Second-Order Reliability Methods

The SORM is an improvement over the FORM by accounting for the curved nature of the failure boundary given by $g(\vec{X}) = 0$ around the design point $\vec{x}^{*}$; thus, providing a better approximation of the probability of failure $P_{f}$.

Curve-Fitting Method

The CF method fits a hyper-paraboloid surface with a vertex at the design point $\vec{x}^{*}$ and the principal curvatures matching the principal curvatures of the failure boundary given by $g(\vec{X}) = 0$ at that point. The probabilities $P_{f}$ of failure are estimated using Hohenbichler and Rackwitz (1988) and Breitung (1984) approximations of the exact solution provided by Tvedt (1990). The calculated probabilities of failure $P_{f}$ are then used to estimate the generalized reliability indices $\beta$, which account for the curved nature of the failure boundary given by $g(\vec{X}) = 0$ around the design point $\vec{x}^{*}$.

Point-Fitting Method

The PF method fits a series of hyper-semiparaboloid surfaces with a vertex at the design point $\vec{x}^{*}$. The principal curvatures of each surface are estimated using fitting points found at the intersections of a hyper-cylinder with axis coinciding with the design point $\vec{u}^{*}$ and the failure boundary given by $g(\vec{U}) = 0$ in $U$-space. The PF method provides a better estimate of the probability of failure $P_{f}$ than the CF method since it provides a better approximation of highly non-linear failure boundaries given by $g(\vec{X}) = 0$ that are unsymmetrical about the design point $\vec{x}^{*}$.

Note

A great description of both methods can be found in Der Kiureghian (2022).

API

Fortuna.solve — Method

solve(Problem::ReliabilityProblem, AnalysisMethod::SORM; FORMSolution::Union{Nothing, HLRFCache, iHLRFCache} = nothing, FORMConfig::FORM = FORM(), diff::Symbol = :automatic)

Function used to solve reliability problems using Second-Order Reliability Method (SORM).
If diff is:

  • :automatic, then the function will use automatic differentiation to compute gradients, jacobians, etc.
  • :numeric, then the function will use numeric differentiation to compute gradients, jacobians, etc.
Fortuna.SORM — Type
SORM <: AbstractReliabililyAnalysisMethod

Type used to perform reliability analysis using Second-Order Reliability Method (SORM).

  • Submethod::Fortuna.SORMSubmethod
Fortuna.CF — Type
CF <: SORMSubmethod

Type used to perform reliability analysis using Curve-Fitting (CF) method.

  • ϵ::Real: Step size used to compute the Hessian at the design point in $U$-space
Fortuna.CFCache — Type
CFCache

Type used to perform reliability analysis using Point-Fitting (PF) method.

  • FORMSolution::Union{HLRFCache, RFCache, iHLRFCache}: Results of reliability analysis performed using First-Order Reliability Method (FORM)

  • β₂::Vector{Union{Missing, Float64}}: Generalized reliability indices $\beta$

  • PoF₂::Vector{Union{Missing, Float64}}: Probabilities of failure $P_{f}$

  • κ::Vector{Float64}: Principal curvatures $\kappa$

Fortuna.PF — Type
PF <: SORMSubmethod

Type used to perform reliability analysis using Point-Fitting (PF) method.

Fortuna.PFCache — Type
PFCache

Type used to perform reliability analysis using Point-Fitting (PF) method.

  • FORMSolution::iHLRFCache: Results of reliability analysis performed using First-Order Reliability Method (FORM)

  • β₂::Vector{Union{Missing, Float64}}: Generalized reliability index $\beta$

  • PoF₂::Vector{Union{Missing, Float64}}: Probabilities of failure $P_{f}$

  • FittingPoints⁻::Matrix{Float64}: Fitting points on the negative side of the hyper-cylinder

  • FittingPoints⁺::Matrix{Float64}: Fitting points on the positive side of the hyper-cylinder

  • κ₁::Matrix{Float64}: Principal curvatures on the negative and positive sides

  • κ₂::Matrix{Float64}: Principal curvatures of each hyper-semiparabola