Effect of temperature on machining precision of machine tools

• 1
  Where the heat comes from and what it moves
• 2
  Machine tool thermal growth and geometric error
• 3
  Workpiece heat, thin walls and surface finish
• 4
  Controls that actually reduce thermal error
• 5
  When thermal control is not the answer
• 6
  Step by step: holding tolerance on a warm machine
• 7
  Rough thermal error budget for common features

Every machine tool holds its geometry at one temperature. When the temperature changes, castings, ballscrews, spindles and workpieces all change length, and they do it at different rates. Steel grows about 11.5 μm per meter per °C, aluminum about 23 μm, cast iron around 10.5 μm. A 1,000 mm aluminum part that warms 5 °C from cutting heat grows roughly 115 μm. That is 23 times the ±0.005 mm tolerance we hold on tight features.

The heat has four main sources. Ambient air in the shop, spindle and axis motor losses, friction in guides and bearings, and the cutting zone itself. None of them arrive at the same time. The spindle warms within minutes. The bed takes hours. So the machine is never in one thermal state, it is always drifting toward the next one.

This is why a first-article part and a part cut four hours later can differ even when the program never changed. The error is not random. It is repeatable if you record the thermal history, and it is the reason warm-up routines and in-process probing exist.

• 1
  Ambient airShop temperature swings move the whole machine frame, slowly.
• 2
  Spindle and motorsFast local growth at the front bearing and axis drives.
• 3
  FrictionBallscrew, guideway and bearing losses add heat along the axis.
• 4
  Cutting zoneChip and tool heat enter the part at the contact point.

A spindle grows along its axis as it warms. On a 5-axis machine that shift moves the tool tip in Z, so a face that should be flat comes out with a small step between passes. Ballscrews grow in length, which shows up as pitch error that grows toward the ends of travel. The screw is anchored at one end and free at the other on many machines, so the error is not symmetric across the table.

Geometry suffers too. If one side of a column is warmer than the other, the column tilts. Squareness between X and Y then drifts by a few micrometers over 300 mm. A machine that was square when it was leveled cold is not square at 30 °C spindle temperature.

Measurement happens on a cold coordinate measuring machine. A part cut on a warm machine and measured on a cold CMM can read out of tolerance even though the machine was correct at the moment of cutting. Both sides of that comparison need to be at 20 °C, the reference temperature used in most drawings and ISO standards.

• 1
  Spindle growthAxial shift of the tool tip, typically tens of micrometers.
• 2
  Ballscrew growthPosition error that increases along the axis travel.
• 3
  Column and frame tiltSquareness and straightness drift over the working volume.
• 4
  Cold measurementCMM and part must share the same reference temperature.

Machining heat does not only come from the spindle. Cutting energy goes into the chip, the tool and the part. In dry cutting of steel, most of it leaves with the chip, but in aluminum the part can take a meaningful share. A thin wall has little mass to absorb it, so it bows away from the cutter and springs back after the pass. The wall measures thick in the middle and thin at the ends.

Coolant helps, but it also creates its own gradient. Flooding one face of a plate while the other face stays dry bends the plate. The bow can reach 50 to 100 μm on a 300 mm aluminum plate with a 10 °C difference across the thickness. Rough and finish in the same setup, then let the part stabilize before the finishing pass.

Surface finish follows the same logic. A tool that grows in length changes its effective radius, so the feed marks change depth along one pass. Thermal drift of a few micrometers will not show on a Ra 1.6–3.2 μm as-machined face, but it can push a Ra 0.2–0.8 μm finished face out of spec.

• 1
  Thin wallsLow mass, high deflection, springback after the cut.
• 2
  One-sided coolantGradient across the thickness bends plates and shafts.
• 3
  Rough then finishLet the part settle between passes on tight work.
• 4
  Finish sensitivityFine finishes show drift that rougher faces hide.

The cheapest control is time. Run the spindle at production speed for 30 to 60 minutes before cutting tight features, so the machine reaches a stable state instead of drifting through the first parts. Log spindle temperature and let it flatten before the first measurement. On a 5-axis center, warm the rotary axes with a short exercise cycle, not just the spindle.

The second control is symmetry. Apply coolant to both sides of a plate, or cut dry with air blast. Keep the part in the same thermal environment from roughing to finishing. Where the geometry allows, leave 0.3 to 0.5 mm for the finishing pass and take it after the part has cooled.

For work we cannot hold by time alone, we machine close to the reference temperature and probe the part on the machine. Probing gives the actual position of a datum, and the control offsets the remaining error. On a 4,000 mm part, that matters more than any compensation table, because the error changes along the length of the travel.

• 1
  Warm-up30–60 minutes at cutting speed before tight features.
• 2
  SymmetryCool both faces, or cut dry with air blast.
• 3
  Rough and restLeave 0.3–0.5 mm, finish after the part stabilizes.
• 4
  Probe on the machineMeasure the datum, then offset the remaining error.

Not every tight-looking tolerance is a thermal problem. If a bore is out of round by 40 μm on a machine that has been warm for hours, look at the spindle, the toolholder or the fixture first. Thermal error tends to be directional and repeatable, not roundness-shaped. A warm machine shifts a position, it rarely makes a hole lobed.

If the same program gives good parts in the morning and bad parts in the afternoon, and the shop has no air conditioning, temperature is the first suspect. If the error appears on one feature and not on the one next to it, it is usually clamping or tool deflection.

There is also a cost line. A temperature-controlled room and a 60-minute warm-up only pay off when the tolerance is below about ±0.01 mm on long features. For a ±0.05 mm bracket, the same money is better spent on fixtures and probing. Choose the control that matches the tolerance, not the one that sounds most advanced.

• 1
  Directional driftPoints to temperature or screw growth, not to roundness.
• 2
  Local errorPoints to clamping, tool runout or fixture stiffness.
• 3
  Time-of-day patternPoints to ambient swings in the shop.
• 4
  Cost lineThermal control pays below roughly ±0.01 mm on long parts.