Nights, Weekends and Holidays: Who Watches the Greenhouse?

A commercial hydroponic house runs 168 hours a week and the crew is in the building for a fraction of them. On most sites the crop spends roughly two-thirds of its life with nobody there, so greenhouse monitoring on nights, weekends and holidays is not an add-on — it is the normal condition. What matters is who takes the call at 03:00 on a Sunday.

Written from the commissioning side: which failures arrive in unattended hours, how sites staff the gaps, and the response targets worth writing down. It assumes a working alarm system exists; thresholds are a separate subject.

What Breaks While Nobody Is in the Building

The faults are the same ones that happen on a Tuesday morning. What changes at night is the delay before anyone touches them. Six events account for most out-of-hours losses:

Only two of the six are visible without instrumentation. Flow, power state and vent position must be measured as states, not inferred from temperature, or the platform reports the consequence hours later.

Why a Weekend Failure Costs More

Two delays add up — detection and response — and the second is usually the larger one: a fault starting at 22:00 on a Saturday is often confirmed by a Monday walk-through rather than by an alert. Response is where the weekend hurts: the on-call person may be twenty minutes away or ninety, may need a second person for a safe intervention, and may find the stores locked. On NFT in warm weather, two to four hours without flow before wilting shows is a common planning assumption — shorter under high radiation, longer in a deep water bed. Set your own number from a drill.

Covering Greenhouse Monitoring on Nights and Weekends

There are three ways to cover the unattended hours, and the choice is about who can act rather than who can see.

An on-call rota a hydroponics crew can sustain needs four to six names for one site once leave, sickness and training are taken out, on a one-week rotation with a named deputy. The duty has to be paid — allowance plus callout — and that person should not run a full shift after a 04:00 visit.

Write the escalation down: if the first name does not acknowledge inside the tier window, the call goes to the second, then the manager. A rota with no second name fails the night the first phone is on silent.

Severity Tiers and Response Targets

Give every alarm a tier and a clock, short enough for a tired person. The ranges below assume the on-call technician lives 20–30 minutes away, holds keys, and can authorise a callout.

• Tier 1, crop at risk within the hour: no flow, pump circuit power loss, high temperature with vents closed. Acknowledge in 5–10 minutes, on site in 30–45 minutes.
• Tier 2, damage within hours: temperature drift, EC or pH out of band, vent position not matching command. Acknowledge within 30–60 minutes, resolved in the next working window.
• Tier 3, information: trends and maintenance reminders — logged, reviewed next working day, no call.

State per tier what the responder can do remotely — switch to the standby pump, change a setpoint, force a vent open — and what needs a drive. Remote alarm escalation only works if the notification path survives the event it reports: a cut that drops the router drops the alarm, so comms need an independent supply.

Alarm Fatigue Is a Staffing Problem

Fatigue is a rota that has stopped trusting the phone, and it comes less from bad thresholds than from being woken for nothing: acknowledgement times stretching, phones muted overnight, one person carrying the duty. What holds up:

• One event, one notification — group repeats rather than calling every cycle.
• Suppression during planned maintenance, tied to a work order.
• A monthly alarm-count review: anything repeating three times a week gets an engineering fix, not a settings change.
• Fair rotation and visible payment for callouts.

Threshold tuning belongs with the alarm documentation; the point here is that a rota degrades on a predictable schedule if nobody maintains it.

What Unattended Greenhouse Monitoring Covers, and What It Does Not

A platform detects, records and escalates. It cannot intervene beyond what is motorised and networked, and it only knows what is instrumented — both matter when a site weighs cutting a shift.

• Covered: flow and pressure, EC and pH, air and solution temperature, tank level, power status, vent and screen position, history for claims, and remote setpoints where actuators exist.
• Not covered: crop appearance, since wilt, pest and disease are visual; mechanical faults no sensor sees; anything downstream of a dead network or controller; physical work like opening a manual valve; and slow leaks.

Monitoring replaces presence, not people: it cuts detection to minutes and leaves the response to whoever can get there.

Before a Public Holiday: The Cover Checklist

Holiday cover fails in a consistent way — the rota is published and nothing else is checked. Work through this a week out:

1. Publish names, numbers and the deputy for every slot, plus days either side.
2. Test the escalation path with a real call and confirm a phone rings.
3. Check generator fuel and agree who refuels.
4. Pull critical spares to site — pump head, solenoid, fuses, sensors — and top up dosing tanks.
5. Review setpoints against the forecast; stage the crop so no contracted harvest falls due on day one back.
6. Name who reviews the log each morning — ten minutes a day catches slow drift.

After the closure, review the log even if nothing happened: a quiet holiday with a hundred suppressed alarms says the same as a noisy one.

FAQs About Out-of-Hours Greenhouse Cover

How many people does an on-call rota need?

Four to six for one site — more than most first drafts propose — on a one-week rotation. Fewer names and one person ends up permanently on duty.

Can a monitoring platform replace the rota?

No. It shortens detection and handles escalation, but someone still has to reach the site and touch a pump or a breaker.

What response time should we commit to?

Tie it to a tier and to the drive: 30–45 minutes on site for crop-at-risk alarms, assuming a 20–30 minute drive.

Should on-call duty be paid?

Yes, as a standby allowance plus a callout rate. Unpaid rotas erode quietly: acknowledgement times lengthen and the duty settles on whoever cannot refuse.

What is the most common gap in holiday cover?

An escalation path never tested end to end. The rota exists, the platform is configured, and nobody has checked that a 03:00 fault rings a phone that is on.

Related Reading

• Hydroponic Alarm Systems: Thresholds, Escalation and Avoiding Alarm Fatigue — Threshold setting and the escalation chain behind the tiers above
• Smart Hydroponics: IoT and Remote Monitoring — What sensors and connectivity deliver, and where they stop
• Backup Generator Sizing for Hydroponic Farms: kW, Fuel Days — Sizing the standby supply that carries an overnight outage
• Surge Protection and UPS for Hydroponic Controllers — Keeping controllers and comms alive through the disturbance
• Hydroponic Farm Risk Management: Power Failure, Crop Loss and Business Continuity — The continuity plan that out-of-hours cover sits inside

Next Step: What to Send for a Quotation

Out-of-hours cover is partly a specification question: how many zones report, which states are measured, and whether the panel supports remote switching all belong in the greenhouse monitoring scope for nights, weekends and public holidays, not a retrofit.

Planning a new house or upgrading the controls on an existing one? For RFQ or project consultation, send: greenhouse dimensions and usable growing area; crop plan and target weekly or annual output; local climate range and raw water analysis; number of zones to be monitored; which states need measuring (flow, power, vent, temperature, tank level); automation and alarm level required, including remote switching; delivery schedule; and the budget band. G AND N FORTUNE LIMITED supplies NFT channels, DWC systems and matched greenhouse packages, with control scope quoted to those parameters.