Repairable-Item Uptime and Steady-State Availability Worksheet
Estimate operating-time and restoration measures from one documented observation window,or evaluate a two-state steady-state model from documented mean up and down times. The worksheet does not turn a reciprocal mean into an instantaneous hazard rate.
Define the data basis
Use one repairable item or a documented homogeneous population. Define exactly which intervals are up,down,externally disabled,preventive,administrative or excluded;different boundaries produce different availability measures.
Observed up-time fraction
Steady-state availability model
MTBF / mean operating time
Mean time to restoration
Observed failure frequency
Included total time
Failure/restoration counts
Calculation basis
State boundary
Equations and interpretation
| Основа | Equation | Значение |
|---|---|---|
| Aggregated observation | up fraction=U/(U+D) MTBF estimate=U/Nf MTTR estimate=D/Nr frequency=Nf/U | Descriptive estimates for the entered window. Zero failures yields no finite MTBF point estimate;an open/censored restoration is not a completed restoration. |
| Documented two-state means | Ass=μU/(μU+μD) reciprocal mean=1/μU | Steady-state/asymptotic availability only under an applicable alternating up/down renewal model with finite means. |
When the observation contains equal positive failure and restoration counts and the same complete cycles,substitution makes (U/N)/(U/N+D/N)=U/(U+D). The equality is not asserted when counts or state boundaries differ. All time inputs use the same selected unit;percentages are dimensionless.
IEC definitions and standard boundary
IEC IEV192-01-23 defines availability and requires all relevant intervals to be assigned to up/down states or excluded;it notes effects from reliability,maintainability,supportability,maintenance and delays. IEV192-08-07 defines steady-state availability and says that under certain conditions it can be expressed as mean up time divided by mean up plus mean down time.
IEV192-05-13 defines MTBF as the expectation of operating time between failures and restricts it to repairable items. This differs from time between failures,IEV192-05-03,which includes up and down time. IEV192-07-23 defines MTTR as mean time to restoration and marks “mean time to repair” deprecated.
IEV192-05-08 defines instantaneous failure intensity through a limiting expected event count. Therefore the observed ratio Nf/U or reciprocal mean is not labeled λ or an instantaneous hazard. IEC61703:2016,Edition2,valid with stability date2028,provides mathematical expressions and models for IEC60050-192 measures. Exact model clauses and additional conditions remain NEEDS_LICENSED_SOURCE.
Data safeguards
- Define the required function and observation exposure. Calendar duration,enabled time,operating time and up time are not automatically interchangeable.
- Specify whether fault-detection,administrative,logistic,preventive-maintenance and externally disabled intervals are in D or excluded.
- Use separate failure and completed-restoration counts. A window can begin/end inside a cycle;right censoring and an unresolved final outage affect naive means.
- For pooled assets,justify a homogeneous population and exposure aggregation. A fleet event count divided by one asset’s time is invalid.
- Report uncertainty/confidence and distribution/model checks separately. A mean alone does not describe ageing,bathtub behavior,repair queues,redundancy or common-cause failures.
Что было исправлено
The former input was labeled total operating time,but code subtracted total repair time before dividing by failures while the FAQ stated operating time/failures. It also silently treated blank repair time as zero,used one failure count for both failure and restoration means,called1/MTBF a failure rate λ,and assigned universal90/95/99% quality grades.
The replacement starts blank,requires a state/source record,separates aggregated observations from a documented steady-state model,and uses distinct failure/restoration counts. Zero failures/restorations and censored windows return not evaluated where a finite mean cannot be estimated. Presets,prefix parsing,automatic grades,innerHTML and ambiguous “mean time to repair” language were removed.
