Skip to content
Ordanis
PT

Guide · Industrial reliability

MTBF, MTTR and availability, with the math in plain sight.

Industrial reliability is an asset's ability to do its job for the planned time. In practice a plant tracks three numbers: MTBF, MTTR and availability. This guide explains what each one measures and how Ordanis calculates it.

What each metric measures

  • 01

    MTBF — mean time between failures

    How long, on average, the asset runs between one failure and the next. Higher means more reliable.

  • 02

    MTTR — mean time to repair

    How long, on average, it takes from the failure until the asset runs again. It measures diagnosis, parts and repair speed.

  • 03

    Availability

    Share of planned production time in which the asset was running. It combines the effect of failures and repair time.

How Ordanis calculates

These are the engine's formulas, with the identifier and version stored with every result.

MTBF
Version: mtbf.v1
runtime_h / failure_count

Running time in the period divided by the number of failures. What counts as a failure comes from an explicit definition, not from the stop class; failures can come from telemetry, imports or manual records, merged without double counting.

MTTR
Version: mttr.v1
Σ (return_to_running − failure_start)_h / reliable_repair_count

Sum of the times from failure start back to the RUNNING state, divided by the repairs with a reliable start and end.

Availability
Version: availability.v1
runtime_in_schedule / (planned_production_time − unknown_in_schedule)

Running time inside the planned schedule, divided by planned time minus the time in which the machine state is unknown. It is not the OEE availability factor.

When there is no number

A reliability metric only helps if it can be trusted. That is why every result has a status.

  • 01

    VALID

    Closed period and data coverage above the definition's valid minimum.

  • 02

    PROVISIONAL

    There is a number, with a caveat: open period or coverage between the minimum and the valid level.

  • 03

    INSUFFICIENT_DATA

    No number is published. The reason is shown: zero failures, too few failures, low coverage, no runtime or no reliable repairs.

Before telemetry exists (for example, before the Edge is installed), MTBF and MTTR can be calculated from imported history only. Runtime is then a declared assumption (24 h a day or declared hours per day), the formula becomes mtbf.history.v1 / mttr.history.v1 and the result is always PROVISIONAL, with low confidence.

What every calculation records

  • Formula and version (for example, mtbf.v1).
  • The definition used (what a failure is, what a microstop is), with code and version.
  • Period, scope (asset, line, plant) and data coverage.
  • Events used and events excluded, with the reason.
  • A hash of the inputs, so the same calculation can be reproduced later.

Limits, plainly

  • The calculations were tested with simulator and test data, not yet with a customer's production data.
  • MTBF is only as good as the failure definition agreed with the plant's maintenance team.
  • Reliability availability and OEE availability are different metrics with different formulas.
Ordanis

One Industrial Reality.

Start with one operational problem. Expand into one industrial reality.

Request Pilot