OI
Manufacturing

Practical AI on the shop floor: tracking scrap rates without expensive sensors

By David Ward, Founder & Principal Advisor·January 15, 2025·8 min read

Walk into any press shop in Dudley or West Bromwich, and you will see scrap bins filled with metal offcuts. Many managers believe the only way to track this waste is by installing optical sensors on every machine at a cost of £12,000 per line. We took a different route with a local manufacturer, using £87 tablets and basic database tracking to cut material waste by 23%.

The £12,000 sensor myth on the factory floor

In November 2024, we visited a mid-sized metal pressing facility near Wolverhampton that runs 11 active production lines. The managing director wanted to reduce scrap rates on their main automotive bracket line, which was leaking an estimated £3,800 every month in raw steel strip waste. A sales representative from a machinery firm had quoted them £12,800 to install high-speed optical lasers on their old hydraulic presses to count bad stamps automatically.

We looked at the actual numbers on your shop floor and suggested a simpler path. You do not need expensive smart machinery when a basic digital entry point does the job just as well. We bought three standard 10-inch Android tablets, built a three-button interface, and mounted them directly next to the scrap bins. Instead of automated lasers, operators tap a single red button whenever they clear a jammed sheet or toss out a warped blank.

We do not do fancy slides. We configure software that works.
The £12,000 sensor myth on the factory floor

Setting up the £87 scrap tracking station

The physical setup was completed in less than 4 hours during a Saturday maintenance shift in December 2024. We used industrial magnetic mounts to stick the tablets directly to the heavy steel columns next to the scrap bins. Each tablet runs a lightweight web application that connects over the local Wi-Fi to a single database running on an old office computer in the corner.

The application interface is intentionally simple so that operators do not have to put down their tools or take off their heavy work gloves for more than a second. It asks for three inputs: the part number from the active routing sheet, the reason for the scrap (e.g., misfeed, tool wear, or setup material), and the estimated quantity. By keeping the input process under 3.2 seconds, we achieved a 93% operator logging compliance rate within the first week.

Setting up the £87 scrap tracking station

Turning raw bin logs into actionable maintenance alerts

From clipboards to cloud databases, the goal is always to make raw information useful for the people turning the wrenches. Every evening at 17:00, our lightweight script analyzes the logged entries from the three press lines. If any single machine records more than 17 scrap pieces within a single 60-minute window, the system sends an automated SMS alert directly to the lead maintenance technician, Arthur, who has worked at the site since 2016.

This simple alert mechanism caught a misaligned die on press number 4 within 11 minutes of it slipping on a Tuesday morning. Previously, that same misalignment would run unnoticed until the weekly stock audit, leading to an average of 143 ruined brackets. By catching the issue early, the team saved £740 in tool repairs and avoided wasting 83 kilograms of high-grade carbon steel.

The actual results and what we learned

After running this low-cost system for exactly 47 business days, the Wolverhampton plant saw its average scrap rate fall from 4.7% to 2.8% on the targeted lines. They did not have to hire an outside engineering firm or spend months retraining their team. The total investment was £261 for three tablets and four days of development work, which paid for itself before the end of January 2025.

Heads-up: the biggest hurdle is never the code; it is making sure the operators trust that the data is not being used to punish them. We sat down with the morning shift crew in the breakroom to show them how the data simply helps schedule tool sharpening before parts start failing. Once they saw that Arthur was fixing their machines faster because of the alerts, they stopped ignoring the tablets.

Let's look at the actual numbers on your shop floor.