pystra.decision.targets.TargetReliability#

Use it: User guide · Worked example · Theory

class TargetReliability(pf, beta, method, k1=None, cost_class=None, variability=None, design=None, objective=None, converged=None, message='', source=None, metadata=dict())[source]#

Bases: object

Target failure probability and reliability index from a calibration.

A single result type for every target-reliability route in this module: the rounded LQI table lookup (LQI.lookup_target()), the LQI marginal optimization (LQI.derive_target()), and the Rackwitz/Steenbergen code-calibration model (RackwitzTargetModel.calibrate()). Only pf, beta and method are always populated; fields that do not apply to a given route are left as None.

Parameters:

  • pf (float) – Annual failure probability or rate at the target.

  • beta (float) – Reliability index corresponding to pf.

  • method (str) – Route that produced the target, e.g. "lqi-table", "LQI marginal" or "Rackwitz/Steenbergen".

  • k1 (float | None) – LQI safety cost ratio, when the target comes from the LQI route.

  • cost_class (str | None) – Discrete cost-class label from the rounded LQI table.

  • variability (str | None) – Variability class used for the rounded LQI table.

  • design (float | None) – Optimizing design parameter for calculated targets (the mean resistance-to-load ratio in the Rackwitz examples).

  • objective (float | None) – Objective value at the optimizing design.

  • converged (bool | None) – Whether the underlying optimization converged.

  • message (str) – Solver message for calculated targets.

  • source (str | None) – Literature source carried with a looked-up target.

  • metadata (Mapping[str, Any]) – Additional model parameters or labels.

Methods

for_period

Return the target converted to a reference period of years.

to_dict

Return the populated scalar quantities as a dictionary.
to_dict()[source]#

Return the populated scalar quantities as a dictionary.

Return type:

dict

for_period(years, dependence_interval=1.0)[source]#

Return the target converted to a reference period of years.

The annual failure probability pf is compounded over years, assuming the governing maximum renews every dependence_interval years. dependence_interval=1 treats the annual maxima as independent (the most onerous case), while dependence_interval=years treats the period as fully dependent (a single renewal) and leaves the annual pf unchanged.

The conversion uses pf directly rather than re-deriving an annual probability from beta; for rounded table targets the two are not an exact pair, and compounding pf is the correct quantity.

Parameters:

  • years (float) – Reference period in years.

  • dependence_interval (float) – Renewal interval of the governing maximum, in years; must lie in (0, years].

Returns:

A new target at the reference period, with the original method and class labels retained and reference_period_years / dependence_interval recorded in metadata.

Return type:

TargetReliability