Scrap reduction in manufacturing comes down to time and precision: how soon you know a defect has started, and how exactly you know what it touched. Most plants find out late, at the next manual check or at final inspection, and then write off more than they need to because they can’t tell good from bad. Tristar’s inline inspection works on both.
The method behind a scrap program, from measuring the baseline to picking the first targets, is laid out in our guide to scrap reduction in manufacturing. Read that for the approach. This page is about what Tristar puts on the line to carry it out.
Where scrap piles up
A die chips. A seal bar fouls. Degraded resin breaks loose in the extruder. The process keeps running, and every cycle after that adds to the pile. If the next check is a while away, the pile is large by the time anyone sees it.
Then comes the second loss. Nobody knows exactly when the defect began, so the hold covers everything since the last known good check. Good parts get sorted by hand or scrapped along with the bad. On a web line the whole roll goes, because finding the bad stretch inside it isn’t practical.
And the third: the defect comes back next week, because the count on the scrap report never said which tool, setting or shift it followed.
How automated inline inspection cuts scrap
| Lever | What changes on the line | What makes it possible |
|---|---|---|
| Catch it at the source | The defect is seen at the station that made it, before more value is added downstream | A camera on every part or the full web |
| Act in seconds | Operators and the PLC respond while the first bad parts are still on the conveyor | Real-time, sound and mass alerts, plus reject and flag signals |
| Scrap only the bad section | The affected length of a roll, or the affected parts in a lot, is removed and the rest ships | Roll maps with defect location, size and time, and traceability to lot |
| Fix the cause | The die, setting or changeover behind the defect is identified and corrected | Trends by line, tool and shift |
The first two levers shorten how long a defect runs. The third limits how much good product goes out with it. The fourth is the one that keeps the scrap rate down after the first drop, and it’s the reason to treat inspection data as process data. Production line monitoring covers the alerting and trend side in more detail.
What Tristar installs to reduce scrap in manufacturing
Cameras at the source
Standard industrial cameras and lighting go where the defect is created: after the press, over the web, at the mold or the bag machine.
A model that knows your product
Tristar trains it from your samples on top of a foundation model that already understands materials. It runs on an edge server in your plant.
Connections to the line
Outputs to the PLC drive reject gates, kick-off systems, flaggers and edge markers. Alerts reach operators by screen and sound.
A record of every defect
Roll reports, roll maps and a dashboard, with data passed to your MES or ERP.
A weekly review
Your team and ours tune tolerances and add defect types, so the system rejects what should be rejected and passes what shouldn't.
That last step matters for scrap in a way that’s easy to miss. An inspection system set too tight creates scrap of its own. Tracking false positives by defect type, and reviewing them, keeps good product out of the reject bin.
What two deployments measured
On four high-speed plastic film lines, a flexible packaging producer cut scrap by 25% and customer claims by 50%. Annual savings came to $814,000, and the system paid back in four months.
In steel stamping, across seven press lines in three facilities, scrap also fell 25%. It began trending down within 120 days. Throughput rose 3%, about 800,000 additional parts per year. One event shows the mechanism clearly: a defect spike on a night shift was flagged within 10 seconds, and more than 10,000 defective parts were avoided. The stamping case study has the full account.
These are two plants’ numbers, not a forecast for yours. We’ll tell you what to expect after we run your parts.
Where to start
Manufacturing scrap reduction doesn’t begin with every SKU. It begins with the 5–10 part and defect combinations that cost you the most, because a camera pays for itself there first and the data is easiest to act on.
A workable first conversation needs only a few things from you:
- Your scrap report, broken down by defect and line as far as you have it.
- Good and bad samples of the top offenders.
- A description of how each one is caught today, and how long it typically runs before that happens.
From there we can say which defects a camera will see, where the station should go, and what the line signal should do. For the material-specific detail, see AI defect detection for plastic film or metal surface defect detection.
