Lifetime Modeling of SiC Power Modules in Automotive Traction Inverters (Python Implementation)

---

📋 Overview

This repository contains the comprehensive Python implementation for predicting thermal fatigue lifetime in Silicon Carbide (SiC) Power MOSFETs used in high-performance automotive traction inverters.

The work successfully combines:

• ⚡ Software modeling (Python tool with MATLAB thesis model)
• 🔬 Experimental characterization methods
• 📊 Industry-standard compliance (ECPE AQG 324)

---

🎯 Motivation and Context

The automotive industry's transition to high-performance Electric Vehicles (EVs) demands:

• Smaller, lighter, more powerful components
• Silicon Carbide (SiC) technology for superior performance
• Long-term reliability under extreme operating conditions

---

🚀 What This Tool Does

Convert Driving Cycles → Lifetime Prediction

Electric Motor Profile (Speed, Torque, Time)
              ↓
    [Electro-Thermal Model]
              ↓
┌─────────────────────────────────────┐
│  Coupled Iterative Solution Loop:   │
│                                     │
│  T_j(i-1) → P_loss(i) → ΔT → T_j(i) │
│      ↑                         ↓    │
│      └─────────────────────────┘    │
│                                     │
│  Converges when thermal stability   │
│  is reached at each time step       │
└─────────────────────────────────────┘
              ↓
    Junction Temperature (T_j)
              ↓
    [Rainflow Cycle Counting]
              ↓
     Thermal Stress Cycles
              ↓
    [Bayerer Model - CIPS08]
    ΔT_j, T_j_max, t_on → N_f
              ↓
   Cumulative Damage (Miner's Rule)
              ↓
 ✓ Lifetime Consumption: X.XX%
 ✓ Extrapolated Life: XXXX hours
 ✓ AQG 324: PASS/FAIL

Three Mission Profiles

The tool analyzes three standardized automotive scenarios with progressive intensity:

Scenario	Speed Range	Torque Range	Thermal Stress	Use Case
Urban	0-8000 RPM	10-50 Nm	Low	City driving, stop-and-go
Highway	5000-14000 RPM	20-320 Nm	Medium	Highway cruise, overtaking
Performance	0-15000 RPM	20-450 Nm	High	Track driving, max acceleration

---

📦 Installation

Prerequisites

Python >= 3.7
pip >= 20.0

Quick Install

# Clone repository
git clone https://github.com/matteototaro/sic-thermal-lifetime.git
cd sic-thermal-lifetime

# Install dependencies
pip install -r requirements.txt

# Run the tool
python thermal_lifetime_prediction.py

Dependencies

numpy>=1.19.0
pandas>=1.0.0
matplotlib>=3.2.0
scipy>=1.5.0
tqdm>=4.50.0  # Optional, for progress bars

---

📊 Usage

Basic Execution

Simply run the script - it will automatically analyze all three scenarios:

python thermal_lifetime_prediction.py

The script will:

1. Load pre-generated mission profiles from CSV files (in profiles/ folder)
2. Perform coupled electro-thermal simulation
3. Extract thermal cycles using rainflow counting
4. Calculate cumulative damage and lifetime
5. Generate comprehensive visualization

Output

The tool generates:

1. Console Output

================================================================================
  LIFETIME CONSUMPTION: 0.012500%
================================================================================
  Extrapolated: 8000 hours (0.9 years)
  AQG 324: 0.13 / 1000 PC cycles
  ✓ PASS - Margin: 999.87 cycles (99.99%)

2. Comprehensive Visualization

A publication-quality comparison plot (docs/example_output.png) showing:

• Row 1: Motor Speed profiles (all 3 scenarios side-by-side)
• Row 2: Motor Torque profiles (all 3 scenarios)
• Row 3: Junction Temperature evolution (all 3 scenarios)
• Row 4: Comparative summary table with all key metrics

3. Results Dictionary

results = run_lifetime_test_all_scenarios()

# Access individual scenario data
profile, T_j, P_loss, cycles, damage_df, total_damage = results['urban']
profile, T_j, P_loss, cycles, damage_df, total_damage = results['highway']
profile, T_j, P_loss, cycles, damage_df, total_damage = results['performance']

---

🔬 Technical Methodology

1. Electro-Thermal Coupling

The tool implements proper feedback between electrical and thermal domains through an iterative process at each time step:

T_j(i-1) → P_loss(i) → Thermal Network → T_j(i)

Power Loss Components:

• Switching losses: E_on(T_j) × f_sw
• Conduction losses: I_rms² × R_DS_on(T_j)
• Diode losses: V_F × I_avg

All temperature-dependent parameters are updated at each iteration based on the previous junction temperature.

2. Thermal Network

4-layer RC network using Explicit Euler integration:

C_th × dT/dt = Q_in - Q_out
T(t+Δt) = T(t) + Δt × dT/dt

Stability condition: Δt < 2 × min(R_th × C_th)

3. Rainflow Cycle Counting

Implements ASTM E1049-85 standard:

• Extracts fatigue-relevant thermal cycles
• Tracks temperature swing (ΔT), mean temperature, and dwell time
• Converts continuous temperature history into discrete damage events

4. Lifetime Prediction - Bayerer Model (CIPS08)

Power Cycling Lifetime Model (Bayerer et al., 2008):

N_f = K × (ΔT_j)^β₁ × exp(β₂/T_j_max) × t_on^β₃

Where:

• β₁ = -3.483 (Coffin-Manson: thermal strain dependency)
• β₂ = 1917 K (Arrhenius: temperature activation energy)
• β₃ = -0.438 (Time-dependent creep/diffusion effects)
• ΔT_j: Junction temperature swing [K]
• T_j_max: Maximum junction temperature [K]
• t_on: Heating time / power-on duration [s]
• K: Technology-dependent constant calibrated from test data

Note: The original Bayerer (CIPS08) model includes a fourth parameter (β₄) for current per bond wire. However, for a fixed module technology and specific application, this parameter becomes a constant absorbed into K. Therefore, this implementation uses the simplified three-parameter form commonly applied in automotive qualification standards.

Acceleration Factor (without K, comparing field to test conditions):

The acceleration factor relates field conditions to test conditions:

AF = (ΔT_field / ΔT_test)^β₁ × exp[β₂ × (1/T_max_field - 1/T_max_test)] × (t_on_field / t_on_test)^β₃

N_f_field = N_f_test × AF

This allows prediction of lifetime under actual operating conditions based on standardized qualification test results.

Miner's Rule (Cumulative Damage):

D_total = Σ(n_i / N_f_i)

Failure predicted when D_total ≥ 1.0 (100% life consumed)

⚠️ NOTE ON CONFIDENTIALITY

This work was initially developed on MATLAB in collaboration with Ferrari S.p.A. for my Master's Thesis.

Due to intellectual property (IP) considerations:

• ✅ Full thesis documentation is publicly available
• ✅ Python demonstration tool with generic parameters is provided
• ❌ Production MATLAB model with Ferrari-specific data remains confidential

The Python tool uses publicly available Infineon datasheet parameters to demonstrate the complete methodology while respecting confidentiality requirements.

---

📄 Accessing the Thesis

📖 Full Thesis PDF: Totaro_Matteo_MastersThesis.pdf

The thesis includes:

• Comprehensive literature review on SiC power modules
• Detailed mathematical derivation of lifetime models
• Experimental test setup and procedures
• Results validation and sensitivity analysis
• Future work and recommendations

📝 Citation

If you use this work in your research, please cite:

@mastersthesis{totaro2022sic,
  author  = {Totaro, Matteo},
  title   = {Lifetime Modeling of SiC Power Modules in Automotive Traction Inverters},
  school  = {ALMA MATER STUDIORUM - Università di Bologna},
  year    = {2022},
  type    = {Master's Thesis},
  note    = {In collaboration with Ferrari S.p.A.}
}

---

📖 References

1. ECPE AQG 324 - Qualification of Automotive Electronic Components
2. Infineon FS03MR12A6MA1B Datasheet - 1200V 310A SiC Power Module
3. ASTM E1049-85 - Standard Practices for Cycle Counting in Fatigue Analysis
4. Bayerer, R., Herrmann, T., Licht, T., Lutz, J., & Feller, M. (2008) - "Model for Power Cycling lifetime of IGBT Modules - various factors influencing lifetime", 5th International Conference on Integrated Power Electronics Systems (CIPS), Nuremberg, Germany, pp. 1-6
5. Miner (1945) - Cumulative Damage in Fatigue, Journal of Applied Mechanics
6. IEC 60749-34 - Power Cycling Test to Failure for Power Semiconductor Devices
7. Held, M., et al. (1997) - "Fast power cycling test of IGBT modules in traction application", LESIT Project

---

🔧 Contact

Matteo Totaro
Email: tmatteos@gmail.com
Website: mtotaro.com
LinkedIn: linkedin.com/in/m-totaro

---

📜 License

This project is licensed under the MIT License - see the LICENSE file for details.