Reliability Analysis Skill
Component and system reliability prediction and analysis for electronic hardware.
Purpose
This skill provides comprehensive capabilities for predicting and analyzing the reliability of electronic components and systems. It supports industry-standard reliability methodologies, failure rate calculations, and life testing data analysis.
Capabilities
MTBF/MTTF Calculations
- Mean Time Between Failures (MTBF) for repairable systems
- Mean Time To Failure (MTTF) for non-repairable components
- Series and parallel system reliability modeling
- Redundancy calculations (active, standby, k-out-of-n)
- Mission reliability vs operational availability
Failure Rate Databases
- MIL-HDBK-217F failure rate predictions
- Telcordia SR-332 methodology
- FIDES reliability methodology
- IEC 62380 electronic component reliability
- Custom component database management
Derating Analysis
- Component stress ratio calculations
- Temperature derating curves
- Voltage and power derating
- Derating guideline compliance (NAVSEA, JPL, ESA)
- Stress analysis documentation
FMEA/FMECA Support
- Failure Mode and Effects Analysis facilitation
- Criticality analysis (CA) calculations
- Risk Priority Number (RPN) computation
- Severity, occurrence, detection ratings
- FMEA worksheet generation
- Action tracking and verification
Reliability Block Diagram Analysis
- RBD construction and visualization
- Series, parallel, and complex configurations
- Active and standby redundancy modeling
- Common cause failure analysis
- System reliability calculation
Fault Tree Analysis (FTA)
- Fault tree construction (AND, OR, k-of-n gates)
- Minimal cut set identification
- Top event probability calculation
- Importance measures (Birnbaum, Fussell-Vesely)
- Common cause failure modeling
Accelerated Life Testing Data Analysis
- Arrhenius model for temperature acceleration
- Eyring model for multi-stress acceleration
- Inverse power law for voltage/mechanical stress
- Acceleration factor calculation
- Life projection to use conditions
Weibull Distribution Fitting
- Two-parameter and three-parameter Weibull
- Maximum Likelihood Estimation (MLE)
- Probability plotting
- Goodness-of-fit testing
- Confidence interval estimation
- B-life calculations (B1, B10, B50)
Thermal Derating Curves
- Junction temperature estimation
- Thermal resistance modeling
- Safe operating area verification
- Thermal runaway analysis
- Heatsink selection guidance
Prerequisites
Installation
pip install numpy scipy pandas matplotlib reliability weibull
Optional Dependencies
# For advanced reliability modeling
pip install surpyval lifelines
# For report generation
pip install jinja2 openpyxl
Usage Patterns
MTBF Calculation with MIL-HDBK-217
import numpy as np
class MIL217Calculator:
"""MIL-HDBK-217F failure rate calculator"""
# Base failure rates (per 10^6 hours) - simplified examples
BASE_RATES = {
'resistor_film': 0.0037,
'capacitor_ceramic': 0.012,
'capacitor_electrolytic': 0.12,
'diode_general': 0.024,
'transistor_bipolar': 0.074,
'ic_digital': 0.16,
'ic_linear': 0.21,
'inductor': 0.0017,
'connector_pin': 0.00066,
'pcb_layer': 0.00042,
}
# Temperature factors (simplified)
@staticmethod
def temp_factor(temp_c: float, component_type: str) -> float:
if 'capacitor_electrolytic' in component_type:
return np.exp((temp_c - 25) / 15)
return np.exp((temp_c - 25) / 20)
# Environment factors
ENV_FACTORS = {
'ground_benign': 1.0,
'ground_fixed': 2.0,
'ground_mobile': 5.0,
'airborne_inhabited': 4.0,
'airborne_uninhabited': 8.0,
'space_flight': 0.5,
}
def calculate_component_fr(self, component_type: str, temp_c: float,
environment: str, quantity: int = 1) -> float:
"""Calculate failure rate for component type"""
base_rate = self.BASE_RATES.get(component_type, 0.1)
temp_factor = self.temp_factor(temp_c, component_type)
env_factor = self.ENV_FACTORS.get(environment, 2.0)
return base_rate * temp_factor * env_factor * quantity
def calculate_system_mtbf(self, components: list) -> dict:
"""Calculate system MTBF from component list"""
total_fr = sum(c['failure_rate'] for c in components)
mtbf = 1e6 / total_fr # Hours
return {
'total_failure_rate': total_fr,
'mtbf_hours': mtbf,
'mtbf_years': mtbf / 8760,
'components': components
}
# Example usage
calc = MIL217Calculator()
components = [
{'type': 'resistor_film', 'qty': 100, 'temp': 55},
{'type': 'capacitor_ceramic', 'qty': 50, 'temp': 55},
{'type': 'ic_digital', 'qty': 10, 'temp': 65},
]
for comp in components:
comp['failure_rate'] = calc.calculate_component_fr(
comp['type'], comp['temp'], 'ground_fixed', comp['qty']
)
result = calc.calculate_system_mtbf(components)
print(f"System MTBF: {result['mtbf_hours']:.0f} hours ({result['mtbf_years']:.1f} years)")
Weibull Analysis
from reliability.Fitters import Fit_Weibull_2P
from reliability.Probability_plotting import Weibull_probability_plot
import matplotlib.pyplot as plt
# Life test data (hours to failure)
failures = [1200, 1500, 1800, 2100, 2400, 2800, 3200, 3800, 4500, 5500]
censored = [6000, 6000, 6000] # Units still running at test end
# Fit Weibull distribution
fit = Fit_Weibull_2P(
failures=failures,
right_censored=censored,
show_probability_plot=False
)
print(f"Beta (shape): {fit.beta:.3f}")
print(f"Eta (scale): {fit.eta:.1f} hours")
print(f"B10 Life: {fit.distribution.quantile(0.1):.1f} hours")
print(f"B50 Life: {fit.distribution.quantile(0.5):.1f} hours")
print(f"Mean Life: {fit.distribution.mean:.1f} hours")
# Reliability at specific time
time = 2000 # hours
R_2000 = fit.distribution.SF(time)
print(f"Reliability at {time} hours: {R_2000:.4f} ({R_2000*100:.2f}%)")
Fault Tree Analysis
from typing import List, Dict
class FaultTreeNode:
def __init__(self, name: str, gate_type: str = None, probability: float = None):
self.name = name
self.gate_type = gate_type # 'AND', 'OR', 'VOTE'
self.probability = probability # For basic events
self.children: List['FaultTreeNode'] = []
self.k = None # For k-out-of-n voting gates
def add_child(self, child: 'FaultTreeNode'):
self.children.append(child)
def calculate_probability(self) -> float:
if self.probability is not None:
return self.probability
child_probs = [c.calculate_probability() for c in self.children]
if self.gate_type == 'AND':
result = 1.0
for p in child_probs:
result *= p
return result
elif self.gate_type == 'OR':
result = 1.0
for p in child_probs:
result *= (1 - p)
return 1 - result
elif self.gate_type == 'VOTE':
# k-out-of-n gate
from itertools import combinations
from functools import reduce
import operator
n = len(child_probs)
k = self.k
prob = 0
for i in range(k, n + 1):
for combo in combinations(range(n), i):
term = 1.0
for j in range(n):
if j in combo:
term *= child_probs[j]
else:
term *= (1 - child_probs[j])
prob += term
return prob
# Example: Power supply failure fault tree
top = FaultTreeNode("Power Supply Fails", "OR")
primary_fails = FaultTreeNode("Primary Supply Fails", "AND")
primary_fails.add_child(FaultTreeNode("AC Power Loss", probability=0.01))
primary_fails.add_child(FaultTreeNode("UPS Fails", probability=0.001))
backup_fails = FaultTreeNode("Backup Supply Fails", probability=0.005)
top.add_child(primary_fails)
top.add_child(backup_fails)
system_probability = top.calculate_probability()
print(f"Top event probability: {system_probability:.6f}")
Accelerated Life Test Analysis
import numpy as np
from scipy.optimize import curve_fit
class ArrheniusModel:
"""Arrhenius acceleration model for temperature stress"""
def __init__(self):
self.activation_energy = None # eV
self.k_boltzmann = 8.617e-5 # eV/K
def acceleration_factor(self, temp_test: float, temp_use: float,
activation_energy: float) -> float:
"""Calculate acceleration factor between test and use conditions"""
temp_test_k = temp_test + 273.15
temp_use_k = temp_use + 273.15
af = np.exp((activation_energy / self.k_boltzmann) *
(1/temp_use_k - 1/temp_test_k))
return af
def estimate_activation_energy(self, temps: List[float],
failure_rates: List[float]) -> float:
"""Estimate activation energy from multi-temperature test data"""
temps_k = [t + 273.15 for t in temps]
inv_temps = [1/t for t in temps_k]
ln_rates = [np.log(r) for r in failure_rates]
# Linear regression: ln(rate) = A + Ea/(k*T)
slope, intercept = np.polyfit(inv_temps, ln_rates, 1)
self.activation_energy = slope * self.k_boltzmann
return self.activation_energy
# Example usage
model = ArrheniusModel()
# Multi-temperature test results
test_temps = [85, 105, 125] # Celsius
failure_rates = [0.001, 0.005, 0.02] # failures per 1000 hours
ea = model.estimate_activation_energy(test_temps, failure_rates)
print(f"Estimated activation energy: {ea:.2f} eV")
# Project to use conditions
af = model.acceleration_factor(125, 55, ea)
use_life = 2000 * af # If 2000 hours at 125C
print(f"Acceleration factor: {af:.1f}x")
print(f"Projected life at 55C: {use_life:.0f} hours")
Usage Guidelines
When to Use This Skill
- New product reliability predictions
- Design for reliability (DfR) activities
- Warranty cost projections
- FMEA and FMECA development
- Life test planning and analysis
- Field failure analysis support
Best Practices
- Use appropriate failure rate models for the application environment
- Consider temperature derating for all components
- Document all assumptions in reliability predictions
- Validate predictions with field data when available
- Update failure rates based on actual performance
- Include manufacturing defects in early-life reliability models
Process Integration
- ee-environmental-testing (life test analysis)
- ee-hardware-validation (reliability verification)
- ee-dfm-review (reliability design reviews)
Dependencies
- reliability: Python reliability engineering library
- scipy: Statistical analysis
- numpy: Numerical computations
References
- MIL-HDBK-217F Reliability Prediction
- FIDES Reliability Methodology Guide
- IEEE 1413 Methodology for Reliability Prediction
- SAE JA1000 Reliability Program Standard