Guide
Which leak testing method fits a machined aluminum casting, and how tight should the spec be?
Leak testing is the family of checks that tell you whether a sealed part lets a fluid or gas pass through its walls, joints or ports, and, in the better methods, how much passes in a given time. On an aluminum casting the question is almost always porosity: pores left by casting that only become a path to the outside once a drill, end mill or tap cuts into them.
By Saola Manufacturing engineering team · Published · Updated
This guide is written for the engineer who owns a pressure-tight housing, such as a regulator body, valve body, meter case, pump cover or manifold, and has to choose a test, write the acceptance limit on the drawing and agree it with a supplier. It works from the ASTM practices that define each method, not from any one test-equipment maker’s catalog, so the numbers you see are the limits those standards state for themselves.
On this page
What leak testing methods are used on machined aluminum castings?
Four methods cover nearly all cast housings: bubble testing, pressure decay (air decay), hydrostatic testing with a liquid, and tracer-gas testing with helium. They differ in whether they find where the leak is, measure how big it is, or both.
ASTM E432, the guide ASTM publishes for choosing among leak test methods, lists the methods for locating leaks and for measuring them in order of increasing sensitivity, and asks the engineer to weigh sensitivity, cost and reliability when picking one (ASTM E432). That order matters for castings. A bubble test shows the exact pore that weeps but gives no number. A pressure-decay test gives a number but cannot say where the leak is. Helium can do both, at a cost in fixtures, gas and cycle time that only makes sense when the leak limit is tight.
| Method | ASTM practice | Locates or measures | Sensitivity stated in the scope | Where it fits on castings |
|---|---|---|---|---|
| Bubble emission (immersion or soap solution) | E515 | Locates; the standard says a quantitative measure is not practical | 1 × 10−5 std cm3/s (4.5 × 10−10 mol/s), normal limit | Finding the leaking pore; failure analysis; low-volume parts |
| Hydrostatic (liquid inside, under pressure) | E1003 | Locates by visible leakage | Procedures limited to leaks of 1 × 10−4 std cm3/s (4.5 × 10−9 mol/s) or larger | Water-service parts; proof testing where the medium is a liquid |
| Pressure decay (air decay) | E2930 | Measures total leak rate from pressure, temperature and time | Not stated; depends on test volume, pressure, time and temperature stability | Production go/no-go on every part |
| Helium tracer, sniffer probe or accumulation | E499/E499M | Locates (probe) or measures (accumulation) | 1 × 10−7 Pa·m3/s or greater | Tight gas limits; finding leaks too small for bubbles |
Two things in that table surprise people. The first is that a hydrostatic test, the one many drawings default to, is the least sensitive method listed; it is a strength and gross-leak check, not a fine-leak check. The second is that the most common production method, pressure decay, has no sensitivity figure in its standard at all, because the answer depends entirely on how the test is set up.
How do you compare leak rates stated in different units?
Convert everything to one unit before you compare a drawing limit with a tester reading or a standard’s threshold. One std cm3/s of leakage reads 60 in sccm, about 1 in mbar·L/s and about 0.1 in Pa·m3/s, a spread of roughly 600 times, which is easy to lose in a quote exchange.
US drawings often use std cm3/s (also written atm·cc/s or sccs) or std cm3/min (sccm). European and helium-equipment documents usually use mbar·L/s, and the SI unit is Pa·m3/s. ASTM E1003 pairs its threshold with a molar flow, which is useful because molar flow does not depend on which reference temperature someone assumed. The E499/E499M scope also warns that its SI and std-cc values are not exact equivalents and should not be mixed within one test, which is good advice for any drawing note.
| Unit | Equal to 1 std cm3/s | Common use |
|---|---|---|
| std cm3/s (atm·cc/s) | 1 | US drawings, ASTM inch-pound documents |
| std cm3/min (sccm) | 60 | Air-decay and mass-flow tester displays |
| mbar·L/s | about 1.01 | Helium leak detectors, European documents |
| Pa·m3/s | about 0.101 | SI; ASTM and ISO leak testing standards |
| mol/s | about 4.5 × 10−5 | ASTM E1003 and E515 thresholds |
Why does a casting pass as cast and then leak after machining?
Because the dense skin that forms against the die or mold seals most as-cast porosity, and machining removes that skin. Drilling a port, facing a sealing surface or tapping a thread can open a path through pores that were closed before.
This is why the test belongs after final machining, and why a leak result from the foundry is not a substitute for one on the finished part. MIL-STD-276A, the public defense standard for impregnating porous castings, requires parts to be completely machined before the preliminary pressure test and retested if ports or bolt holes are cut afterward (MIL-STD-276A). The same logic applies whether or not your part will ever be sealed.
Leakers from machining tend to cluster. If most rejects fail at the same port or the same milled face, the cutter is breaking into a heavy section where shrinkage porosity collects, and a design change (coring the boss, moving the port, evening out the wall) usually fixes more than any test or sealant. If rejects are scattered, the casting process itself is the place to look. NADCA’s design guidance makes the economic point that pressure testing every casting and impregnating only the failures keeps cost down (NADCA, secondary processing), which also keeps the reject rate visible as a process signal. The ADC12 vs A380 guide covers how alloy choice affects castability, and the die casting page describes the ADC12 aluminum and Zamak 3 zinc part families we source from partner foundries and machine at our lead CNC plant.
When is air decay enough, and when do you need helium?
Air decay is enough when the leak limit is loose enough to resolve reliably in the cycle time you can afford, which in practice covers most water and low-pressure housings. Move to helium when the limit is too small for air decay to separate from temperature noise, or when you must find the exact leak location on a failed part.
Pressure decay reads a leak as a falling pressure in a sealed volume, and per ASTM E2930 the conversion to leak rate rests on the ideal gas law using measured pressure, temperature and time. That makes temperature the enemy. Working the ideal gas law that ASTM E2930 relies on, a trapped charge of air near room temperature that cools by 1 °C loses roughly 0.3% of its absolute pressure (1 K out of about 293 K) with no leak at all, which can be larger than the signal from a small leak. Castings come off a machining center or a parts washer warm, and a large internal volume dilutes the pressure drop, so the practical limit of air decay is set by part volume, thermal stability and cycle time more than by the instrument.
Helium changes the physics. The detector responds to the tracer gas itself, not to a pressure change, so part volume and temperature matter far less, and ASTM E499/E499M places its detection threshold in the 10−7 Pa·m3/s range (ASTM E499/E499M). The cost is fixturing, gas recovery or consumption, and a slower cycle. A common production pattern is air decay on every part and helium or bubble testing on the rejects, to locate leaks and feed the results back to casting and machining.
The mistake to avoid is specifying a tighter limit than the function needs. Every step down in allowed leak rate moves you toward a more expensive method and more rejected castings, and ASTM E432 orders methods by sensitivity and asks you to weigh sensitivity, cost and reliability together, so set the limit from the function first and let it pick the method.
How should you specify a leak test on the drawing?
Write the test medium, the test pressure and direction, the maximum allowed leak rate with its unit, and the stage in the routing at which the test applies. A note that says only “leak test” or “no leaks” cannot be verified, because every method has a threshold below which it reads zero.
A workable note has five parts. The medium: dry air, nitrogen, helium or water. The pressure and direction: internal pressure or vacuum, gauge or absolute, and whether it is the working pressure or a higher proof pressure. The limit: a leak rate in one unit, with the reference conditions if you use std cm3. The stage: after final machining and before or after finishing and sealing. And the method or standard, if you care how it is done, such as pressure decay per ASTM E2930 or bubble emission per ASTM E515.
Keep proof testing and leak testing separate in the note. A proof test asks whether the part survives a pressure above its working pressure without yielding or bursting; a leak test asks how much escapes at a stated pressure. MIL-STD-276A, for example, has impregnated parts proof tested to the applicable drawing, and only when the drawing is silent does it default to air under water (or a brushed soap solution) at double the working pressure, not less than 10 psig, with a liquid-fill test at the same pressure as the alternate. A drawing that borrows that clause without a leak-rate number has specified a no-visible-leakage criterion, whatever tester is used. The RFQ preparation guide shows where test notes belong in a quote package, and the tolerances guide covers the general-tolerance notes that sit beside them.
What changes for gas and water distribution parts?
The medium changes the method and the limits. Water-service parts are usually proof tested with liquid and leak tested with air, while fuel-gas parts normally carry tighter gas leak limits and rule out some sealing options entirely.
For water parts, a hydrostatic test matches the service medium and doubles as a strength check, but it leaves the part wet. Internal water in an aluminum housing that is then boxed for ocean freight is an invitation to staining and corrosion, so the drawing or purchase order should require drying, or use dry air decay for the production leak test and reserve water for proof or sample testing. Any product standard your part must meet will set its own pressures; put those on the drawing rather than leaving the supplier to look them up.
For gas parts, air or helium testing is closer to the service condition, and limits tend to be tighter. Check early whether the part may be impregnated at all: MIL-STD-276A prohibits impregnation with organic sealants for parts exposed to oxygen at any pressure (sodium silicate, being inorganic, is the stated exception), and forbids it for parts welded afterward or run above the sealant’s rated temperature. The gas and water distribution page covers the part families in that sector, and the quality page describes inspection practice.
How do you split leak-test responsibility with an overseas supplier?
Decide in writing which operation owns the leak test, which method and limit it uses, and how that result is correlated with your own receiving test. Most disputes over leaking castings come from two parties running different tests and both being right.
When casting, machining and finishing move to Vietnam or elsewhere in Southeast Asia, they may sit in one plant or several. Either way, ask for three things before the first production lot. First, a master part or a calibrated reference leak, so the supplier’s tester and yours can be checked against the same known leak. Second, a defined retest rule, since a part that fails, is re-clamped and then passes should be logged, not quietly shipped. Third, lot-level reject counts, so you can see a casting process drift before it reaches your line.
If the part is die cast and machined in the same supply chain, the test can sit right after final CNC milling and before surface finishing, which is the order the standards above point to. Saola’s ISO 9001 certification is in progress, not held, so if your program requires a registered quality system, say so in the RFQ and plan the audit accordingly. To review a pressure-tight part, send the drawing with its leak-test note, or write to rfq@saolamfg.com.
Frequently asked
Is a hydrostatic test more sensitive than an air test?
Usually not. ASTM E1003 limits its hydrostatic procedures to fairly large leaks, and a well-set-up air or helium test can detect much smaller ones. Hydrostatic testing is the right choice for proof pressure and for water service, not for fine leaks.
Can a pressure-decay tester tell me where the leak is?
No. It measures the total leak of the sealed volume. To locate the leak, follow up with a bubble test or a helium sniffer probe on the rejected part.
Should the leak test come before or after anodizing or coating?
Test after final machining and before finishing, then seal any rejects if impregnation is allowed. A coating can temporarily bridge a pore and hide a leak that opens later in service. The finishing comparison guide covers finish choices for cast aluminum.
What does “zero leakage” mean on a drawing?
Only “no leak detectable by this method at this pressure.” Every method in ASTM E432 has a detection threshold, so state the limit as a leak rate or name the method whose threshold you accept.
Does impregnation replace leak testing?
No. Sealing closes leak paths, but the part still needs a test afterward to show it worked. MIL-STD-276A requires a proof test after impregnation.
Sources
- ASTM E432-91(2022), Standard Guide for Selection of a Leak Testing Method, ASTM International, reapproved 2022 (scope summary)
- ASTM E515-11(2022), Standard Practice for Leaks Using Bubble Emission Techniques, ASTM International, reapproved 2022 (scope summary)
- ASTM E1003-13(2022), Standard Practice for Hydrostatic Leak Testing, ASTM International, reapproved 2022 (scope summary)
- ASTM E499/E499M-11(2025), Standard Practice for Leaks Using the Mass Spectrometer Leak Detector in the Detector Probe Mode, ASTM International, reapproved 2025 (scope summary)
- ASTM E2930-26, Standard Practice for Pressure Decay Leak Test Method, ASTM International, 2026 (scope summary)
- MIL-STD-276A, Impregnation of Porous Nonferrous Metal Castings, Department of Defense, 30 December 1992 (public copy)
- Secondary Processing, NADCA Design, North American Die Casting Association, accessed September 2026