Operations

Operations covers everything done to run the plant day to day: monitoring, feeding, observation, cleaning, maintenance, stock movements, record keeping, handover between shifts, and the procedures that define how each is carried out and by whom.

• Most incidents are detected by a person noticing something, so the quality of routine observation determines how early problems are found.
• Records are the only means of distinguishing a trend from a bad morning, and they are what any later investigation depends on.
• Consistency of feeding, cleaning and handling drives feed conversion, growth and survival more than any individual intervention.
• Handover between shifts is where information is lost, and lost information is how a developing problem becomes an event.
• Procedures are what allow the farm to run correctly when the most experienced person is not there.

The layers matter because each catches a different class of problem. Continuous monitoring catches fast events; the shift walk catches what instruments do not measure, principally the behaviour of the fish; the weekly and periodic layers catch slow drift that neither of the others would ever reveal.

• Design the routine alongside the plant, since layout, access and instrumentation determine how long the daily round takes.
• Write procedures for the tasks that are done rarely and under pressure — outage, mortality event, emergency harvest — rather than only for the routine ones.
• Structure records so they can be compared with the financial and design models directly.
• Provide the means to record where the work happens, since anything requiring a later trip to an office is recorded late or not at all.
• Decide the staffing model against the degree of automation and the number of independent loops, including out-of-hours cover.
• Define who is authorised to make which decisions, particularly for treatments and emergency actions.

• Observe the fish before reading the instruments. Behaviour, appetite and distribution frequently change before any measured parameter does.
• Record at the time of the observation, not at the end of the day.
• Review records rather than only collecting them; a log nobody reads has no operational value.
• Make handover explicit and written, not a conversation in a doorway.
• Train every operator on the manual procedures, so the plant can run without automation.
• Investigate anomalies rather than absorbing them, and record what the investigation found.
• Keep procedures current with the plant, since a procedure that describes equipment no longer installed teaches operators to ignore procedures.

• Review procedures against how the work is actually done, and correct whichever is wrong.
• Audit record completeness periodically rather than discovering gaps during an investigation.
• Refresh training, particularly on the rare, high-consequence procedures.
• Review the alarm and event log for patterns that individual entries do not show.
• Reconcile stock records against physical counts at each opportunity, typically at grading and harvest.
• Update the routine after any change to the plant.

• Collecting records that nobody reviews.
• Recording the plan rather than what actually happened.
• Relying on verbal handover for information that matters.
• Writing procedures for routine work only, leaving the rare emergency undocumented.
• Depending on one experienced person whose knowledge exists nowhere else.
• Feeding to a table rather than to observed response.
• Treating an anomaly as noise without investigating or recording it.
• Leaving procedures unchanged after the plant has been modified.

Structured records are the prerequisite for every analytical method the farm might later adopt. Consistent parameters, consistent sampling points, timestamps, and events recorded alongside measurements are what allow performance to be modelled at all. A farm that has kept such records for several cycles has an asset that cannot be acquired retrospectively, and one that has not cannot buy its way to the same position.

Operational culture is the harder part. The behaviours that matter — recording an inconvenient observation, reporting a mistake, investigating an anomaly that could be ignored — depend on whether doing so is treated as diligence or as a problem. That is a management decision rather than a technical one, and it determines whether the records describe the farm or describe what was expected of it.

Operations is where a designed system becomes a producing farm. Build the routine in layers so that fast events, human observation and slow drift are each caught. Record what actually happened, at the time, and review it. Make handover written, document the rare procedures as carefully as the daily ones, and keep both current with the plant.