Leak Testing for Welds: Methods, Standards & Acceptance Criteria

Leak testing is the acceptance test that answers a question none of the other weld NDT methods can: will this joint actually hold pressure, or exclude contaminants, once it is in service? A weld can pass radiographic testing with a clean film, pass magnetic particle testing with zero surface indications, and still leak — a tight lack-of-fusion plane or a pinhole porosity cluster can be too small for volumetric or surface methods to flag reliably, yet large enough to pass gas under load. For welding engineers and quality managers responsible for pressure vessels, piping systems, and sealed fabrications, leak testing is the functional check that closes that gap.

This guide covers the four methods fabricators actually use, the codes that govern them, how to choose the right one for a given application, a step-by-step procedure, the failure modes that produce bad calls in the field, and how leak testing fits alongside volumetric and surface NDT in a complete weld quality program.

Key takeaways

• Leak testing verifies the functional tightness of a welded joint under service pressure — it complements volumetric and surface NDT, it does not replace them.
• Four methods cover nearly all applications: hydrostatic, pneumatic, vacuum box/bubble, and tracer gas (helium mass spectrometer) — sensitivity rises by several orders of magnitude from hydrostatic to helium.
• ASME Section V, Article 10 is the primary US governing standard; ISO 20485 and EN 1779 fill the same role internationally.
• API 510 and API 570 set in-service leak-test requirements for pressure vessels and piping respectively, and ASME PCC-2 governs post-repair testing.
• Method selection depends on required sensitivity, part geometry and accessibility, the applicable code, and whether the vessel can safely be pressurized with gas at all — pneumatic testing carries stored-energy risk that hydrostatic testing does not.
• The most common field failure is treating a passing hydrostatic test as proof of leak-tightness for a service that actually needs helium-level sensitivity.
• Acceptance criteria and personnel qualification (ISO 9712 or ASNT SNT-TC-1A) both have to be fixed in a written procedure before testing starts, not decided after an ambiguous indication appears.

Table of contents

1. What leak testing verifies in a welded joint
2. Leak testing methods: hydrostatic, pneumatic, vacuum box, and tracer gas
3. Standards governing weld leak testing
4. Choosing the right method
5. Step-by-step leak test procedure
6. Common failure modes and how to fix them
7. Validating and verifying leak test results
8. Where leak testing fits in a broader weld quality program

What leak testing verifies in a welded joint

Every other weld NDT method asks a version of the same question: does a discontinuity exist, and how big is it? Leak testing asks a different question entirely — does the joint, as built, actually pass gas or liquid under the pressure and conditions it will see in service? That distinction matters because the two questions do not always have the same answer. A weld can carry a discontinuity too small to reject on a radiograph or too tight to register on an ultrasonic scan, yet still form a continuous leak path once the assembly is pressurized. Conversely, a joint can carry a rejectable-looking indication on a surface method that has no through-wall path at all.

Leak testing is applied at three points in a fabrication’s life: after initial construction (code-mandated hydrostatic or pneumatic testing before a pressure vessel or piping system is placed in service), during in-service inspection intervals under an owner-operator’s inspection program, and after any weld repair, where a localized leak test on the repair area substitutes for re-testing the entire system. Each context has a different governing standard and a different acceptable sensitivity, which is why method selection is not a single fixed choice — it depends on what stage of the asset’s life the test is verifying.

Leak testing methods: hydrostatic, pneumatic, vacuum box, and tracer gas

Hydrostatic testing fills the vessel or piping system with a liquid, almost always water, and pressurizes it to a multiple of design pressure — typically 1.3x for ASME Section VIII vessels — while inspectors check welds and connections for visible weeping or pressure drop. It is the default code-mandated test for new pressure vessels because liquid’s low compressibility means a rupture releases far less stored energy than a gas-filled system at the same pressure, making it the safer choice for the largest pressure excursion a fabrication typically sees.

Pneumatic testing uses compressed gas instead of liquid, usually air or nitrogen, and is reserved for systems that cannot tolerate liquid residue — food, pharmaceutical, and some low-temperature or vacuum-service piping. Because compressed gas stores far more energy than liquid at the same pressure and volume, pneumatic testing carries materially higher risk in the event of a sudden failure, and codes impose extra precautions (lower test pressures, exclusion zones, staged pressurization) that hydrostatic testing does not require.

Vacuum box and bubble testing covers a family of low-sensitivity, low-cost techniques: coating a weld with a soap solution and applying pressure or vacuum, or immersing a small assembly in liquid and watching for a stream of bubbles. Under ASTM E515, the normal sensitivity limit for bubble emission techniques is on the order of 1x10^-5 std cm3/s — coarse compared to tracer gas methods, but fast, inexpensive, and adequate for locating gross leaks or for a quick field check where a laboratory-grade instrument is not practical.

Tracer gas testing introduces a gas the test equipment can detect at very low concentrations — almost always helium — and uses a mass spectrometer or a handheld sniffer probe to locate where it escapes. Under ASTM E499’s detector-probe method, sensitivity reaches roughly 1x10^-8 std cm3/s or finer, several orders of magnitude beyond a bubble test, which is why tracer gas methods are the default for cryogenic vessels, vacuum equipment, hermetically sealed enclosures, and any application where even a very small leak is unacceptable.

Standards governing weld leak testing

ISO 20485 superseded the earlier EN 13185 tracer-gas standard, and the European method-selection framework in EN 1779 (which compares bubble, pressure-change, and tracer-gas techniques on sensitivity and applicability) is still widely referenced for choosing between them even where hydrostatic testing is excluded from its scope. On the personnel side, ISO 9712 qualification covers leak testing as a certifiable NDT method alongside eddy current, ultrasonic, and radiographic testing, generally excluding straightforward hydraulic pressure tests from the certification requirement.

Choosing the right method

No single leak test method is correct for every application. The decision comes down to four variables: required sensitivity, whether the part can safely be pressurized with gas, part geometry and accessibility, and the governing code or contract requirement.

Where a weld needs both a leak-tightness verdict and confirmation of internal soundness, leak testing is typically paired with a volumetric method rather than substituted for one — see radiographic testing weld acceptance criteria or phased array ultrasonic testing (PAUT) for weld inspection for the volumetric side of that pairing, and welding inspection methods compared for how all the common methods sit alongside each other on a single fabrication.

Step-by-step leak test procedure

1. Confirm the written procedure and required sensitivity against the governing code (ASME Section V Article 10, API 510/570, or ISO 20485) before testing — method, test medium, test pressure, hold time, and acceptance criteria all have to be fixed in advance, not decided on the shop floor.
2. Verify mechanical integrity first. A leak test pressurizes the assembly; any weld, support, or restraint not qualified for the test pressure has to be checked or excluded before pressurization begins, and a hydrostatic or pneumatic test should never proceed without a documented pressure-relief and exclusion-zone plan.
3. Prepare the test medium and instrumentation. For hydrostatic testing, fill slowly and vent trapped air; for tracer gas testing, calibrate the mass spectrometer or sniffer against a certified reference leak standard immediately before use.
4. Pressurize in stages, holding at intermediate steps to check for gross leaks before reaching full test pressure — this catches a major defect before it becomes a hazard at full pressure rather than after.
5. Hold at test pressure for the specified dwell time, allowing the system to stabilize thermally and mechanically; a premature reading during pressure or temperature drift produces a false leak-rate result.
6. Scan all welds, connections, and closures systematically, following a documented scan pattern so no joint is missed — for bubble testing, watch continuously rather than making a single pass; for tracer gas testing, hold the sniffer probe at the specified standoff distance and scan speed.
7. Record and disposition every indication against the acceptance criteria in the governing code or contract, including instrument calibration data, ambient conditions, and the final pass/fail determination, before depressurizing and releasing the assembly to service.

Common failure modes and how to fix them

Treating a hydrostatic pass as leak-tight for a service that needs finer sensitivity. A vessel that holds water at 1.3x design pressure can still leak at a rate unacceptable for vacuum, cryogenic, or hermetic-seal service — hydrostatic testing was never designed to catch that scale of leak. Fix: match the test method’s sensitivity to the actual service requirement during procedure planning, not after a field failure exposes the gap.

Uncalibrated or drifted tracer-gas instrumentation. A mass spectrometer or sniffer that has not been checked against a certified reference leak standard can under-report or over-report leak rate by an order of magnitude or more. Fix: calibrate immediately before use against a traceable reference leak, and re-verify at defined intervals during extended test campaigns.

Background contamination masking or mimicking a tracer-gas signal. Residual helium in the test area from a previous test, or a poorly purged system, produces a false background reading that either masks a real leak or triggers a false-positive call. Fix: purge the test area and allow background levels to stabilize before starting, and record baseline background as part of the test report.

Skipping the staged-pressurization step. Bringing a system straight to full test pressure without intermediate holds means a gross defect is discovered at the point of maximum stored energy rather than at a safer, lower pressure. Fix: always stage pressurization with documented hold points, regardless of how routine the test seems.

Testing before mechanical qualification is confirmed. Pressurizing an assembly whose supports, restraints, or temporary closures were not verified for test pressure has caused documented equipment failures unrelated to the weld itself. Fix: complete and sign off a mechanical-readiness checklist before any pressure is applied, separate from the leak-test procedure itself.

Validating and verifying leak test results

A leak test result is only as defensible as the calibration and documentation behind it. Verification should run at three levels:

• Pre-test instrument calibration against a certified, traceable reference leak standard, re-checked at defined intervals through extended campaigns and whenever probe, cable, or instrument settings change.
• Independent confirmation of borderline indications, either with a repeat test at the same sensitivity or with a more sensitive method (escalating from bubble to tracer gas) before an indication is accepted or rejected.
• Full documentation of test conditions — medium, pressure, hold time, ambient temperature, background readings, and calibration records — retained as part of the fabrication’s quality record, not just the final pass/fail line.

Programs that record only a final pass/fail result, without the underlying test conditions and calibration trail, are the ones that cannot defend a leak test result during an audit or a later field failure investigation — the corrective action becomes re-testing from scratch instead of reviewing a documented trail that would have settled the question in minutes.

Where leak testing fits in a broader weld quality program

Leak testing answers “does this finished joint hold pressure and exclude contaminants as built?” It does not answer “was this weld made with the right parameters?” — that is a process-control question, and a joint can pass a leak test cleanly while still having been welded with drifted heat input, an unstable arc, or a cooling rate outside the qualified window that produces a defect only later, in service, after thermal or mechanical cycling opens what the leak test could not yet detect.

That is why finished-weld leak testing and in-process monitoring are complementary rather than substitutable. Real-time welding monitoring system coverage — tracking thermal signature, arc stability, and travel speed as the weld is deposited — closes the process-control gap that a post-weld leak test, run after the fact, cannot see. Where the concern is specifically the thermal history of a pressure-boundary joint, a thermographic inspection plan provides a complementary look at heat input and cooling behavior that a leak test’s pass/fail result does not capture on its own. For fabrications governed by pressure vessel welding quality requirements or ASME B31.3 process piping, converging the leak-test record with the same traceability record that captures in-process thermal weld monitoring data means a quality manager can answer an auditor’s question about a joint’s functional tightness and its welding process from one record set, instead of reconciling two separate paper trails after the fact.

FAQ

What is leak testing in welding inspection? Leak testing (LT) verifies that a welded joint or fabricated assembly is tight enough to hold pressure or exclude contamination at its intended service condition. Unlike radiographic, ultrasonic, or magnetic particle testing, which look for discontinuities in the weld metal itself, leak testing checks functional integrity — whether any flaw, however small, actually creates a through-path for gas or liquid under load.

What are the main leak testing methods for welds? The four methods most fabricators use are hydrostatic testing, pneumatic testing, vacuum box or bubble testing, and tracer gas testing. Sensitivity increases roughly in that order, with tracer gas methods detecting leaks orders of magnitude smaller than a hydrostatic test can.

What standard governs leak testing of welded pressure equipment? ASME Boiler and Pressure Vessel Code, Section V, Article 10 is the primary US standard. ISO 20485 and EN 1779 fill the same role internationally, API 510 and API 570 govern in-service pressure vessels and piping, and ASME PCC-2 governs post-repair leak testing.

Is hydrostatic testing enough to qualify a welded pressure vessel? For most new-construction ASME code vessels, yes — hydrostatic testing at 1.3x design pressure is the code-mandated qualification test. Applications needing tighter margins, such as cryogenic or vacuum service, commonly add a supplementary helium test even after a passing hydrostatic result.

What is the difference between a bubble test and a helium mass spectrometer test? A bubble test under ASTM E515 has a normal sensitivity limit around 1x10^-5 std cm3/s. A helium mass spectrometer test under ASTM E499 reaches roughly 1x10^-8 std cm3/s or finer — several orders of magnitude more sensitive, at the cost of specialized equipment.

Can leak testing replace radiographic or ultrasonic weld inspection? No. Leak testing only detects a through-wall path large enough to pass gas or liquid under test conditions; it cannot see a tight lack-of-fusion plane or a crack that has not yet broken through the wall. Volumetric methods remain necessary to characterize internal weld soundness.

Who can certify a leak test on a code-governed weld? ISO 9712 and the ASNT SNT-TC-1A recommended practice both define leak testing as a certifiable method with Level I-III tiers. Most contracts require the test procedure and report to be signed off by a certified Level II or higher technician.

Further reading on leak testing fundamentals and calibration is available from ASNT’s overview of leak testing as an NDT method, NIST’s leak artifact calibration services, and ISO’s standard catalog entry for ISO 9712 personnel certification.