Interactive prototype – simulated example data only. No live data, no PLC connection and no change to the Selmo Standard.
Where does the time go?
Loss time by measurable reason, in station minutes.
Availability: when is AutoRel lost – and why?
A sequence is available as long as its automatic release (AutoRel) is present. Every AutoRel loss is an availability loss with a fixed, measured reason.
Station
AutoRel losses by reason
Interlock
Manual end of cycle
Manual mode without fault
Availability
Performance: where does a step take longer than its target?
Measured only while AutoRel is present and the sequence is running. Every step time above the target step time is performance loss.
Station / step
Target step time
Actual (median)
Steps above target
Performance loss
Share
If a step exceeds the time t max specified in the Behavior Reference (BR), this is detected as a deviation from the defined behavior. The resulting stop appears as an availability loss with the reason “Interlock” – not as a performance loss.
Bottleneck: which sequence sets the cycle?
Sequence time per part, measured against the specified target cycle time. A cycle time improvement only takes effect at the bottleneck.
Bottleneck candidateSequence time per partTarget cycle time
Waiting times
Blocked and starved times are measured. Whether they count as a loss is still to be specified.
Station / Sequence
Parts/cycle
s/part
Gap to target cycle time
Blocked
Starved
A bottleneck candidate is not proof of an effective line bottleneck. Stations are linked only in the cockpit; blocked and starved times alone do not evidence a cause.
Measurement model: what is measured – and how the cause is established
Loss type
Definition
Measured reason
Status
Availability
The sequence’s AutoRel is lost
Interlock · manual end of cycle · manual mode without fault
defined
Performance
Step time above target step time while AutoRel is present
Station, sequence, step, duration of the deviation
defined
Bottleneck
Sequence time per part against target cycle time
Station with the highest time per part
Candidate
Waiting
AutoRel present, but no product (starved) or output occupied (blocked)
Blocked and starved time per station
Assignment open
Quality
NOK parts
Only where NOK coding exists
Outside this model
Scope
The evaluation does not extend the scope defined in PTF (Process · Technology · Function) and the Behavior Reference (BR). Only what is specified there as measurable and decidable is measured. The basis remains the production process, hardware and interfaces.
Actions
The cockpit shows the loss and the measured reason. The user derives and decides the action. Outlook: AI support in deriving actions is conceivable as a later option. It would work only on these measured reasons and decide nothing itself.
Architecture
Level
Task
Boundary
Process analysis / PTF / Behavior Reference (BR)
Target behavior, target step times, t max
No automatically known physical cause
Selmo Plant / HWZ / sequence / zone
AutoRel, interlock, operating mode, step times as standard signals
No new PLC logic planned
ST10 … ST80
Independent station controllers according to the Selmo Standard
No higher-level line control required
OEE Cockpit
Read-only evaluation, linking, Pareto, bottleneck
No control commands, no invented root causes
User / operator
Derive and decide actions, check effectiveness
Selmo displays, it does not decide
Acceptance rule: Every loss position must be traceable to Selmo signals and process references; unassigned time remains visible.
What the cockpit does not do
What is detected is the deviation from the defined behavior. Selmo replaces neither commissioning nor experience — it reduces the space of possible causes. And what is checked is always the defined behavioral space: what is not modeled is not covered — it appears as unassigned time, not as a cause. What never occurred in the period under review has not been tested — it simply did not come up.
