In This Article
One of the most confusing failures in sealed aluminum castings is a part that appears sound as a blank, may even pass a blank-stage leak check, and then starts leaking after CNC machining. This is usually not a purely casting problem or a purely machining problem. It is the interaction of internal casting discontinuities, machining stock, final wall thickness, sealing interfaces and test sequence.
Why leakage after machining is a typical sealed-casting problem
Aluminum castings can contain internal discontinuities. ASTM E155 provides reference radiographs for discontinuities in aluminum and magnesium alloy castings, while ASTM E505 covers aluminum and magnesium die castings and includes categories such as porosity and shrinkage.
An internal discontinuity does not necessarily reach the surface or create a through-path in the blank.
A typical sequence is:
internal porosity exists -> sound metal still separates both surfaces -> blank does not leak -> CNC removes stock -> defect is opened -> continuous path forms -> final leak failure
Industry guidance from the Metal Casting Institute also notes that leakage in aluminum castings can result from porosity in wall sections being opened by machining cuts.
Separate gas porosity, shrinkage and cracks
| Discontinuity | Typical characteristic | Sealing risk |
|---|---|---|
| Gas porosity | Rounded voids associated with gas | Can become leak paths when opened |
| Shrinkage porosity | Irregular void network from insufficient feeding during solidification | Can create connected paths in hot sections |
| Crack | Continuous fracture | More direct route to through-leakage |
All three can appear as “Leak NG” after machining, but their casting root causes differ.
A useful failure report should record defect type, defect location, distance from final machined surfaces, whether it is through-connected, associated thick/hot section, and the machining operation after which leakage first appears.
Why a defect can be invisible at the blank surface
Sealing is about connectivity, not simply whether any void exists.
If a void is located inside a 5 mm wall while intact metal remains on both sides, the blank may be completely leak-tight.
After 1.2 mm is removed from one side, the void can be exposed. If the internal void network connects to the enclosure cavity, a leak path can form.
The important question is therefore not only:
Is porosity present?
It is:
Where is the porosity, how much metal remains to the final machined surface, and can it form a continuous path?
Why more machining allowance is not automatically safer
Excess machining stock can increase the probability of opening internal porosity.
| Excess stock effect | Possible consequence |
|---|---|
| More material removed | Greater chance of opening internal voids |
| Longer machining time | Higher cost |
| More stress redistribution | Flatness and thin-wall distortion |
| Lower final wall | Reduced structural margin |
| Deeper cutting near hot spots | Entry into shrinkage-prone regions |
Too little stock is also risky because blank variation may not clean up and critical datums may remain incomplete.
Machining allowance should be established from:
casting capability -> blank tolerance -> defect distribution -> machining stock -> final wall -> CTQ
Why flanges and O-ring regions are high-risk zones
Sealed housings often receive significant machining around flanges:
- large sealing-face cleanup;
- O-ring grooves;
- fastener holes;
- fittings and plug holes;
- final datum surfaces.
These zones combine high material removal with high sealing sensitivity.
A flange leak can therefore come from either a porosity path or geometric sealing problems such as warpage, groove error or surface damage.
What can X-ray tell us?
ASTM E155 and E505 provide radiographic references to classify and communicate casting discontinuities and severity.
X-ray is useful for identifying internal volumetric discontinuities, screening high-risk blanks and comparing defect severity.
Its limitation is that exact depth location can be ambiguous in a two-dimensional projection. A visible pore may be close to the final machined surface or located near the middle of the wall.
Why industrial CT is better for defect-to-machining correlation
Industrial CT can provide three-dimensional data for defect location, size, shape and relation to the final CAD surface.
The analysis can be framed as:
CT defect cloud -> overlay final 3D geometry -> calculate removed stock -> identify defects that will be exposed
This is especially useful for new-tool validation, high-value sealed housings, repeated leakage in the same region and machining-stock optimization.
CT still does not replace final leak testing because a visible defect and an actual through-leak path are different questions.
Leak testing answers the final question
X-ray and CT characterize internal structure.
Leak testing asks:
Is there an actual measurable path from the pressure side to the low-pressure side?
| Method | Primary question |
|---|---|
| X-ray | Are significant internal defects present? |
| CT | Where are defects and how close are they to final surfaces? |
| Leak test | Is a through-leak path actually present? |
The methods are complementary.
Why final leak testing must follow critical machining
A blank may pass a leak check and then undergo flange machining, O-ring grooving, fitting-hole drilling and thin-wall machining.
Any of these operations can change the sealing boundary.
The process plan should identify:
What is the last machining operation capable of changing the pressure boundary?
The final leak gate should be after that operation.
If O-rings, fittings or plugs are installed later, an additional assembly leak test may be required to validate the complete seal chain.
Leak, proof pressure and burst are not the same test
| Test | Purpose | Destructive? |
|---|---|---|
| Leak test | Detect a leakage path | No |
| Proof pressure | Confirm integrity at specified pressure | Normally no |
| Burst test | Determine ultimate failure pressure | Yes |
Passing one does not automatically prove the others.
The specification should define medium, pressure, hold time, temperature where relevant and allowable leak rate.
Test medium changes what can be detected
Possible media include dry air, nitrogen, helium, water or a customer-specified fluid.
Different media have different physical properties and detection methods.
Therefore:
A casting that does not visibly leak water is not automatically proven gas-tight to a stricter quantitative criterion.
The RFQ should define:
test medium + pressure + time + allowable leak rate + method
rather than simply “leak test required.”
Machining damage can also create leaks
Not every post-machining leak is caused by casting porosity.
| Machining issue | Leak risk |
|---|---|
| Scratch or tool mark on seal face | Bypass around O-ring |
| Incorrect groove geometry | Insufficient seal compression |
| Thread drilled too deep | Breakthrough into cavity |
| Deep-hole drift | Low local wall thickness |
| Aggressive deburring | Damage to sealing edge |
| Coating dimensional change | Seal-fit change |
| Fixture damage | Local seal-face defect |
Leak analysis must distinguish casting-induced leakage from machining/assembly-induced leakage.
How to identify casting versus machining root cause
A practical method is to correlate leak-path location with process data.
| Observation | More likely direction |
|---|---|
| Same casting hot spot repeatedly leaks | Casting shrinkage/porosity |
| Leak rate rises after a tool change | Machining factor |
| Leakage around O-ring perimeter | Seal face/groove/assembly |
| Leakage near deep drilling | Bore drift, thin wall or porosity |
| Strong lot-to-lot difference | Casting batch |
| Leak appears only after coating | Coating/interface factor |
An effective 8D links leak location to casting position + CNC operation + tool + blank batch.
Can impregnation solve casting leakage?
Vacuum impregnation and other porosity-sealing processes are used in industry for some porous castings.
They should not be treated as a default substitute for casting quality.
The engineering review should confirm whether impregnation is permitted, whether the defect is sealable porosity rather than structural damage, media and temperature compatibility, interaction with later surface treatment, process timing and traceability.
Cracks, severe shrinkage or insufficient wall thickness cannot simply be converted into good structural parts by impregnation.
Build a defect map instead of treating every failure separately
For stable production, historical data can be combined into a defect map.
| Data | Purpose |
|---|---|
| X-ray/CT defect coordinates | Identify high-risk regions |
| Leak location | Identify through-path hot spots |
| CNC cut depth | Correlate machining exposure |
| Final wall thickness | Find weak regions |
| Mold cavity number | Detect cavity-specific behavior |
| Casting batch | Track process drift |
| Leak rate | Measure severity |
| Scrap/rework | Quantify quality cost |
The useful question after several months is:
Where does this part tend to leak, which cavity and machining depth are associated with it, and why?
That is production quality engineering.
How leak gates can evolve from prototype to production
| Stage | Main focus |
|---|---|
| Prototype | Identify thin-wall and defect-risk regions |
| DVT | Correlate X-ray/CT with final machining |
| Pilot | Establish final leak test and failure localization |
| Mass production | 100% or defined-frequency leak gate + SPC/traceability |
Not every part requires X-ray or CT on every unit. A common economic strategy is to use NDT for process qualification and high-risk screening while final leak testing verifies the finished function.
Typical quality chain
casting-process design
-> material/batch verification
-> blank dimensional and visual inspection
-> X-ray/CT where required
-> rough machining
-> defect-exposure review
-> critical sealing/structural machining
-> deburring and cleaning
-> flange/O-ring/interface CTQ inspection
-> final leak test
-> proof pressure when required
-> surface treatment/assembly
-> assembly leak test when required
-> data traceability
The objective is not to add tests indiscriminately. Each gate should address a specific known risk.
Failure-cause-countermeasure matrix
| Failure | Possible cause | Response |
|---|---|---|
| Sudden leak after CNC | Internal porosity opened | Defect map + stock optimization |
| Flange-area leak | Shrinkage or flatness/O-ring problem | NDT + sealing CTQ |
| Leak near deep bore | Bore drift, thin wall, porosity | Bore position + wall control |
| Lot leak rate rises | Casting-process drift | Blank-batch SPC |
| One mold cavity fails more | Local filling/solidification issue | Cavity traceability |
| Random distributed leakage | Gas-porosity variation | Melt/casting-process control |
| Leak after coating | Interface/coating/assembly change | Post-process re-test |
| Tester results unstable | Fixture/equipment problem | MSA, calibration, fixture maintenance |
RFQ inputs for sealed castings
| RFQ input | Why it matters |
|---|---|
| Casting route | Different defect mechanisms |
| Alloy/heat treatment | Material and distortion |
| 3D + 2D | Final machined surfaces |
| Final wall thickness | Structural/defect risk |
| Machining stock | Direct relation to defect exposure |
| Sealed medium | Leak-test relevance |
| Working pressure | Functional boundary |
| Allowable leak rate | Acceptance |
| Test medium | Detection sensitivity |
| Proof/burst requirement | Structural validation |
| X-ray/CT requirement | NDT gate |
| Impregnation permission | Repair strategy |
| O-ring/seal design | Sealing CTQ |
| Surface treatment | Interface dimensions |
| Annual volume | Test automation |
| Traceability | Batch/cavity/process linkage |
If an RFQ only says “must not leak” without defining medium, pressure and allowable leak rate, the requirement is not sufficiently defined for manufacturing acceptance.
Conclusion
A sealed cast-aluminum housing that leaks only after machining is fundamentally a case where internal casting discontinuities intersect the final machining boundary.
The effective solution is not simply to demand “zero porosity.”
It is to manage:
defect type -> defect location -> machining stock -> final wall thickness -> sealing interface -> leak gate
Mature production uses X-ray/CT data, CNC depth, leak location and casting-batch traceability to build a defect map and remove high-risk regions from design and process rather than repeatedly repairing final failures.
FAQ
Why can a cast-aluminum housing be leak-tight as a blank but leak after machining?
Internal gas porosity, shrinkage or microcracks may exist without reaching either surface. CNC machining can remove material that previously isolated the defect and create a continuous path to the enclosure cavity or outside. Blank-stage leak results therefore cannot replace final leak testing after critical machining.
What is the difference between X-ray, industrial CT and leak testing?
X-ray is mainly used to identify volumetric casting discontinuities and severity. Industrial CT adds three-dimensional defect location, size and relation to machined surfaces. Leak testing answers a different question: whether an actual through-path exists. The methods are complementary rather than interchangeable.
Does leaving more machining allowance prevent leakage in cast housings?
Not necessarily. Excess stock increases material removal and residual-stress redistribution and may make it more likely that internal porosity is opened. Machining allowance should be set from casting capability, blank tolerance, defect distribution, final wall thickness and CTQs rather than simply maximized.
At what stages should leak testing be used in production of sealed aluminum castings?
Depending on risk, leak-test gates can be placed at the blank stage, after critical rough machining, after final machining and after assembly. High-risk sealed parts should normally receive final leak verification after sealing faces, O-ring grooves, ports and other major machining are complete, with additional gates defined by the product risk and customer specification.
