In August 2020, we commissioned our first smart farming project. A 130-acre farm in Melaka, divided into four to five independent blocks of roughly 30 acres each. Each block had its own pump, filtration system, dosing injector, solenoid valves, and sprinklers. Soil moisture sensors throughout. Cloud dashboard. The kind of setup that looked, on paper, like exactly what modern precision agriculture should be.
The project failed. Not because of the sensors. Not because of the software. Because the irrigation hydraulics were fundamentally wrong.
What Actually Happened
Water was not being distributed uniformly across the field. Plants at the far end of each block were receiving less water than plants near the pump — the classic symptom of an undersized pump pushing through an undersized pipe network across 30 acres of head loss. The farmer was experienced. He had been farming that land for years. But like most farmers, he had designed his irrigation system by feel and by what the supplier recommended, not by calculating the actual hydraulic requirements.
The result was stunted growth in the far zones. Then disease, because stressed plants are vulnerable and because uneven moisture creates microclimates that pathogens exploit. The farmer could see the problem in his crops. What he couldn’t see — and what made this particularly damaging — was that the soil moisture sensors were faithfully recording all of it.
The data looked precise. The readings were real. But they were describing a broken system, not a healthy one. A sensor in a dry zone at the far end of the block was reading low moisture — correctly — but the correct response was not to open the valve longer. The correct response was to fix the pipe sizing and pump selection. The automation was making decisions based on data that was accurate but not representative of what a well-engineered system should look like.
The Lesson We Took From It
Good technology amplifies good engineering. It cannot rescue bad engineering.
That’s the principle we drew from Melaka. It sounds obvious in retrospect. But it isn’t obvious when you’re selling automation systems and a farmer is ready to spend on IoT. The temptation — for the supplier and for the buyer — is to believe that more data and more control will solve the problem. It won’t, if the physical system is wrong.
Soil moisture data is meaningless if water distribution is not uniform. If some plants receive twice the water of others, you don’t have a data problem. You have a hydraulics problem. Adding more sensors to a poorly designed irrigation network gives you a more detailed picture of the inequality — it doesn’t fix it.
What We Changed
After Melaka, we changed how we approach smart irrigation projects. Before any discussion of sensors, controllers, or cloud platforms, we now work through the hydraulics: pump sizing, pipe diameter, flow velocity, pressure loss across the network, emitter uniformity. We calculate the distribution uniformity coefficient — the ratio of the average low-quarter application to the average overall application. If it’s below 0.85, the physical system isn’t ready for automation. We fix the system first.
This has meant, on some projects, telling a customer to spend their budget on better pipe and a correctly sized pump before we install a single sensor. That’s not the conversation a technology supplier usually wants to have. But it’s the right conversation, and it’s the one that produces outcomes the farmer can actually measure — in yield, in water consumption, in disease incidence.
Why This Matters for IoT in Agriculture
The smart farming industry has a structural problem. The companies selling sensors and automation are incentivised to sell sensors and automation. The companies selling pumps and pipe are incentivised to sell pumps and pipe. Nobody is incentivised to sit down with the farmer, calculate the full hydraulic system, verify distribution uniformity, and only then specify the control layer.
That gap is where most precision agriculture projects fail. Not in the technology. In the assumption that the physical system beneath the technology is already correct.
The farmer in Melaka was not making a naive mistake. He was doing what most farmers do — trusting that suppliers would flag problems, and trusting that technology would compensate for whatever was imperfect. It doesn’t. It amplifies whatever is already there. If the system is good, automation makes it more efficient and more predictable. If the system is flawed, automation makes the flaws more expensive and harder to diagnose.
The Question Before the Question
When a farmer asks us about soil moisture monitoring, our first question now is not “how many sensors do you need?” It’s “do you know your distribution uniformity?” When they ask about automated fertigation, we ask about head loss across the network and whether the dosing injector is sized for the actual flow rate at the far end of the field, not the flow rate at the pump outlet.
These are irrigation questions, not technology questions. But they’re the questions that determine whether the technology investment will pay off.
If you’re planning a smart irrigation or precision agriculture project and want to work through the hydraulics before the hardware, contact us at info@autoflotechnology.com.