In a precision engineering workshop, each non-conforming part costs twice: once in material and machine time, and again in processing costs (rework, sorting, scrap, customer complaints). Yet many workshops still accept scrap rates as inevitable, when they should be a signal to be analyzed. Reducing rejects in machining is not a question of good will or increased vigilance on the part of operators. It's a question of method, measurement and variability control.

This article details the most frequent causes of scrap in mechanical production, the levers available to reduce them, and how modern tools now make it possible to deal with the problem at source.

Why is the scrap rate so high in so many workshops?

The first mistake is to look for the cause of rejects in the operator. In the vast majority of cases, non-conformities are the result of systemic causes: raw material variability, machine thermal drift, tool wear, imprecise initial settings.

A workshop running at 3 or 4 % of scrap generally doesn't have a human skills problem. It has a process control problem.

Some common causes observed in precision mechanics:

• Uncompensated thermal drift The machine heats up during production, dimensions drift progressively, and parts come out of tolerance.
• Approximate initial setting The first good part is validated visually or with a manual instrument, without statistical analysis. The operator starts from a sub-optimal centering.
• Tool wear not detected in time Without real-time variability monitoring, wear-related drift is only detected after several non-conforming parts have been produced.
• Material variability not taken into account Two batches of raw materials from the same supplier do not always behave identically during machining.
• Unreliable measuring instruments A caliper out of calibration, a comparator incorrectly mounted - the measurement is wrong, and so is the decision based on that measurement.

What your scrap really measures: an introduction to capability

Before looking to reduce scrap, it's useful to understand what the statistics say about your process. The Ppk index measures the actual capability of a process: it expresses the difference between observed variability and imposed tolerances.

A Ppk below 1 means that your process is statistically producing non-conformities. A Ppk of 1.33 is often considered the threshold for a capable and stable process. Below this level, scrap is not a surprise: it is mathematically predictable.

This statistical reading is fundamental because it shifts the question from «who made the part bad» to «what generates this variability». And there, the answers change completely.

Concrete ways to reduce machining scrap

1. Control variability before mass production

The initial setting determines everything. If the machine already starts out out-of-step, corrections during production will simply make up for a permanent delay. The most effective approaches are based on :

• systematic dimensioning of the first parts,
• a centering analysis before starting the series,
• machine offset correction before volume production.

This is exactly what an automated process control system (APC): it analyzes the first parts, calculates the offset and corrects the machine correctors without manual intervention. The first good part is produced right from the start, not after ten test parts.

2. Detecting drifts in real time with the SPC

Statistical process control (SPC) consists of tracking measurements during production on control charts. When a drift appears, even if the parts are still within tolerance, the SPC detects it and alerts the operator.

That's the difference between curative and preventive: we stop before producing non-conformities, not after.

Well-deployed SPC monitoring also enables us to distinguish between common causes (natural process variability) and assignable causes (material batch change, tool wear, machine incident). Without this distinction, corrective actions are often poorly targeted.

3. Reliable measurement

A measurement result is only of value if the instrument that produced it is reliable. Studies Gage R&R allow us to quantify the proportion of variability due to the measurement system itself. In some workshops, this variability represents a significant proportion of the tolerance, meaning that sorting or scrapping decisions are partly based on measurement noise.

What's more, using an instrument that has been taken out of its calibration interval is a fault that a quality audit will sanction without hesitation. Rigorous management of the instrument pool (life cards, calibration alerts, automatic blocking) eliminates this risk.

4. Intelligent supply control

Purchased raw materials and components make a significant contribution to production variability. Systematic incoming inspection, based on the actual quality of the supplier, can detect problematic batches before they enter production.

Standards ISO 2859 and ISO 3951 The more reliable the supplier over time, the less stringent the controls. The less reliable, the tighter the controls. This dynamic approach reduces the control load while maintaining the same high standards.

5. Automate machine setting: close the loop

This is the most structuring lever for precision engineering workshops. An APC (Automated Process Control) system connected directly to machine tools can :

• read measurement results in real time,
• calculate the necessary correction,
• modify machine correctors without human intervention.

The measurement → decision → correction loop is fully automated. The operator is informed, but does not have to intervene for each correction. This type of system can bring a Ppk from less than 1 to more than 1.4, radically transforming the rate of non-conformities.

How much does scrap really cost?

The classic mistake is to count scrap at its material cost. The reality is much harsher.

Scrap includes :

• the cost of the material,
• machine time consumed,
• operator time,
• detection and sorting costs,
• processing costs (rework or scrap),
• and in some cases, the cost of late delivery or customer complaints.

On high value-added parts - as is often the case in precision mechanics - this total cost can be 5 to 10 times the material cost alone. A scrap rate of 4 % reduced to 1.5 % represents very tangible savings, which can amount to tens or hundreds of thousands of euros per year, depending on volumes.

What automation means for small production runs

A frequent obstacle to the adoption of process control systems is the diversity of part numbers. A workshop producing 2,000 part numbers a year in short runs finds it hard to justify the high set-up times required for each new part.

It is precisely for this reason that automatic configuration generation from the machining program (CAM) is a major differentiator. When the system automatically configures itself from the digital drawing, the cost of implementation per part number plummets, and the benefits of process control become accessible even on the shortest production runs.

Conclusion: reducing machining scrap requires a systemic approach

There is no magic bullet for eliminating machining scrap. What works is an approach that tackles the real causes: process variability, initial set-up, measurement reliability, supply control.

Statistical tools and automated control systems now make it possible to do what used to be the job of an expert: analyze variability, detect deviations and correct the machine, without adding to the operator's workload.

Reducing scrap in machining means first deciding to measure what's going on properly, and then acting on what the data reveals.

FAQ - Frequently asked questions about reducing machining scrap

What scrap rate is considered acceptable in precision mechanics? There is no universal threshold, but a rate of more than 2 % on stable series generally indicates an insufficiently controlled process. Below 1 %, the process is considered efficient. The basic objective is to aim for zero defects by stabilizing variability.

What's the difference between scrap and rework? If a part is scrapped, it cannot be salvaged; it is downgraded or scrapped. Rework brings the part back into tolerance by means of an additional operation. Both generate costs, but rework often masks a process problem that scrap makes more visible.

Is SPC enough to reduce machining scrap? The SPC is a detection tool: it signals deviations before they produce non-conformities. But corrections must always be made, either manually or automatically. An SPC without rapid corrective action has only a limited effect. APC goes further, automating correction directly on the machine.

Can these tools be deployed on small production runs? Yes, as long as configuration by reference is rapid. Systems that configure themselves automatically from the CAM program drastically reduce the cost of implementation per new part, making deployment viable even on highly diversified workshops.

How do you convince management to invest in waste reduction? The simplest ROI to present is the actual cost of current scrap (material + machine time + sorting + processing), multiplied by annual volume. Comparing this amount with the cost of an improvement project usually gives a very quick justification, often less than a year's ROI.