Guide · Downtime management
Every stop has a class — or shows up as unknown.
Downtime management starts by separating what is planned, what is a failure, what is a microstop and what comes from outside the machine. Ordanis does this from the machine state, with versioned rules and with the final word of the people on the line.
The seven stop classes
01
Planned
Maintenance window, break, scheduled stop.
02
Changeover
Planned format or product change.
03
Unplanned
Unplanned stop lasting at least the microstop maximum.
04
Microstop
Short unplanned stop within the asset's microstop definition.
05
Starved
Machine ready, but no material coming from upstream.
06
Blocked
Machine ready, but the next process is full.
07
Unknown
The data does not support a class. It stays visible until someone classifies it.
A microstop is a definition, not a guess
Each plant defines what a microstop is: minimum duration (below it is noise), maximum duration (from it on it is unplanned downtime) and the machine states that count. The definition can apply to the organization, plant, line, asset class or a single asset — the most specific wins — and has a version and a start date.
count(unplanned stops with min ≤ d < max); Σ d in the periodNumber and total time of stops within the definition valid at each event's start.
Σ duration(unplanned stops with d ≥ microstop max)Time lost to unplanned stops, prorated to the analysed period.
The people on the line have the final word
- The operator reclassifies a stop with the new class and a reason; the decision is recorded and audited.
- An operator reclassification takes priority over the automatic classification.
- Whether a stop counts as a failure (for MTBF) is decided by the failure definition, never by the class.
- Stored calculations do not change: a new reading uses the new classification.
Limits
- Automatic classification needs a reliable machine state from telemetry; without it, failures and work orders can come from imports or manual records, but there is no automatic classification.
- Not yet validated with a customer's production data.
One Industrial Reality.
Start with one operational problem. Expand into one industrial reality.
