CNC machining in aviation

• 1
  Key takeaways
• 2
  What actually removes metal in an aviation cut
• 3
  Matching alloy to feature, not to habit
• 4
  Why setup count is the real cost driver
• 5
  Holding ±0.005 mm and proving it
• 6
  Where the process stops being the right answer
• 7
  From drawing to inspected part
• 8
  Which machine class fits the feature

Every cut is a balance between three things: the stiffness of the tool, the stiffness of the workpiece, and the heat the cut generates. In aviation work the workpiece is usually the weak link. Ribs 1.5 mm thick, thin-walled housings and long shafts deflect under cutting force long before the spindle runs out of power. A 12 mm carbide end mill in a 40-taper holder can push hard, but a 1.5 mm rib will bend away and spring back, leaving a taper the probe will find later.

The second mechanism is heat. Aluminium 6061 and 7075 carry heat away with the chip, so they cut fast and stay cool. Titanium TC4 (Ti-6Al-4V) does the opposite. It holds heat at the cutting edge, and it work-hardens if the tool rubs instead of shearing. That is why titanium feeds are aggressive rather than gentle: light passes on a blunt edge burnish the surface, raise hardness and shorten tool life.

Stainless 17-4PH and 15-5PH sit between the two. They machine cleanly when the tool is sharp and the coolant reaches the edge, and they distort when a heavy roughing pass leaves residual stress in the web. Precipitation-hardened grades also move during heat treatment, so the sequence of rough, treat, finish matters more than the finish pass itself.

Fixtures decide how much of this you can control. A part held on three points with a light clamp will chatter; the same part supported under the cut line and clamped with equal force cuts quietly at the same parameters. When a program behaves differently on two machines, the fixture usually explains it.

None of this changes the basic promise of the process. A computer-controlled machine repeats a programmed path within ±0.005 mm on a well-supported part, and it does so on the first part and the four-hundredth. The engineering question is whether the part is stiff enough for that path to mean anything.

The alloy list for aircraft hardware is short and familiar. Aluminium 6061-T6 covers brackets, panels and housings. 7075 carries higher load in the same envelope but machines with a gummier chip and needs sharper geometry. 2024 sits between them and is common on structural fittings. Stainless 303 and 304 handle corrosion-exposed hardware, while 17-4PH and 15-5PH take the higher-strength brackets and actuator parts.

Titanium TC4 and Ti-6Al-4V appear wherever weight and temperature meet. They cut at roughly one third the surface speed of aluminium and demand rigid setups. Inconel raises the temperature ceiling again and is normally reserved for engine-side hardware where the geometry is simple enough to machine without constant re-clamping.

Magnesium AZ31B and AZ91D cut very fast with sharp tools and high feed, but the chip is a fire risk. Fine magnesium swarf has to be collected wet or handled under a controlled program, and it should never sit dry in a bin. That is a shop-floor constraint, not a design one, but it affects price and scheduling.

Composites behave differently again. Carbon fibre laminates are abrasive, so tool life is measured in meters rather than hours, and delamination at the exit face is the defect to watch. Most composite parts are trimmed and drilled rather than milled to shape.

A practical rule: pick the alloy the load case needs, then check the drawing for the one feature that alloy is worst at. Titanium is fine until the wall is 1 mm. Aluminium is fine until the part needs hardness at a wear surface.

A part that comes off in one setup is cheaper, straighter and easier to inspect than the same part spread over four operations. Every re-clamp adds a positional error that stacks on the machining tolerance. On a bracket with holes on two faces, a 3+2 setup on a five-axis machine can hold hole-to-hole position in one coordinate frame; the same part on a three-axis machine needs two fixtures and a datum transfer.

For long parts, the practical limit is the machine envelope. Our largest travel is 4,000 × 400 × 150 mm, which suits long spars, rails and stringer sections. Medium work sits in the 750 × 1,150 × 550 mm and 600 × 600 × 600 mm class, and compact housings fit the 500 × 500 × 450 mm and 500 × 310 × 200 mm machines. A Ø400 mm rotary table handles round flanges and hubs in one continuous path.

Five-axis simultaneous cutting is not automatically better. It pays off when the feature is genuinely non-prismatic: blade roots, ported housings, compound-angle faces, or a part that would otherwise need six setups. For a flat plate with a pattern of holes, a three-axis machine with a good fixture is faster and easier to verify.

Tool access decides the rest. A pocket deeper than four times its width needs a long, thin tool that deflects. That is the point where the design should change, not the program. Widening a corner radius from 2 mm to 4 mm can remove a whole operation and a probe check.

A tolerance is a promise about measurement as much as about cutting. Our shop holds ±0.005 mm (about ±0.0002 in) on supported features, and surface finish from Ra 0.2–0.8 μm on sealing faces down to Ra 1.6–3.2 μm on as-machined surfaces. Those numbers only hold where the feature is rigid enough to be touched by a probe without moving.

Temperature is the first obstacle. A 300 mm aluminium part grows about 0.007 mm over a 10 °C swing. If the drawing tolerance is tighter than the shop's thermal control, the inspection result depends on when it was measured. Good practice is to let the part stabilize before final inspection and to record the temperature alongside the numbers.

The second obstacle is datum choice. Aviation drawings often call out a datum that is not a practical clamping face. When that happens we agree on a functional datum in the DFM review, machine to it, and report the relationship to the drawing datum so the engineer can close the loop.

Inspection covers raw material checks, in-process monitoring and a final pass before shipment. Reports are available on request. Where a feature cannot be measured without a CMM stylus, we say so early rather than promising a number the shop cannot verify.

CNC machining is subtractive, so it competes with its own chip pile. A part that removes 80 percent of its billet by volume is usually a casting or a forging with a light finish pass. Aluminium die casting and vacuum casting cover that shape at far lower cost per part, and the machined surfaces can be finished afterward to the tolerance that matters.

Very thin walls and tall unsupported webs are the second boundary. Below roughly 0.8 mm in aluminium and 1.5 mm in titanium, cutting force and residual stress dominate, and the yield loss climbs. Sheet metal fabrication handles uniform-thickness panels better, and it does so without a billet.

Third is quantity. For one prototype to a few hundred parts, machining needs no tooling and no minimum order quantity, which is why it dominates early flight-test hardware. Above roughly 10,000 parts a year, the tooling cost of a casting or forging amortizes and the per-part price inverts.

Surface treatment is a related boundary. Anodizing, plating, powder coating and laser marking all change dimensions slightly, and a hardcoat anodize layer on a thread can make a nut bind. Call out the finish before the final pass, not after.

The honest summary: machining wins on low volume, tight tolerance and complex geometry. It loses on high volume, uniform thin sections and shapes better formed than cut.