Probabilistic Design and Reliability | Module 3 of 10

Performance Functions and Limit States

The limit state is the line between acceptable and unacceptable performance. Reliability work becomes clear once that boundary is written as g(X) = 0.

01

Readiness check

Tick only what you can do closed-notes before using this module for design decisions.

  • Compute axial stress as force divided by area.
  • Explain yield as a material limit.
  • Read a normal distribution probability for a margin.
  • Distinguish resistance from load effect.
  • State a failure event as an inequality.
0 or 1 weak itemsContinue with this module.
2 weak itemsReview axial stress in Mechanics of Materials, Module 2.
3 or more weak itemsReview stress and material strength in Mechanics of Materials, Module 1.
02

The core idea

A limit-state function g(X) is positive in the safe region, zero on the boundary, and negative in the failure region. For stress-strength interference, g = strength - stress.

g(X) = R(X) - S(X)safe: g(X) > 0P_f = P[g(X) <= 0]

Limit states turn a vague design concern into a computable event. If a pin yields when stress exceeds yield strength, the performance function can be g = S_y - P/A. If a beam fails a deflection requirement, the function can be g = delta_allow - delta(P,E,I). If a fastener loses clamp load, g can be remaining preload minus required clamp load. The safe region is not a subjective feeling; it is the set of inputs for which g is positive. The failure region is where g is zero or negative. In stress-strength interference, two distributions overlap. The amount of overlap is failure probability, not a mysterious reliability factor.

The skill works when: the failure event, units, assumptions, distributions, and checks are visible.
The skill breaks down when: a probability is reported without the mechanical model and evidence boundary.
The concept. A limit-state function g(X) is positive in the safe region, zero on the boundary, and negative in the failure region. For stress-strength interference, g = strength - stress.
03

The skills, taught in order

Define the quantity of interest

State what must remain acceptable: stress, deflection, pressure, temperature, preload, life, or clearance.

Choose sign convention

Use g > 0 as safe and g <= 0 as failure throughout the analysis.

Map variables to mechanics

Write the mechanical equation before adding probability.

Use analytical margins when possible

If resistance and load effect are independent normal variables, the margin is normal and failure probability is direct.

Interpret the result physically

A probability without the failure event, units, and assumptions is not useful engineering evidence.

Engineering connection: use these skills to turn variability into a design decision, not just a plot.

04

Worked example 1: stress-strength interference for an axial member

An axial tie has area A = 250 mm^2. Load P is normal with mean 80 kN and standard deviation 8 kN. Yield strength S_y is normal with mean 380 MPa and standard deviation 25 MPa. Estimate failure probability for yielding using g = S_y - P/A.

  1. ProblemAn axial tie has area A = 250 mm^2. Load P is normal with mean 80 kN and standard deviation 8 kN. Yield strength S_y is normal with mean 380 MPa and standard deviation 25 MPa. Estimate failure probability for yielding using g = S_y - P/A.
  2. AssumptionsLoad and strength are independent. Area is treated as fixed. Units use 1 MPa = 1 N/mm^2.
  3. Modelsigma = P / Amu_sigma = 80000 / 250 = 320 MPasigma_sigma = 8000 / 250 = 32 MPag = S_y - sigma
  4. SolveMean margin is mu_g = 380 - 320 = 60 MPa. Margin standard deviation is sigma_g = sqrt(25^2 + 32^2) = 40.6 MPa. beta = 60/40.6 = 1.48. Therefore P_f = Phi(-1.48) = 0.069.
  5. CheckThe nominal factor is 380/320 = 1.19, so a nontrivial failure probability is plausible. The high load scatter drives much of the overlap.
  6. ConclusionEstimated yield failure probability is about 6.9 percent under the stated model.
Result. Estimated yield failure probability is about 6.9 percent under the stated model.
05

Worked example 2: beam deflection as a serviceability limit state

A beam is acceptable if tip deflection stays below 5 mm. A preliminary model predicts deflection as normal with mean 3.8 mm and standard deviation 0.6 mm. Estimate probability of exceeding the limit.

  1. ProblemA beam is acceptable if tip deflection stays below 5 mm. A preliminary model predicts deflection as normal with mean 3.8 mm and standard deviation 0.6 mm. Estimate probability of exceeding the limit.
  2. AssumptionsThe deflection model and uncertainty are credible near the service limit. Failure means serviceability failure, not fracture.
  3. Modelg = delta_allow - deltamu_g = 5.0 - 3.8sigma_g = 0.6P_f = Phi(-mu_g / sigma_g)
  4. Solvemu_g = 1.2 mm and beta = 1.2/0.6 = 2.0. P_f = Phi(-2.0) = 0.0228.
  5. CheckThis is a serviceability probability, not a life-safety claim. Consequence changes the target probability.
  6. ConclusionThe predicted probability of exceeding the 5 mm deflection limit is about 2.3 percent.
Result. The predicted probability of exceeding the 5 mm deflection limit is about 2.3 percent.
06

Python activity

Engineering question: Estimate failure probability for an axial member by analytical margin and Monte Carlo.

Assumptions and units: Use the units shown in the code comments. Keep the random seed fixed while learning so the result is reproducible.

from math import erf, sqrt
from random import Random

def phi(x):
    return 0.5 * (1.0 + erf(x / sqrt(2.0)))

rng = Random(7)
n = 200_000
area = 250.0
failures = 0

for _ in range(n):
    strength = rng.gauss(380.0, 25.0)       # MPa
    load = rng.gauss(80_000.0, 8_000.0)     # N
    stress = load / area                    # MPa
    if strength - stress <= 0.0:
        failures += 1

pf_mc = failures / n
mu_g = 380.0 - 80_000.0 / area
sd_g = (25.0**2 + (8_000.0 / area)**2) ** 0.5
pf_formula = phi(-mu_g / sd_g)

print("Monte Carlo Pf", pf_mc)
print("Analytical Pf", pf_formula)
Numerical checks: The Monte Carlo estimate should be close to the analytical result, within sampling error.
Limitations: The calculation ignores area tolerance, stress concentration, and nonnormal tails.
07

Misconceptions and diagnostics

MistakeDiagnostic questionCorrection
The limit state is just a formula from a textbook.Ask: what real unacceptable outcome does g <= 0 represent?Correct the assumption, then rerun the calculation or rewrite the report.
Stress-strength reliability always means yield failure.Ask: is the governing failure mode yield, fracture, fatigue, buckling, wear, or service deflection?Correct the assumption, then rerun the calculation or rewrite the report.
The mean margin is the reliability.Ask: how much scatter surrounds that mean margin?Correct the assumption, then rerun the calculation or rewrite the report.
If beta is positive, the design is safe.Ask: positive by how many standard deviations and for what consequence?Correct the assumption, then rerun the calculation or rewrite the report.
08

Practice ladder

Recognize

For bearing load capacity C and applied equivalent load P, write a safe-sign performance function.

Solution guidance

Write the governing equation, state units and assumptions, then compare the answer with the relevant limit state or design decision.

Calculate

If strength is 300 MPa and stress is 210 MPa at nominal values, compute nominal margin.

Solution guidance

Write the governing equation, state units and assumptions, then compare the answer with the relevant limit state or design decision.

Diagnose

A model uses g = stress - strength but says failure is g <= 0. Find the sign error.

Solution guidance

Write the governing equation, state units and assumptions, then compare the answer with the relevant limit state or design decision.

Design

Write two different limit states for a bolted joint, one for yielding and one for loss of clamp function.

Solution guidance

Write the governing equation, state units and assumptions, then compare the answer with the relevant limit state or design decision.

09

Working with AI, and proving it yourself

Useful AI support

Ask for alternative explanations of terminology, candidate code tests, or a checklist of assumptions to verify.
Ask it to challenge a derivation, then prove the result with units, limiting cases, and an independent calculation.
Do not let AI invent material distributions, fabricate failure data, claim standards compliance, or make final safety decisions.

Portfolio evidence

Create one short reliability note for this module. It must include the engineering question, assumptions, input definitions, equations, calculation, checks, interpretation, and limitations.

Verification required: dimensional analysis, a benchmark case, convergence or sample-size check, and a plain-language residual-risk statement.
10

Retrieval and spaced review

Closed notes. Answer out loud, then reveal.

What sign convention does this course use for g?

g > 0 is safe, g <= 0 is failure.

What is stress-strength interference?

The overlap between stress and strength distributions.

What is a safe region?

The set of input values for which the performance function is positive.

What does beta represent in a normal margin model?

Mean margin divided by standard deviation of margin.

Why must units be checked before probability?

Probability cannot fix a mechanically inconsistent model.

TodayFinish the review and first two practice levels.
+1 dayRebuild one worked example from a blank page.
+3 daysChange one assumption and predict the effect.
+7 daysConnect this module to the capstone bracket.
+30 daysReuse the method in a new mechanical component.
11

References and source discipline

This course synthesizes original MechCompass explanations from established reliability engineering sources. Confirm current editions and standards before using them for formal design approval.

Topic in this moduleReference direction
Limit-state reliability methodsHaldar and Mahadevan, Probability, Reliability, and Statistical Methods in Engineering Design
Stress-strength interference and mechanical reliabilityLe, Reliability-Based Mechanical Design, Volume 1
Reliability index conceptsDer Kiureghian, Structural and System Reliability