Case Study · Cold Exposure

Two Antennas per Door. The Reading Order Tells the Rest.

If the outer antenna reads first, the person is going in. If the inner one reads first, they are coming out. That is how 3,000 workers have their cold-room time counted without doing anything — no badge, no button, no queue.

3,000 workers Passive RFID Food processing
The client

Friato Nutriza

A food processor with around 3,000 employees working daily in refrigerated environments. Under Brazilian labour law, continuous exposure below 15 °C is limited to 1 hour and 40 minutes, followed by 20 minutes of rest in an area with thermal and acoustic comfort, seating and drinking water.

At that scale, complying with the rule is one thing; being able to demonstrate each individual's compliance on a given day is another — and it is the second one a labour inspector asks for.

The challenge

The Crew Enters Together. The Exposure Is Individual.

The usual approach in the sector is to form teams that enter and leave together, supervised by a lead. It works as coordination and fails as a record — because the rule counts each person's time, and crew-level control cannot tell who stayed longer, who came back early, or who did not stop at all.

Crew control, individual liability
Grouping moves people together, but the legal requirement is per worker — and the two never meet.
A spreadsheet filled by the lead
The record depended on manual entry, which raises a question of accuracy exactly when it is being examined.
Lost productivity
Moving a whole crew because one person must stop costs time from everyone who could still be working.
No way to know who skipped
Without individual identification, there was no way to name who missed a break — or to find out why.
The solution

Every Door Becomes a Direction Sensor

Passive RFID was chosen for one specific requirement: the record could not depend on any action by the worker. Biometrics and facial recognition were evaluated and ruled out for that reason — they require stopping, looking or touching.

Two Antennas on Every Door
One outside, one inside. The reading sequence defines direction: outer first is entry, inner first is exit.
Automatic Status
On entry the person becomes “Working”. On exit, “On break”. Each clock runs on its own, with no manual entry.
Every Door Covered
Eight readers and sixteen antennas ringing the cold area — so no passage exists without control, which is where the record would be lost.
Tag in the Helmet Liner
And in the ear protector, for areas that use hairnets. Two tags per helmet, so at least one is read.
Reader in the Ceiling Liner
Away from ambient humidity, which shortens the life of the most expensive equipment in the installation. Antennas with aluminium base and ABS housing.
Always-Current Records
An API synchronises the employee register daily with the HR system — joiners and leavers enter the count on their own.

On equipment: the installation uses fixed readers and antennas from market vendors — the readers are Zebra FX7500. The platform works with hardware from multiple manufacturers; the brand named is the equipment chosen for this project, not an exclusivity.

Why this matters more than it seems

The record happens without the person noticing it is being made. They walk through the door the way they would anyway, and the data is born from that.

Any method that requires stopping, touching or looking at a reader competes with production — and loses on peak days, which are the days exposure is highest. A control that only works when there is spare time is not a control.

The infrastructure

What Was Installed

3,000
workers with individually counted time
8 + 16
fixed readers and antennas ringing the cold area
7,500
RFID tags in helmets and ear protectors
Oct/24 → Jan/25
from installation to production
The proof of concept

They Tried the Boot. It Didn't Work.

Before installing across the plant, a portal was built on a single door to decide where the tag should sit. Two hypotheses were tested — helmet and boot — and the boot failed for three separate reasons.

The human body blocks the signal
With the tag on the boot, one person walking near another caused interference. The body is a serious source of blocking, because of how much water it holds.
Leather boots don't take adhesive
Rubber ones do. Leather would require stitching or riveting the tag — another process, another cost.
Boots last six months
Each worker would consume two tags a year on footwear replacement alone, multiplying project cost over time.
Floor antennas, ruled out earlier
The cold-room floor is washed several times a day, and the film of water left behind blocks the radio wave.
What the test settled

Tag in the helmet liner and in the ear protector — positions where the tag faces the antenna with nothing in between, and no contact with the body.

Two tags per helmet, to guarantee one read. Plus a name sticker on the outside, so supervisors can spot a PPE swap by eye.

This is the kind of decision that never appears in a commercial proposal and that determines whether the project works.

What changed

The Coordinator Gets a Name, Not a Total

That is the change that alters the working day. Before, nobody could say who had missed a break. Now the coordinator receives the name of whoever is running over time and can go and find out why — which is sometimes operational, not carelessness.

Supervisors follow crew presence from a mobile app in the field, and Friato's IT team took the passage data into their own BI dashboard, where HR and production consult it daily. Data born from a legal requirement also became a reading of movement and idle time.

Who, by name, missed their break yesterday?

Not how many breaks were taken — the name of whoever did not take one. If the answer depends on asking the crew lead, the control is collective and the inspector's question is individual.