High-altitude flight accuracy in CNC machined aerospace parts

• 1
  Why high-altitude flight accuracy starts with a temperature argument
• 2
  What the machine does to high-altitude flight accuracy before the part ever flies
• 3
  When 5-axis motion improves high-altitude flight accuracy and when it does not
• 4
  Material choice and its effect on high-altitude flight accuracy
• 5
  How we verify high-altitude flight accuracy before shipment
• 6
  Choosing the machine and material for a flight part

A machined bracket is inspected in a metrology room at 20 °C. It then sits in a wing bay at -55 °C, or next to a gearbox at 120 °C. The metal moves. Aluminium 6061 grows about 23 μm per metre per °C; titanium TC4 grows roughly 9 μm. Over a 400 mm bracket that is a 0.2 mm swing between cold soak and hot soak. High-altitude flight accuracy is not a single number on a drawing. It is the behaviour of a tolerance stack across a temperature range.

This is why we ask what the part does, not just what the print says. A hole pattern that is checked at 20 °C may be judged against a different datum when the airframe is cold-soaked on the ramp. If the mating part is titanium and the bracket is aluminium, the two materials separate at a different rate. The joint load can shift. Bolts that were sized for a room-temperature fit may not carry the same preload at altitude.

The practical answer is not to tighten every tolerance. It is to identify which dimensions carry the load path and which are cosmetic. A ±0.005 mm bore on a bearing seat earns its cost. A ±0.05 mm clearance on a cable clamp does not. Spending the tight tolerance where it does not matter is the most common way to raise the price of high-altitude flight accuracy without buying any.

• 1
  Check the datum firstA tight dimension referenced to the wrong face is still wrong at 11,000 m.
• 2
  Match the CTEMixed aluminium and titanium joints move at different rates under cold soak.

Most of the error in a machined aerospace part is created on the shop floor, not in the sky. Spindle growth, ballscrew heating and fixture movement all push the cutter away from the nominal path. On a long roughing cycle in 7075, a spindle can grow 15–30 μm within the first hour. If the finishing pass follows immediately, the first pocket and the last pocket of the same program are cut at different depths.

The usual countermeasure is a warm-up cycle and a rough-then-finish split. We rough with stock left on, let the machine and the part reach thermal equilibrium, then finish in a separate setup or a separate program. For thin webs under 1.5 mm, we also reduce radial engagement and use high-speed toolpaths so the cutting force stays low. Force is what pushes a thin rib out of position, not spindle speed.

Vibration is the other mechanism. A long, slender tool on a deep pocket will chatter. Chatter leaves a surface that looks fine under a fingernail but measures Ra 3.2 μm or worse, and the tool wears on one flute. For deep pockets we switch to a shorter tool, a larger diameter, or a reduced step-down. Sometimes the answer is a different toolpath, not a different machine.

High-altitude flight accuracy also depends on how the part is held. A fixture that clamps on a thin wall will deflect the wall while cutting and spring back after unclamping. We prefer to clamp on a thick boss or a dedicated tab, and to machine the datum faces first so every later operation references something that was cut in the same setup.

• 1
  Warm up before finishingLet the spindle stabilise, then cut the tight features.
• 2
  Rough and finish separatelyStock-on roughing keeps thermal drift out of the final pass.
• 3
  Clamp on stiff featuresThin-wall clamping causes spring-back after the part is released.

A 5-axis machine does not automatically make a part more accurate. It removes setups. Every time a part is unclamped and re-fixtured, the datum shifts by a small amount. On a part with six faces and three tight bores, that stacking of setup errors can dominate the tolerance budget. Cutting the same features in one continuous 5-axis setup removes most of that stack.

The second benefit is tool access. A compound-angle boss or a contoured flange that would need a special form tool on a 3-axis machine can be reached with a standard ball nose at the correct lead angle. The tool stays short and rigid. Surface finish improves because the cutter is not reaching around a corner.

There is a limit. Five-axis motion adds rotary axis error and post-processor risk. If a part is a simple plate with holes on one face, a 3-axis machine with a good fixture will hold the same tolerance for less money and with less programming risk. We quote 5-axis when the geometry or the setup count justifies it, not by default.

For parts that fit inside a 500 × 500 × 450 mm envelope, we often run them on a 4-axis mill with a Ø400 mm rotary table. That gives access to four faces in two setups and keeps the part on a smaller, stiffer machine. The choice between 4-axis and 5-axis is usually about face count, not about prestige.

• 1
  Count the setupsFewer setups means less datum stack-up.
• 2
  Keep the tool short5-axis access lets a rigid tool reach compound angles.
• 3
  Do not overspecifyA flat plate with simple holes does not need simultaneous 5-axis.

Titanium TC4 (Ti-6Al-4V) is common in aerospace because it keeps strength at temperature. It is also the material that punishes a weak setup. It work-hardens, it conducts heat poorly, and it pushes back on the tool. A light finishing pass with a sharp cutter works better than a heavy pass with a worn one. We keep a separate tool set for titanium so the edges stay sharp.

Aluminium 7075 machines fast and holds a good finish, but it is more sensitive to residual stress than 6061. A part hogged out of a thick plate can move after the last cut as the internal stress rebalances. For a thin, long part we may rough, stress-relieve, then finish. That adds a step, but it prevents a part that was accurate on the machine from being out of tolerance a week later.

Inconel and other nickel alloys are used where the part sees exhaust or high heat. They cut slowly and generate a lot of heat at the edge. High-altitude flight accuracy in these alloys comes from conservative parameters, rigid tooling and patience. Trying to run them at aluminium speeds is how tools break and parts get scrapped.

The material also affects inspection. A titanium part and an aluminium part of the same nominal size will not agree with a steel gauge at every temperature. We record the inspection temperature and report it, so the customer can compare the measurement against the design condition.

• 1
  Titanium rewards sharp toolsLight finishing passes beat heavy cuts on TC4.
• 2
  7075 can move after cuttingStress relief before finishing helps long thin parts.
• 3
  Record inspection temperature20 °C readings and flight-condition readings are not the same number.

Inspection is not a final gate. It is a series of checks. We check the raw material certificate before any cut, monitor the first article, and inspect the finished part against the drawing. A report can be issued on request. The point is to catch a drift before it reaches the last part in the batch, not after.

For critical features we use a CMM and compare against the CAD model rather than a hand gauge. A hand gauge tells you whether a hole is the right size; a CMM tells you whether the hole is in the right place relative to the datum. On a part with a true-position callout, that difference matters.

Surface finish is checked where the drawing calls for it. A Ra 0.8–1.6 μm finish is a normal machined finish for a sealing face or a bearing bore. A Ra 0.2–0.8 μm finish needs a different insert and a lighter pass, and it costs more. We only apply it where it is specified.

We hold a 99.99% qualification rate across production, and every part is inspected before it ships. That number is not a promise about any single order. It is a description of how the process runs when the drawing, the material and the fixture are all agreed before cutting starts.

• 1
  First article before the batchCatch drift on part one, not part fifty.
• 2
  CMM for positionHand gauges check size, not true position.