac{S_i^*}{V_i^{(k)*}} - \sum_{j=1, j\neq i}^n Y_{ij} V_j^{(k+1 \text{ or } k)} \right]

Computes next voltage estimate for bus i using most recent values for buses 1 to i−1 and previous values for buses i+1 to n.

Variables:
SymbolNameUnitDescription
V_i^{(k+1)} Updated voltage at bus i pu (per-unit) Complex voltage phasor after (k+1)th iteration
S_i^* Complex conjugate of net injected power pu S_i = P_i + jQ_i; negative for loads, positive for generation
Y_{ii} Self-admittance of bus i siemens (pu) Sum of all line admittances connected to bus i
Y_{ij} Mutual admittance between bus i and j siemens (pu) Negative of line admittance between i and j
Typical Ranges:
Transmission system (345 kV): 10–50 pu
Distribution feeder (11 kV): 0.5–5 pu

💡 Worked Example

Problem: Given a 3-bus system with Y_bus (per-unit): Y₁₁ = 5.0−j20.0, Y₁₂ = −1.0+j5.0, Y₁₃ = −4.0+j15.0; P₂ = −0.3 pu, Q₂ = −0.2 pu; V₁ = 1.0∠0° (slack), V₂⁽⁰⁾ = 1.0∠0°, V₃⁽⁰⁾ = 1.0∠0°. Compute V₂⁽¹⁾.
1. Step 1: Use PQ bus formula: V₂⁽¹⁾ = (1/Y₂₂) × [(P₂ − jQ₂)/V₂⁽⁰⁾* − Σⱼ≠₂ Y₂ⱼ Vⱼ⁽⁰⁾]
2. Step 2: Compute conjugate: V₂⁽⁰⁾* = 1.0∠0° → (P₂ − jQ₂)/V₂⁽⁰⁾* = −0.3 + j0.2
3. Step 3: Sum off-diagonal terms: Y₂₁V₁ + Y₂₃V₃ = (−1.0+j5.0)(1.0∠0°) + (−2.0+j8.0)(1.0∠0°) = −3.0 + j13.0
4. Step 4: Numerator = (−0.3 + j0.2) − (−3.0 + j13.0) = 2.7 − j12.8
5. Step 5: Assuming Y₂₂ = 3.0−j18.0 → 1/Y₂₂ ≈ 0.0164 + j0.0982 (via complex division)
6. Step 6: V₂⁽¹⁾ ≈ (0.0164 + j0.0982)(2.7 − j12.8) ≈ 1.297 − j0.086 → |V₂⁽¹⁾| ≈ 1.30 pu, ∠ ≈ −3.8°
Answer: V₂⁽¹⁾ ≈ 1.30∠−3.8° pu — magnitude exceeds typical generator voltage limits (0.95–1.05 pu), signaling need for reactive support or algorithmic damping.

🏗️ Real-World Application

In 2019, a rural microgrid in Rajasthan (India) experienced load flow divergence during monsoon due to increased line reactance and weak interconnection. Engineers reverted to Gauss-Seidel with acceleration factor α = 1.2 to stabilize convergence—diagnosing the issue as low X/R ratio (<2) degrading diagonal dominance. Post-analysis, they added shunt capacitors at key nodes, restoring convergence in <8 iterations—demonstrating how Gauss-Seidel serves both as solver and diagnostic tool.

📚 References