Remote Irrigation Alarm Design: Choose Thresholds Farmers Can Act On
A useful remote irrigation alarm identifies an abnormal condition, the affected block and the action a named operator should take. Set flow and pressure thresholds for the actual operating state, confirm that the measurement is current, and allow only a justified time to distinguish a brief fluctuation from a sustained fault. Then test whether detection, notification and the field response fit the available time. A dashboard alert alone does not establish that a valve closed or a pump stopped.
Pressure gauge and control display on fertigation equipment. This equipment photograph does not show an active alarm or establish remote notification capability. Photo: IrriNex.
This guide develops an alarm worksheet for a remote irrigation monitoring and control system. Its numbers are invented teaching examples, not recommended equipment settings or measurements from an IrriNex installation. Local pump, valve and injection protections require their own engineered response; the examples below concern operational notifications and must not delay those protections.
Start with the operator's decision, then choose the signal
Write the intended response before creating the alarm. “Check zone A's delivery and follow the approved isolation procedure if a leak is confirmed” describes work someone can perform. “Water problem” does not identify a location, urgency or next step. If nobody needs to assess or act on an event, consider keeping it in the maintenance log instead of sending an interrupting notification.
The UK Health and Safety Executive's alarm management guidance emphasizes relevant warnings, a defined operator response and enough time to perform it. These are useful human-factors principles for a farm worksheet; the guidance does not supply agricultural flow limits or certify a controller.
For each proposed alarm, record the block, sensor location, measured quantity, units, permitted operating states, trigger, confirmation period, return-to-normal rule, recipient and required response. Identify the consequence of missing the warning. An irrigation supervisor should be able to explain why this particular alarm exists and what evidence would justify changing it.
Establish a separate flow baseline for each allowed state
Compare measurements with the configuration that is actually running. A mainline meter serving a single block cannot use the same expected value when another permitted block opens or a filter backwash draws water through that meter. Record which flows cross the measurement boundary. A command acknowledgment does not, by itself, establish the physical valve state.
University of Minnesota Extension's discussion of irrigation flow and pressure measurement explains why readings differ across a farm and distinguishes instantaneous flow from accumulated volume. Use verified measurements at the relevant location when establishing a baseline. The example below assumes those checks have already been completed.
| Operating state | Assumed reference flow | Illustrative notification band | Meaning |
|---|---|---|---|
| Zone A alone, steady irrigation | 30 m³/h | 27–33 m³/h | Reference ±10%; investigate sustained readings outside the band |
| Zone B alone, steady irrigation | 18 m³/h | 16.2–19.8 m³/h | A different baseline is needed for the smaller block |
| A and B together | No approved state in this example | No combined-flow band assigned | Resolve the unexpected configuration; do not silently learn it as normal |
| Startup, flushing or backwash | Separate state assessment required | No steady-irrigation band assigned here | Apply a documented transition or maintenance rule |
The ±10% band is an invented inspection threshold, not a universal tolerance or proof of acceptable application uniformity. Real limits must account for the installed equipment, source variation, measurement uncertainty and the consequences of delayed detection. Verify hydraulic performance separately. Do not let an automatic baseline update absorb a persistent leak or a blocked lateral into the definition of healthy operation.
Specify trigger, persistence and recovery independently
A threshold states which readings qualify as abnormal. A persistence period states how long valid qualifying observations must continue before a notification activates. A recovery threshold provides separation from the trigger so a value near the boundary does not repeatedly activate and clear the alarm. A recovery period can additionally require stable evidence before returning to normal.
| Condition | Activation rule | Return-to-normal rule | First assessment |
|---|---|---|---|
| High mainline flow | Flow >33 m³/h for 40 s | Flow ≤31.5 m³/h for 20 s | Check actual zone state, additional outlets and possible leakage |
| Low mainline flow | Flow <27 m³/h for 40 s | Flow ≥28.5 m³/h for 20 s | Compare upstream/downstream pressure, valve state and source delivery |
| Low pressure at the defined block inlet | Pressure <1.6 bar for 20 s | Pressure ≥1.75 bar for 20 s | Confirm the local reading and investigate the delivery route |
| Required measurement unavailable | Invalid quality, or measurement age >45 s | Fresh valid observations meet the site's recovery check | Follow the approved loss-of-visibility response |
The pressure values, time limits and recovery levels are also hypothetical. The unavailable-data rule has no additional persistence timer in this example: the age limit already represents an allowed interval without a fresh observation. An invalid quality flag activates that condition when evaluated. Separate communication checks may be needed even when a sensor still sends plausible numbers.
HSE's control-system guidance on alarm processing discusses deadbands, timers and management of mode-dependent suppression. Apply those concepts through documented settings and testing. A bounded startup allowance must identify when the normal rule becomes eligible and what detects failure to finish startup. It must not suspend required equipment protection.
Walk through the timer with actual timestamps
Assume zone A has already reached its approved steady state at time 0. Fresh, valid flow samples arrive every 10 s, without a moving-average filter. The high-flow confirmation timer starts at the first sample above 33 m³/h. Any subsequent valid sample at or below 33 before activation cancels that pending timer. A missed expected valid sample resets the pending timer because it cannot establish uninterrupted qualifying evidence. The unavailable-data alarm is evaluated separately: invalid quality, or measurement age >45 s. A single missed 10 s sample therefore does not automatically activate the age alarm.
| Time after eligibility | Flow | Result |
|---|---|---|
| 10 s | 34.0 m³/h | Start the pending high-flow timer |
| 20 s | 32.5 m³/h | Cancel the pending timer |
| 30 s | 34.5 m³/h | Start a new timer |
| 40, 50 and 60 s | 34.2, 34.1 and 34.4 m³/h | Continue pending: 10, 20 and 30 s have elapsed |
| 70 s | 34.3 m³/h | Activate: 70 −30 =40 s of qualifying sampled evidence |
| 80 s | 32.2 m³/h | Remain active; flow has not reached the recovery threshold |
| 90 s | 31.4 m³/h | Start the return-to-normal timer |
| 100 s | 31.3 m³/h | Continue recovery confirmation |
| 110 s | 31.2 m³/h | Return to normal: 110 −90 =20 s |
Five samples at 30, 40, 50, 60 and 70 s span 40 s. Counting four samples as “40 seconds” would activate at 60 s, after only 30 s. The trace establishes persistence at the sampled points; it does not reveal unmeasured fluctuations between them. Choose the acquisition method and interval to suit the phenomenon being monitored.
For pulse-derived flow, check the actual observation window and quantization before applying this trace. The flow-meter pulse calculation guide explains why no pulse in a short interval can coexist with real water movement. A last-known positive value must not remain valid indefinitely when new evidence stops.
Budget the complete time from fault to response
Persistence is only one part of the delay. A sensor may sample slowly, a gateway may buffer data, an application may wait for an upload, and the assigned operator may need time to inspect and act. Write the time budget from the physical condition becoming detectable to completion of the necessary field response. Include filtering or averaging latency if present.
In a separate hypothetical review, assume the responsible designer assigns 150 s for an operational response after a sustained high-flow condition begins in an eligible steady state. This is an invented time allowance, not a safe exposure limit. Assume a verified acquisition chain reaches its first qualifying sample within 10 s, confirmation takes a further 40 s, notification delivery takes at most 15 s, the operator reads it within 30 s, assessment takes 25 s and the approved action completes within 20 s.
Total assumed delay =10 +40 +15 +30 +25 +20 =140 s. The nominal allowance left is 150 −140 =10 s. These values assume no additional smoothing delay or travel time; if either exists, add it. The arithmetic only passes the stated budget if every bound and the required response are substantiated. It does not prove reliability from a single successful demonstration.
An operator who checks messages every 5 minutes does not meet the assumed 30 s reading bound. A disconnected phone provides no finite delivery guarantee. If the response cannot fit, revise the measurement, communication, staffing or engineered local action. Do not preserve an attractive timer by omitting inconvenient delays. The pump and valve interlock worksheet addresses local operating permission and protective response as a separate design task.
Give the recipient a practical, evidence-based response
A message should identify the block and sensor, measured value and unit, active operating state, threshold, event time, data freshness and first action. For example: “Zone A high flow: 34.3 m³/h; trigger >33 for 40 s; steady irrigation; valid sample at 70 s; verify open outlets and follow the approved response sheet.” This is an illustrative message, not a claim about an IrriNex software feature.
Flow and pressure combinations narrow the investigation but rarely prove one cause. High flow with low block pressure can prompt checks for an additional open route, a damaged pipe or incorrect instrumentation. Low flow with normal upstream pressure can prompt checks for a restricted route or a valve that has not opened as expected. Account for pressure-tap location and actual configuration before assigning a diagnosis.
For a suspected filter restriction, use the measured filter pressure-loss worksheet instead of assuming that one low-pressure reading establishes clogging. Prioritize the response by consequence and available time. Assign a primary recipient and a backup with the ability to reach the equipment; a long mailing list does not establish responsibility.
Test notification, acknowledgment and physical outcome separately
Use an approved simulation or supervised water-only test with the installer. Do not rupture a line, dry-run a pump or defeat a protective interlock to create an alarm. Exercise the threshold, brief excursion, sustained event, recovery band, invalid input and unexpected operating-state cases. Confirm the timestamps and required action against the agreed worksheet.
Keep separate records for condition activation, notification delivery, operator acknowledgment, equipment command, measured physical response and return to normal. Acknowledgment means the warning was recognized; it does not mean the fault was repaired. Returning to normal does not automatically authorize restarting equipment that has tripped. Preserve the approved reset and restart procedure.
Test the backup contact route and an unavailable gateway as well. The irrigation outage acceptance guide covers the broader distinction between internet access, field communication and local execution. Here, retain evidence that the chosen alarm can still reach someone when a dependency fails, or document the alternative field-check arrangement when it cannot.
Maintain an alarm register as the farm changes
Keep the approved settings beside the zone map, measurement configuration and response sheet. Reassess them after adding laterals, changing emitter flow, altering permitted zone combinations, replacing a sensor or updating control software. Record the change, its reason, the responsible person and the test result. Preserve the earlier version so an event can be interpreted using the settings that were active at the time.
Review repeated alarms with the operator. A recurring event may reveal a real hydraulic problem, a poor state definition, inadequate measurement resolution or a notification that arrives too late to help. Resolve the cause before widening the band. For an IrriNex project review, bring the completed state table, alarm rules, event trace, timing budget and field-test record. Together they define a specific requirement that can be checked against the proposed equipment and operating routine.
