The situation
A fertilizer mixing and loading plant was running on a control scheme with more in common with relay logic than with anything modern. Every step in the process was a manual one. Someone had to physically energize each hopper, each conveyor, and the mixing drum, in sequence. Meanwhile other operators ran loaders to keep the hoppers fed.
That meant a full crew to fill a single order. It also meant the plant's throughput was capped by how fast people could walk between control stations and throw switches — not by the capability of the equipment itself.
What the retrofit had to do
The ask was not to replace the plant. The mechanical equipment — hoppers, conveyors, mixing drum, gates — was sound and correctly sized for the work. The ask was to put a brain on it: a system that could run the whole sequence automatically, or hand a single loader operator complete control from an Android tablet.
That "or" is the important word. A plant that only runs automatically is fragile — the moment an order doesn't fit the recipe, or a hopper bridges, or a customer shows up early, somebody has to override it. A plant that only runs manually is what they already had. We built for both, with the same operator interface either way.
One tablet, one operator
The operator carries an Android tablet. An order comes in, and the breakdown lands on that tablet: each commodity in the blend, and the exact weight of each one needed to fill it.
That is the part that changes the job. The operator is not doing arithmetic on a clipboard or guessing at a hopper level. The tablet tells them what to put where, and how much.
They fill each hopper to its number, then press start.
The mix sequence
One button starts the batch. From there the system takes over:
The mixing drum comes up.
The hopper conveyor runs.
The hopper gates open in sequence.
It runs until the hoppers are empty. Then the gates close and the conveyor shuts down on its own — nobody standing at a panel waiting to catch the end of the batch.
The drum keeps turning for a set mix time, or until the operator presses stop on the tablet, whichever comes first. That gives a repeatable mix for a standard blend, and a manual override for the blends that need a longer or shorter tumble.
Loadout
When the customer's trailer is in position, the operator presses start loadout and the plant fills the trailer.
We mounted cameras on the output spout. The operator watches the trailer fill from the tablet instead of standing at the spout — and, the part drivers appreciate, the camera feed is visible to them too, so they can see when to pull forward or adjust position rather than finding out when product is on the ground.
Overflow at loadout is one of the more expensive ways to lose a sale. A camera on the spout is a cheap fix for it.
What changed on the floor
The plant went from needing a crew of people stationed at switches to running on one loader operator with a tablet. The equipment did not change. The sequence did not change. What changed is that the sequence is now executed by a controller that does not get tired, does not skip a step, and does not need to be told twice.
And when something does need a human decision, that human is holding the whole plant in their hands instead of standing next to one switch.