Precise machining for high volume production

• 1
  Why precise machining for high volume production is a capability question
• 2
  How the process plan decides whether the run holds tolerance
• 3
  Fixtures, tool wear and thermal drift in long runs
• 4
  In-process gauging beats end-of-line sorting
• 5
  Where the cost actually sits in a high volume run
• 6
  When high volume machining is the wrong process
• 7
  Which part features belong in high volume machining

A single good part proves the machine can reach a dimension. Ten thousand good parts prove the process can hold it. Those are different claims, and the gap between them is what high volume manufacturing is about. The number that matters is not the best dimension a machinist ever hit, it is the spread of dimensions across the whole run.

Start with the tolerance band and the observed spread. If a feature has a 0.020 mm band and the process naturally scatters 0.006 mm, you have room. If it scatters 0.018 mm with a slow drift from tool wear, you are one small upset away from scrap. A capability index above 1.33 is a working target for most production features; anything near 1.0 means the process is fighting the tolerance, not holding it.

Capability is set by six things: machine geometry, thermal stability, fixture repeatability, tool wear rate, material batch consistency, and how often you measure. Change any one and the spread moves. That is why a part that ran clean for three months can start drifting after a material batch change, with nothing else altered on the floor.

This is the boundary worth stating at the start. Precise machining for high volume production works when the feature set is stable and the tolerance band is wide enough to absorb normal wear. It stops working when a single feature sits at the edge of what the machine can do and the drawing never changes.

• 1
  Fixed cost, spread variableFixtures and gauges are paid once; dimensional spread is paid on every part.
• 2
  Drift is the real enemyTool wear and thermal growth move the mean slowly, not suddenly.
• 3
  Measure often enough to see driftSampling interval shorter than the time it takes the process to move out of band.

A high volume plan is written backwards from the datum. Pick the surfaces that locate the part in every operation, then design fixtures that repeat on those surfaces. If operation 2 locates on a surface cut in operation 1, any error in operation 1 is carried forward and multiplied by the number of features cut later.

Op count matters more than most people expect. Every extra setup adds a fresh chance for location error, chip trapping and clamp distortion. Consolidating features into one 5-axis setup removes those risks and usually tightens the true position of related holes. Our 16 simultaneous 5-axis centers and 16 mill-turn centers exist for exactly this reason: fewer setups, fewer datums, less stack-up.

Clamping is where thin parts lose their shape. A wall 1.5 mm thick will deflect under a clamp load that a solid block ignores. Use light clamping with support underneath, sequence roughing and finishing so the part relaxes before the finish cut, and expect to leave 0.2–0.3 mm of stock for the final pass.

Plan tool changes around the tolerance, not around tool life alone. A tool that holds size for 400 parts and drifts after that should be indexed at 300, with the offset checked at each index. That is cheaper than sorting scrap after the fact.

• 1
  Datum firstChoose locating surfaces that survive every operation.
• 2
  Fewer setupsEach removed setup deletes a stack-up error.
• 3
  Control relaxationRough, let the part cool, then finish.

A production fixture is a measuring instrument that happens to hold a part. Soft jaws machined in place, hardened locating pins and repeatable clamps all serve one purpose: put the part in the same place on part 1 and part 10,000. If the fixture wears, the part moves and the offset you dialed in becomes wrong.

Tool wear shows up as a slow trend in one direction. On aluminium it is mild; on stainless and titanium it is faster, and on Inconel it can move a bore within a few hundred parts. The practical answer is to log the offset at fixed intervals and adjust before the trend reaches the edge of the band, not after.

Heat is the quieter problem. A spindle running for six hours grows, and so does the part. A shop at 20 °C in the morning and 28 °C by afternoon will see dimensions follow the thermometer. Coolant temperature control, a warm-up cycle before the first cut, and finishing passes scheduled after the machine has settled all reduce that drift.

With 127 high-precision CNC machines across three plants and 7,600 m² of floor, we can dedicate a machine to one part number for the length of a run. Dedicated capacity removes the setup-to-setup variation that shared machines introduce.

• 1
  Log offsets, don't guessRecord the correction at each check and watch the slope.
• 2
  Warm up firstRun a spindle warm-up cycle before the first production cut.
• 3
  Dedicate machinesOne part number per machine for the length of the run.

Inspecting at the end of a run tells you what you already made. In-process gauging tells you what you are about to make. The difference is the cost of the parts between the last good check and the moment the process drifted out of band.

Set the sampling interval from the drift rate. If a bore moves 0.004 mm per 100 parts, and the band allows 0.010 mm of movement before correction, check every 150 parts. That leaves margin. If the drift rate is unknown, start denser and loosen once you have data.

Gauge choice matters too. A micrometer reads a diameter at one point; a bore gauge reads roundness and taper. For position, a CMM or a dedicated check fixture tells you what the drawing actually controls. Use the instrument that measures the feature the drawing defines.

Every shipment leaves after 100% inspection, covering raw material check, in-process monitoring and final inspection, with reports available on request. For long runs we can agree on the sampling plan up front so both sides know what is measured and how often.

• 1
  Sample from drift rateSet the interval so the process cannot leave the band between checks.
• 2
  Match gauge to featureDiameter, roundness, and position need different instruments.
• 3
  Reports on requestRaw material, in-process and final records can be supplied.

Unit price falls with quantity, but not forever. The curve flattens once setup and programming are amortized, and after that the price is dominated by cycle time, material and inspection. Adding quantity past that point buys very little.

Cycle time is the lever with the most room. Trimming 30 seconds from a 6-minute cycle is a 8% cost reduction on the part, and it usually comes from toolpath work, not from running the spindle harder. Higher feed rates on a light finishing pass often cost more in tool life than they save in time.

Inspection is the second lever. If the drawing demands a full CMM report on every part, that cost can rival the machining cost. Agreeing on a sampled plan for stable features and full check only on critical ones usually cuts the total without adding risk, provided the capability data supports it.

Setup cost is the third. The same part spread across four machines costs four times the setup. Keeping a run on one machine with a dedicated fixture is often cheaper even when the machine is not fully loaded.

• 1
  Curve flattensAfter setup is amortized, quantity adds little.
• 2
  Cycle time firstToolpath changes usually beat feed rate increases.
• 3
  Inspection is real costFull CMM on every part can match machining cost.

Machining removes material one part at a time. That is a strength for tight tolerances and frozen designs, and a weakness when the shape is complex and the volume is large. A part with deep internal cavities, thin hollow shells or features that cannot be reached by a tool is a candidate for casting, molding or die casting instead.

The crossover is roughly geometric complexity against tolerance. Simple prismatic parts with tight tolerances stay machined at any volume. Complex parts with generous tolerances move to a forming process once tooling cost is amortized. Parts in the middle, where tolerances are tight and the shape is awkward, stay machined because the alternative cannot hold the tolerance.

Material also sets the boundary. High volume commodity plastics are almost always molded, not machined. Aluminium, stainless, steel and titanium parts with real tolerance requirements stay on the machine.

There is a middle path worth naming. Die casting or molding produces the near-net shape, then precise machining for high volume production finishes the critical features: bores, sealing faces, threaded holes and datum surfaces. That combination often beats either process alone, because the forming step handles the bulk shape and the machine handles the tolerance.

• 1
  Machining wins on toleranceFrozen geometry plus tight bands is the sweet spot.
• 2
  Forming wins on shapeDeep cavities and hollow shells belong to casting or molding.
• 3
  Hybrid often winsForm near-net, then machine the critical features.