Skip to content

Design Problem Authoring

This section covers problem authoring for the design command. Design problems extend the base problem dictionary with DesignProblem keys and can follow two distinct workflows:

  1. Inverse FORM: compute partial factors/design point values that satisfy a target reliability (TargetBeta or TargetPf).
  2. Objective optimization: compute design variable values that minimize an objective while satisfying reliability constraints.

When To Use Which Design Type

  • Use Inverse FORM when code calibration, load/resistance factor calibration, or characteristic-value back-calculation is the goal.
  • Use objective optimization when cost/weight/performance minimization is the goal and reliability is one of the constraints.

Design Documentation Layout

  • problem-authoring/design/design-keys.md: full DesignProblem dictionary reference and validation rules.
  • problem-authoring/design/inverse-form.md: worked Inverse FORM examples (AT625 single-case and AT624 multi-case).
  • problem-authoring/design/optimization.md: worked objective optimization example (Sorensen81Problem.py) including additional optimization keys.

Notes On Cases

  • A design problem can define a single default case or multiple named cases.
  • This applies to both workflows: Inverse FORM and objective optimization.
  • If DesignProblem.cases is omitted, Reliafy creates a default case from the top-level StochasticVariables definition.
  • Multi-case definitions are useful when target reliability remains fixed but statistical definitions or design context vary per scenario.