Two irrigation systems, two chillers, and the soil temperature that showed which one had failed

The situation

This grower cools its irrigation water before it reaches the crop. Cut flowers are sold on stem quality and uniformity, and the temperature of the root zone is part of how that is produced rather than a variable left to the season, so chillers bring the water to temperature on the way to the beds.

There are two irrigation systems on the site, each with its own chiller, each feeding its own beds. That arrangement is normal and sensible. It also creates a blind spot, because nothing in it reports whether the water is still at temperature when it arrives at the plant. The chillers know whether they are running. They do not know what the roots got.

That distinction matters more than it sounds. A chiller that has failed outright announces itself. A chiller that is slowly losing capacity does not: it runs, it reports normally, and the water leaving it is a little warmer every week. In a greenhouse in warm weather, the resulting drift in the beds looks exactly like the weather, which is the most available explanation and the wrong one.

What was installed

Each system was instrumented in its own beds, identically: a tensiometer reading soil water tension, and a temperature and humidity sensor in the root zone, both wired to an Omni Genesis controller on the greenhouse upright. A small solar panel above it powers the station and reporting is cellular, so the sensors could go where the crop is rather than where power happened to be.

Nothing was attached to the chillers. That was the point.

Why the pair is the instrument

The useful measurement here is not the temperature of either system. It is the difference between them.

A single reading in a single bed is weak evidence. Warm soil in July has too many possible causes to act on: the weather, the time of day, a sensor drifting, an irrigation that did not run. Any of those explains it, and a grower who has run this house for years will reach for the weather first, correctly, most of the time.

Two systems in the same greenhouse, under the same sun, on the same day, should move together. When they stop moving together, every shared explanation is eliminated at once. The weather is shared. The season is shared. The house is shared. What is not shared is the chiller, and that is where the difference has to come from.

This is why the second station was worth installing even though it measures the same quantity as the first. It is not redundancy. The comparison is the instrument.

What the tensiometer is for

The temperature sensors would have found a divergence. The tensiometers are what made it interpretable.

The obvious alternative explanation for one system’s root zone running warmer is that it did not get watered: dry soil heats faster and holds heat differently, so a blocked line or a failed valve produces a similar signature. Soil water tension settles that immediately. Once the record shows that irrigation arrived normally in both systems and the temperatures still parted company, the water itself is what differs, and the chiller is the only thing that sets it.

The grower found the failing unit and repaired it before the crop was affected.

Why it matters

The generalisable part is not about flowers or chillers. It is that you can often measure a machine most reliably by not measuring the machine.

Equipment telemetry reports intent: the pump says it is pumping, the chiller says it is chilling, the dosing unit says it dosed. Each of those is true and none of them describes what reached the plant. A sensor in the root zone is downstream of every failure in the chain, including the ones nobody thought to instrument, and it costs one station instead of a monitoring point on each piece of plant.

And where a farm already runs two of something, instrument both. The second station rarely feels justified, because it measures nothing the first one does not. What it adds is a control: a reading that should agree, and whose disagreement is a fact rather than an interpretation.