"Output was low again this month, and nobody can tell me why." If you run a garment unit, a printing press, a plastic or packaging factory, or a rice or flour mill, you have probably said something like this. Machines stop many times a day: a thread breaks, material runs out, a mould or die is changed, the operator steps away, the power goes. Each stop is short, so nobody writes it down. At the end of the month all you see is the shortfall.

The fix is to let the machine report its own stops, and to ask a person only for the reason.

Why the paper register misses most stops

Manual downtime registers usually capture the long breakdowns, the ones where a mechanic was called. The many small stops of a few minutes each go unrecorded, and together they can add up to more lost time than the big breakdowns. Supervisors also fill in the sheet at the end of the shift from memory and round the times off. None of this is dishonest; it is simply hard to keep records by hand while you are trying to keep a line running.

How automatic tracking works

A small device is fitted to each machine you want to watch. It needs to know one thing: is the machine running or not? Depending on the machine, that signal can come from:

  • a clamp-on current sensor around the motor's supply cable, which notices when the motor draws power
  • a vibration sensor on the machine body
  • a signal from the machine's own controller or counter, where one is available

The device timestamps every start and stop and sends it over the factory Wi-Fi to a dashboard. When a stop lasts longer than a limit you choose, a simple screen near the line, or the supervisor's phone, asks for a reason from a short list: no material, breakdown, style or die change, waiting for quality check, operator away, power cut.

The dashboard then shows downtime by machine, by shift and by reason, in plain charts a production manager can read in a minute.

Setting it up, step by step

  1. Pick the machines. Start with the bottleneck: the machine or line that limits your whole output.
  2. Agree on what counts as downtime. Decide whether planned breaks and changeovers are counted separately, and how short a stop can be before it is ignored.
  3. Write a short reason list. Six to eight reasons are plenty. With too many, people just pick the first one.
  4. Fit the sensors. Anything inside an electrical panel or around a supply cable must be installed by a qualified electrician.
  5. Run it quietly for a couple of weeks and compare it with what supervisors believed.
  6. Review the top reasons every week with the people on the floor, and work on one at a time.

Load-shedding, Wi-Fi and the factory floor

Power cuts are part of factory life here, so plan for them. Give each device a small battery backup so it can record that a stop was caused by a power cut, rather than counting it as a machine fault. Devices should store data locally when the Wi-Fi drops and send it once the connection returns. Enclosures and cable routes must also cope with heat, dust and monsoon humidity.

Mistakes to avoid

  • Using the data to punish operators. If people fear the system, the reasons they enter become useless. Present it as a way to find problems in material supply, maintenance and planning.
  • Tracking everything at once. Data from twenty machines that nobody reviews is worth less than three machines reviewed every week.
  • Collecting without acting. The value comes from fixing the top reason, not from the chart.

What you can do yourself, and where help is useful

Your production team knows the machines, the reasons and the bottlenecks, and that is the most important input. An expert helps with choosing a sensing method for each machine, building reliable devices, and setting up the dashboard and reports. See IoT device development.

A small first step

Choose one bottleneck machine, fit one device, and run it for a few weeks alongside your current register. If the data changes how you plan maintenance or material supply, extend it line by line. The IoT pilot project page explains how to structure that trial, and you can get in touch if you would like help planning it.

Related service IoT device development Connected devices from prototype to dashboard: sensors, firmware, connectivity and alerts that reach your phone.