The Gauge Passed on The Machine. A day Later It Had Moved

Both gauges came off the five-axis and checked out clean. Every dimension passed on the CMM at the granite table that same day. We moved them into the temperature-controlled room and let them sit. Twenty-four hours later we re-measured. Same machine, same parts. Flatness and bushing positions had all gone out of tolerance. Nothing had touched them overnight. The stress still locked inside the parts finished the job we thought was done

1.The part

The customer builds multi-station positioning and inspection equipment in the US. The gauge is a large aluminum plate, roughly 720 mm across the bushing pitch circle, cut through with several openings, with pressed SUJ2 (AISI 52100) steel bushings setting the datum holes. Body in Al 6061-T6. Two pieces. The openings are part of why the plate flexes so easily. The bushings locate every reading the machine takes, so bore position and plate flatness are the part.

We machined it from 6061 plate with pressed steel bushings on purpose. A gauge like this stands on dimensional stability and a clean assembly datum. Sheet metal will not hold a 720 mm plate flat to 0.08 mm or place a ±0.005 mm bore. Casting brings porosity and its own residual stress into a part that has to stay put for years. Printing will not hold the bore or the pitch circle. Machined 6061 with pressed SUJ2 carried the tolerance and the wear surface together. This ran as a two-piece low-volume build, so both gauges had to come out identical, not just one good one.

Why steel bushings and not a reamed aluminum hole: 6061 galls and wears at the datum after a few hundred locates, and this gauge gets located far more than that over its life. SUJ2 (around 60 HRC, typical material value, not measured on this part) takes the wear and keeps the hole round. Drop the steel bushing on a low-cycle fixture that only gets located a few hundred times. There a reamed aluminum hole holds fine, and the bushing just adds cost.

2.What failed

We ran the first set through the full sequence. Basic stress relief, in-house bushings checked and in spec, liquid nitrogen cold assembly, five-axis finishing, off-machine inspection passed. After 24 hours of rest, three things had moved:

• Flatness went to 0.148 mm against a 0.08 mm limit. The thin plate sprang back and domed.
• Pitch circle drifted to ±0.015 mm against a ±0.008 mm limit. The bushings followed the body as it deformed.
• Bore deviation ran to 0.016 mm against a ±0.005 mm tolerance (0.010 mm band). The bushings tilted and went oval.

The lesson we took: on a thin plate this size, an off-machine pass does not prove a stable part. The stress that moves it bleeds out over the hours after the cut. Sign off same-day only on a thick, rigid block, where the off-machine reading holds.

3.Root cause

We drilled into why, and it came back to stress that had nowhere to go yet.

• The liquid nitrogen cold fit sat too short before machining. The interference stress had not balanced out. We put the part on the five-axis while it was still loaded, and that stress kept releasing afterward and pulled the bushings off position.
• Vacuum pressure on the five-axis fixture ran too high. The thin plate stretched under the clamp. That clamping stress stacked on top of the cutting and assembly stress and drove the deformation.
• 6061 expands about 23.6 µm/m per °C. SUJ2 expands about 11.9, close to two to one. The aluminum-steel interface carried a standing residual stress, and every swing in shop temperature pumped it.

So on an aluminum-steel interference joint this thin, we now rest the assembly at controlled temperature before finishing. The interference stress needs hours to redistribute before you lock in geometry. A thick, rigid boss can resist the load, so there you can finish sooner and skip the wait.

4.What we changed

• Two-stage stress relief. A 96-hour age after roughing, then a 120°C, 6-hour low-temperature soak after semi-finish, so the second stage drains stress the first age cannot reach. Where the stress relief falls in the machining sequence is what decided whether the plate held.
• Held the assembly at constant temperature for 24 hours after cold fit, until the interference stress released, before five-axis finishing.
• Reworked the five-axis fixture. Multi-point support, lower vacuum pressure, less clamping load written into the plate.
• Fine-bored the bushing bores with low-vibration parameters to pull out the small tilt and ovality.
• Ran the whole job at 20 ± 1°C, then rested the finished part 12 hours and re-measured before delivery.

The 20°C hold is not housekeeping. That is the ISO 1 reference temperature for dimensional measurement. On an aluminum-steel part with a two-to-one expansion gap, read it at any other temperature and the number moves on you.

One more call on the fixture: on a thin plate, multi-point support at lower vacuum won over a single high-vacuum zone. The high clamp load deforms the plate itself, and you machine that stress straight into the geometry. On a thick, stiff plate the high vacuum is fine, and multi-point support only adds setup time.

5.Results, measured on this part

The remade set held. These are measured on this part, not pulled from a datasheet.

Flatness	0.148 mm	0.042 mm	0.08 mm
Pitch circle	±0.015 mm	Φ719.995 to 720.006 mm	±0.008 mm
Bore	0.016 mm	within ±0.005 mm	±0.005 mm

Flatness holding means the plate keeps its datum face, so every station reading off it stays true. The pitch circle landing inside the band means the bushing ring sits where the drawing puts it across the full 720 mm span. The bores back in tolerance mean each bushing locates square, not tilted. For how we read flatness, position, and bore callouts against what a shop can actually hold, see our tolerance and GD&T standards.

Client outcome The client kept the part design as-is. After the process rework, both gauge sets passed full dimensional and accuracy acceptance in a single submission. The customer’s inspection line went into production on its original 12-day install schedule, with no slip to equipment start-up.

6.What we carry into the next part like this

• On large thin aluminum precision parts, a rough-cut age alone will not hold. Add a low-temperature stress relief after semi-finish.
• After liquid nitrogen cold fit, do not machine right away. Let the interference stress release, or the bushing walks later.
• On thin-plate five-axis finishing, drop the vacuum pressure. High clamp load writes its own stress into the part.
• On aluminum-steel interference parts, rest and re-measure after finishing. Lagging stress shows up hours later, not on the machine.

The engineer’s read on why this class of part bites:

“It is large, it is thin, the tolerances are tight, and the plate is cut through with openings. Any stress left inside springs back and puts the part out of tolerance.”

Machining a large thin aluminum part that has to stay flat after it leaves the machine? Send the drawing and get a quote with DFM feedback, and we will flag the stress and fixturing risk before we cut.