A thin-wall aluminium housing that measures inside tolerance in the fixture and outside it on the bench is the most common argument in aluminium bore work.
The bore is not moving for one reason; it moves because four sources of distortion were present at the same time and only one of them was addressed.
This page separates the four sources, gives the control for each, and sets a measurement rule that tells you which source you are looking at.
The four sources are clamping force, cutting heat, residual stress and wall thickness variation. Clamping and heat act while the part is held and released; residual stress acts as material is removed; wall variation makes all three worse at the thin sections.
The measurement rule is three points in time: in the fixture, after release, and after a temperature soak. Each answers a different question, and the pattern of the three results identifies the cause.
The bore tooling itself is covered on the PCD reamer page. This page covers what happens to the part around the bore.
Four sources of distortion
Clamping force is the first and the most frequently misdiagnosed. A thin wall clamped on an unfinished surface is held in a shape that the part does not have when free, and every measurement taken in that state describes the fixture.
Cutting heat is the second. Aluminium expands roughly in the range of twenty-three millionths per degree Celsius, which is an order of magnitude above steel, so a bore that is twenty degrees warmer than the gauge is a bore measured at a different size than the one that will be inspected.
Residual stress is the third. A cast or forged blank carries stress from solidification or from the mill, and removing material releases it unevenly. The side of the bore with more material removed moves further than the other, so the bore goes oval or the wall bows.
Wall thickness variation is the fourth, and it is a property of the casting rather than of the process. A wall that varies by a few tenths changes the stiffness and the heat capacity along its length, so the other three sources act unevenly.
The four are usually present together. A part with a thin wall, clamped hard, cut warm, from a blank with stress, will move in a way that no single-source explanation fits.
The clamping question
The question is not how hard to clamp but where the force goes. A thin-wall housing needs the clamping force to close a loop through the stiffest part of the part, not through the wall that is being machined.
Soft jaws machined to the part contour distribute the force over an area and reduce local deflection. They work when the part has a stable surface to grip and the wall is not the gripping surface.
An expanding mandrel or a diaphragm chuck grips from the bore or from a machined diameter, which is useful when the outside is the thin wall. The limitation is that the gripping surface has to be round enough for the mandrel to seat, so it usually requires a pre-machined reference.
Vacuum clamping spreads the force over a face and removes local point loads, which suits large thin plates and housings with a flat face. It provides little resistance to lateral cutting force, so it is usually combined with a light radial support.
The rule that matters is the same in all three cases: finish the surface that will be clamped before clamping on it, and never clamp on a surface that will be machined later.
Heat: the invisible offset
Heat enters the part from the cut and leaves it through the fixture, the coolant and the air. In a thin wall the path is short, so the wall reaches a higher temperature than the bulk of the housing and moves more.
Dry cutting and minimum-quantity lubrication remove the cooling that a flood system provides, so the part runs warmer and the offset is larger. High spindle speeds add heat at the bore and shorten the time available for it to dissipate.
The practical control is not to eliminate heat but to make it repeatable: a warm-up cycle before the first part, a stable coolant temperature, and a fixed dwell before measurement. Repeatability is what makes the offset a constant that can be compensated, instead of a variable that cannot.
The measurement consequence is a soak rule. A part measured immediately after machining is measured at an unknown temperature; a part measured after a fixed soak is measured at a known one. The machine-side checks that support this are in machine-side requirements before running a PCD reamer.
The thermal check itself, including how to separate spindle growth from part growth, is set out in the machine-side page rather than repeated here.
Stock allocation and operation order
Distortion is controlled by where the material is removed, not only by how. The principle is to remove most of the stock while the part is still stiff and to leave a small, uniform allowance for the finishing pass.
| Operation | Allowance principle | Purpose |
|---|---|---|
| Rough machining | Remove the bulk, leave a uniform allowance on all surfaces to be finished | Release most of the residual stress while the wall is still thick enough to resist it |
| Semi-finish | Equalise the allowance, typically a small fraction of the finish allowance | Make the finishing cut uniform in force, so the wall deflects consistently |
| Finish | Cut the remaining allowance in one pass at low radial engagement | Hold size and roundness with minimal force and minimal heat input |
The allowance split follows the same logic as any aluminium bore: enough to clean up the pre-hole, little enough to keep the force low. The relationship between allowance and pre-hole size is covered in reaming allowance and pre-hole size.
Operation order matters as much as the split. If the bore is finished before the mounting face, the face cut releases stress that moves the bore. If the part is clamped on the face that is finished last, the clamp mark lands on the sealing surface.
A workable order for a thin-wall housing is: rough all faces and the bore, release, semi-finish the mounting face, semi-finish the bore, finish the face, finish the bore.
The exact sequence depends on the geometry, but the principle is that the last cut on each surface happens after the stress that surface will see has been released.
The measurement rule
Three measurements at three times answer three different questions. Taking only one of them is what turns a solvable problem into an argument.
| Measurement point | Question it answers | If it fails here |
|---|---|---|
| In the fixture, immediately after the cut | Did the tool cut the intended size and form? | Tool, program or thermal compensation |
| After release, part free on the bench | Did the fixture hold the part in a different shape? | Clamping force, clamp position or support |
| After a temperature soak | Is the part stable, or is it still moving? | Residual stress, wall variation or design stiffness |
The pattern is more informative than any single reading. In the fixture good and after release bad is a clamping problem. Good in the fixture and good after release but bad after soak is a thermal problem. Good at both short times and drifting after soak is a stress problem.
Bad in the fixture is the easiest case: the tool or the program is wrong, and no amount of fixture work will fix it.
Record the three readings on the same part, with the time and the temperature, and the cause is usually visible in one cycle. Without them, each function can defend its own measurement, and all of them are correct.
What to change when the three measurements disagree
The pattern points to an action, and the action should be the cheapest one that fits.
When the in-fixture reading is good and the released reading is bad, change the clamping before changing the tool. Reduce the force, move the clamp away from the wall, add support opposite the cut, or machine soft jaws to the contour.
When both readings are good and the soaked reading drifts, the cause is stress or heat, and the answer is in the process rather than the fixture: a stress-relief step after roughing, a different blank condition, or a longer stabilisation before finishing.
When the in-fixture reading is bad, the tool or the program is the cause, and the fixture is a distraction. Check the tool wear pattern and the thermal compensation before touching the clamping.
When the readings scatter in a way that does not repeat, the wall thickness variation is the likely cause, and the answer is a casting or blank conversation rather than a machining one.
The casting-side view of the same family of problems
The same wall problem appears in a harder geometry on scroll plates, where a thin wall runs beside a deep involute groove; the feature-by-feature tool choice is in scroll compressor machining and tool selection.
is in porosity in aluminium die castings and reamed hole quality.The application context for these housings, including the features that typically drive the tolerance, is on the EV motor housing machining page.
Distortion is one of the two constraints that decide a thin-wall housing, and the housing family is one row of the automotive aluminium bore tool map.
What published sources say about this
These pages support the tolerance framework, the measurement logic and the tool behaviour used above. Each is cited for what it states, under its own conditions.
- Dimensioning and Tolerancing (ASME Y14.5): the form, orientation and position definitions that decide whether a distortion is a size, roundness or flatness failure.
- Reamers, machine reamers (toolmaker technical overview): reamer geometry and guide behaviour, the reference for what a reamer can and cannot correct in a distorted bore.
- Reaming and fine boring (toolmaker competence page): the finishing families, including the adjustable tool used where size has to be corrected after the pre-hole.
- Tools for fine machining bores (toolmaker interview): single-blade tools and their sensitivity to force and temperature variation in thin sections.
- What is Process Capability? (NIST/SEMATECH e-Handbook): process spread against specification, used here to judge whether a scattered bore size is a capability failure or a measurement artefact.
- Bias and variability (NIST/SEMATECH e-Handbook): the separation of a repeatable offset from a random spread, which is the difference between a thermal offset and a stress-driven drift.
- What are control charts? (NIST/SEMATECH e-Handbook): the time-ordered view that shows a drift within a shift, which a single-part measurement cannot reveal.
Outbound references are listed because the reasoning depends on them, not as endorsements. The thermal expansion value is an industry reference figure; no distortion amount is promised for a specific part.
Frequently asked questions
Why is the bore good in the fixture and bad on the bench?
Because the fixture is holding the part in a shape it does not have when free. The measured size describes the clamped state, not the part.
The fix is a clamping change or a measurement change, usually both; a tool change will not help.
How long should the part stabilise before measurement?
Long enough that the reading stops changing. Measure the same part at intervals after release and note when two consecutive readings agree.
That interval, not a fixed rule, is the soak time for that part and that shop temperature.
Does a PCD reamer reduce distortion?
It reduces the force and the heat compared with a carbide multi-flute tool, which reduces distortion rather than eliminating it.
If the wall is thin enough, the fixture and the stock split dominate, and the tool only sets the size.
Is a stress-relief step always necessary?
No. It is necessary when the blank carries stress that the finishing cuts would otherwise release asymmetrically.
The evidence is a part that is good at release and drifts after soak; without that pattern, the step adds cost for no benefit.
What if the three measurements disagree in no consistent pattern?
Then the wall thickness variation is the first suspect, because it makes every other source act unevenly.
Measure the wall at several positions and compare with the drawing; if it varies, the conversation belongs with the casting or blank supplier.
Updated 13 November 2026: first publication. Scope: the four distortion sources, the three clamping options, the thermal offset and soak rule, the stock allocation table, and the three-point measurement rule. The thermal expansion value is an industry reference; no distortion amount is promised.