Home / Courses / Probabilistic Design and Reliability / Reliability-Based Design

Probabilistic Design and Reliability | Module 9 of 10

Reliability-Based Design

Reliability-based design does not ask only whether a nominal constraint is satisfied. It asks whether the probability of violating the constraint is acceptable for the consequence.

01

Readiness check

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

  • Write a deterministic design constraint.
  • Compute a probability of failure from a normal margin.
  • Explain a target reliability in words.
  • Distinguish requirement, objective, and constraint.
  • Run an independent check on a design result.
0 or 1 weak itemsContinue with this module.
2 weak itemsReview constrained optimization in Optimization, Module 6.
3 or more weak itemsReview limit states in Module 3 before continuing.
02

The core idea

A reliability-based constraint limits failure probability, not only nominal stress or deflection. A target reliability has meaning only with a defined failure mode, mission, consequence, and model basis.

deterministic: g(x_nominal) > 0chance constraint: P[g(X) <= 0] <= P_targetnormal margin: beta = mu_g / sigma_g

Reliability-based design moves probability into the design decision. A deterministic constraint might require stress below allowable stress at nominal values. A probabilistic constraint might require P(stress > strength) below 0.001 for a specified mission. The reliability index beta is a useful normal-margin measure, but it is not a magic universal safety score. Introductory first-order reliability methods approximate the limit-state surface near the important failure point. This course keeps FORM and SORM at awareness level and focuses on defensible design workflow: define the failure event, choose target reliability from consequence and context, design for the chance constraint, then verify with an independent simulation or analytical check. The target reliability is not meaningful without consequence. A removable panel clip and a lifting hook do not deserve the same target.

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 reliability-based constraint limits failure probability, not only nominal stress or deflection. A target reliability has meaning only with a defined failure mode, mission, consequence, and model basis.
03

The skills, taught in order

Turn deterministic limits into chance constraints

Replace only the right constraints with probabilities and keep units visible.

Set targets from consequence

Higher consequence requires stronger evidence, lower allowable failure probability, and often independent review.

Use beta carefully

For normal margins, beta maps to failure probability. Outside that model, state what approximation is being used.

Design, then verify independently

A solver result should be checked by an independent Monte Carlo or analytical benchmark.

Report residual risk

Even accepted designs carry assumptions, exclusions, and remaining risk.

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

04

Worked example 1: shaft redesign for target failure probability

A shaft design has a normal margin with mean 32 MPa and standard deviation 16 MPa, so beta = 2. The target is P_f <= 0.001, approximately beta >= 3.09 for a normal margin. If standard deviation stays 16 MPa, what mean margin is needed?

  1. ProblemA shaft design has a normal margin with mean 32 MPa and standard deviation 16 MPa, so beta = 2. The target is P_f <= 0.001, approximately beta >= 3.09 for a normal margin. If standard deviation stays 16 MPa, what mean margin is needed?
  2. AssumptionsThe margin remains approximately normal and scatter is unchanged by redesign.
  3. Modelbeta_target = mu_g / sigma_gmu_g_needed = beta_target sigma_g
  4. Solvemu_g_needed = 3.09 x 16 = 49.4 MPa. The design needs mean margin increased from 32 MPa to about 49 MPa, or scatter reduced, or both.
  5. CheckIncreasing mean margin may require larger diameter, better material, lower load, or geometry change. Reducing scatter may require tighter tolerances or better process control.
  6. ConclusionThe design must gain about 17 MPa of mean margin if scatter is unchanged.
Result. The design must gain about 17 MPa of mean margin if scatter is unchanged.
05

Worked example 2: risk-informed target selection

Two failures have the same calculated P_f = 0.002. One is cosmetic cover cracking; the other is a lifting hook fracture. Should the same target reliability be used?

  1. ProblemTwo failures have the same calculated P_f = 0.002. One is cosmetic cover cracking; the other is a lifting hook fracture. Should the same target reliability be used?
  2. AssumptionsThe probability estimates are for the same mission length but consequences differ.
  3. Modelrisk = probability x consequence
  4. SolveNo. The lifting hook fracture has much higher consequence and needs a much lower target probability, stronger evidence, conservative assumptions, and likely standards or expert review.
  5. CheckProbability alone does not rank risk without consequence.
  6. ConclusionTarget reliability must be tied to failure mode and consequence.
Result. Target reliability must be tied to failure mode and consequence.
06

Python activity

Engineering question: Search candidate shaft diameters for a target Monte Carlo failure probability.

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 pi
from random import Random

rng = Random(31)
n = 150_000
torque_samples = [rng.gauss(420_000.0, 40_000.0) for _ in range(n)]
strength_samples = [rng.gauss(230.0, 16.0) for _ in range(n)]
target = 0.001

for step in range(11):
    d_nom = 28.0 + 0.5 * step
    failures = 0
    for torque, strength in zip(torque_samples, strength_samples):
        diameter = rng.gauss(d_nom, 0.25)
        tau = 16.0 * torque / (pi * diameter**3)
        if 0.58 * strength - tau <= 0.0:
            failures += 1
    pf = failures / n
    print(round(d_nom, 1), "mm Pf", round(pf, 5))
    if pf <= target:
        print("first candidate meeting target:", round(d_nom, 1), "mm")
        break
Numerical checks: Use a fresh independent simulation for the final selected diameter, ideally with a different seed and larger N.
Limitations: This is a teaching search. Formal reliability-based design may require FORM, variance reduction, standards, and review.
07

Misconceptions and diagnostics

MistakeDiagnostic questionCorrection
A target reliability has meaning without a defined failure mode and consequence.Ask: target for what event, mission, and consequence?Correct the assumption, then rerun the calculation or rewrite the report.
Beta is a universal safety score.Ask: what distribution and limit-state assumptions map beta to probability?Correct the assumption, then rerun the calculation or rewrite the report.
A probabilistic constraint replaces engineering judgement.Ask: who accepts the consequence and residual risk?Correct the assumption, then rerun the calculation or rewrite the report.
The optimizer's final design is automatically verified.Ask: what independent simulation or benchmark challenged it?Correct the assumption, then rerun the calculation or rewrite the report.
08

Practice ladder

Recognize

Rewrite 'stress must be below strength' as a chance constraint.

Solution guidance

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

Calculate

For beta target 3.0 and sigma_g = 12 MPa, find needed mean 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 report gives target reliability but no failure mode. Explain why it is incomplete.

Solution guidance

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

Design

Propose a reliability-based redesign workflow for a bolted bracket.

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 is a chance constraint?

A constraint on the probability of violating a limit state.

What is target reliability tied to?

Failure mode, mission, consequence, and accepted risk context.

What does beta equal for a normal margin?

Mean margin divided by margin standard deviation.

Why independently verify a final design?

To catch solver, model, sampling, or implementation errors.

What is residual risk?

Risk that remains after design decisions, assumptions, and mitigations.

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
Reliability-based design conceptsLe, Reliability-Based Mechanical Design, Volume 1
FORM and reliability index foundationsDer Kiureghian, Structural and System Reliability
Risk-informed engineering reliabilityHaldar and Mahadevan, Probability, Reliability, and Statistical Methods in Engineering Design