Porosity Defect in Welding: Types, ISO 5817 Acceptance B/C/D

Porosity is the most common volumetric weld defect: gas that gets trapped in the weld pool and remains as cavities after solidification. A few small pores rarely compromise a statically loaded joint — but porosity above the acceptance limit reduces the effective load-bearing section, acts as an initiation site under fatigue, and is one of the most frequent causes of NDT rejection and rework.

This guide covers what porosity is, the types defined by ISO 6520-1, what actually causes it on the shop floor, the ISO 5817 acceptance limits per quality level, and how to detect and prevent it.

What is porosity in welding?

Porosity (ISO 6520-1 group 2 — cavities) is a gas-filled cavity in the weld metal, formed when gases — hydrogen, nitrogen, oxygen, or shielding gas itself — are absorbed by the molten pool and cannot escape before the metal solidifies.

The main types classified by ISO 6520-1:

The distinction matters because acceptance limits differ by type: aligned porosity is treated more severely than scattered pores, since a line of cavities behaves more like a planar discontinuity.

Gas Pore vs Porosity: Same Defect, Different Names

“Gas pore” and “porosity” are used almost interchangeably on the shop floor, but they sit at different levels of the same classification. ISO 6520-1 defines the isolated single cavity — the one most welders picture when they say “gas pore” — under reference 2011 (single gas pore). “Porosity” is the umbrella term for the whole family of cavity imperfections the standard groups together: the single pore (2011), uniformly distributed porosity (2012), clustered porosity (2013), linear porosity (2014), and the elongated cavity or wormhole (2015/2016), plus the surface-breaking pore (2017) covered above.

So a weld can be “porous” — meaning it shows porosity — without every cavity in it being a textbook single gas pore, and a single gas pore is porosity but is not automatically the more serious distributed or clustered form. The distinction is not academic: acceptance decisions and repair choices both key off which type is present, since a scattering of isolated pores is treated far more leniently by most codes than a linear chain of cavities along the same fusion line.

This naming looseness causes real friction in nonconformance reports. An inspector who writes “gas pore” on an NCR when three aligned cavities are actually present under-communicates the severity to whoever reads the report later, since linear porosity is scored more strictly than an isolated pore under most quality levels. When logging a finding, record the ISO 6520-1 reference code alongside the plain-language term rather than instead of it — the code disambiguates what a reader might otherwise take for a minor, isolated defect.

Search terms like “gas pore,” “gas pores in welding,” “porous weld,” and “porosities in welding” all describe points on this same spectrum. For the formal definition and the mechanism behind why gas gets trapped in the first place, see the porosity glossary entry; for the cause matrix and inspection route tied to the isolated single-pore case specifically, see the gas pore (2011) defect reference.

What causes porosity

Practically all porosity comes down to one of three sources:

1. Shielding problems

• Gas flow too low — or too high, causing turbulence that aspirates air
• Drafts in the welding area (doors, fans, ventilation above ~2 m/s)
• Clogged or spatter-loaded nozzle, damaged diffuser
• Wrong torch angle or excessive stick-out increasing air entrainment

2. Contamination

• Moisture, oil, grease, rust, or paint on the joint
• Zinc coating (galvanized steel) vaporizing into the pool
• Damp electrodes or flux — hydrogen from moisture is also the driver of cold cracking
• Condensation on cold plates brought into a warm shop

3. Process parameters

• Arc voltage too high (longer arc = more air entrainment)
• Travel speed too fast — the pool freezes before gas can escape
• Power source dynamics mismatched to transfer mode, producing erratic short circuits

A useful diagnostic rule: scattered porosity across many welds points to gas or consumables; porosity repeating at the same location points to contamination or a draft at that position.

ISO 5817 acceptance limits

ISO 5817 sets limits for porosity per quality level (B = stringent, C = intermediate, D = moderate). The reference values for the two most common cases:

t = nominal material thickness, d = pore diameter. Clustered (2013) and linear (2014) porosity carry separate, tighter limits. These values are indicative references — always verify against the current edition of ISO 5817 and any overriding code or customer specification.

For the full picture of how quality levels work and which one your project requires, see the ISO 5817 acceptance criteria guide and the Level B vs C vs D comparison.

How porosity is detected

• Visual testing (ISO 17637) — only catches surface-breaking pores (2017).
• Radiographic testing (ISO 17636) — the reference method: pores appear as rounded dark indications, easy to size and count for the area-percentage limits.
• Ultrasonic testing — detects subsurface porosity but sizing scattered small pores is harder than with RT.
• Real-time process monitoring — thermal imaging and high-speed vision detect the conditions that generate porosity (shielding loss, erratic transfer, pool disturbance) while the weld is being made. A weld-monitoring camera like HeatCam IR-C provides a live thermal view for this type of monitoring. See how this works in practice: real-time porosity detection with thermal imaging.

The economics are simple: a pore found by RT after the fact costs an NCR, excavation, and re-weld. A shielding-gas fault flagged in real time costs one interrupted seam. A welding monitoring system that watches thermal signature and arc behavior live flags that same excursion as porosity risk while the seam is still open, not days later when RT finds the pore. The full comparison of inspection approaches is here: AI weld defect detection — thermal vs vision vs acoustic.

For the visual signature of a weld heading for rejection, watch when a weld goes wrong — the erratic pool pattern real-time monitoring is built to catch.

Prevention checklist

1. Gas: verify flow at the nozzle (not just the regulator), check for leaks in hoses and fittings, shield from drafts.
2. Cleanliness: degrease and remove rust/paint/zinc at least 25 mm from the joint edges.
3. Consumables: store dry, respect re-drying procedures for fluxes and basic electrodes.
4. Parameters: keep arc length/voltage at WPS values; reduce travel speed if porosity persists.
5. Monitor: track the process continuously so shielding or contamination drift is caught on the first weld, not at final NDT — this is exactly the gap in-process monitoring closes. Browse the real-time porosity monitoring hardware to see the cameras and analysers that run this check inline.

Can a Porous Weld Be Repaired?

In most fabrication codes and quality systems, porosity is repairable, and repair is the routine response rather than the exception. Outright rejection of the part is reserved for cases where the cavity affects too much of the cross-section to repair economically, or where the specification bars repair on that particular joint or component class outright.

The repair sequence is consistent across most codes: excavate the affected length back to sound metal by grinding, air-arc gouging, or machining; verify the cavity is fully removed with a visual check (and a dye-penetrant or magnetic-particle check on critical joints, to confirm no residual indication remains at the base of the excavation); then re-weld the prepared cavity following the same qualified WPS used for the original joint, not an improvised parameter set. The repair weld is then re-inspected by the NDT method that found the original porosity, since a repair that has not been re-verified is not a closed finding.

Two practical constraints matter more than the mechanics of the repair itself. First, most welding codes limit how many times a given location can be repaired before the joint design itself, not just the welding, is considered suspect — check the governing code rather than assuming unlimited attempts are available. Second, a repair changes the local heat input history and can introduce new risks — hardness increase in the heat-affected zone, distortion from repeated thermal cycling — that the original weld did not carry, which is why repairs on critical materials are frequently qualified as their own procedure rather than treated as a like-for-like re-run of the production WPS.

Repeated porosity at the same joint location across multiple repair cycles is itself a signal worth escalating. See the ISO 5817 acceptance criteria reference for how the applicable quality level frames what counts as an acceptable outcome after repair, and treat a joint needing more than one repair attempt as a process investigation, not just a rework ticket.

Catching Porosity While the Weld Is Being Made

During multi-pass welding, porosity forms when gas is trapped in the cooling pool or between passes. Traditional quality control catches it after the fact—at the inspection bench, under radiography or ultrasound, sometimes weeks after the weld was made. By then, the pass is cold, the machine setup has moved on, and the cost of rework or scrap is committed.

A thermographic camera observes the weld pool and the solidification signature as they happen in real time. The thermal profile of a sound weld follows a predictable cooling curve for the base metal, shielding gas, and travel speed in use. When gas is trapped—when the pool collapses asymmetrically, or when cooling is interrupted by voids—the thermal pattern deviates from that signature. The deviation appears during the pass, not at the NDT line.

This is not a detection method in the sense of radiography or ultrasound: it does not image the porosity directly or measure its size. It is a process-monitoring window. It answers the question a quality engineer asks in real time: “Is the weld pool behaviour consistent with sound metal formation right now?” An anomaly during the pass becomes a flag to repeat that section, adjust parameters, or investigate the conditions that produced it—while the workpiece is still hot and the setup is still live.

The thermal signature also creates a per-weld record. Every pass is logged with its thermal data: pool temperature, cooling rate, and any anomalies that appeared during solidification. That record becomes the evidence an ISO 3834 or ISO 17635 auditor asks for when inspecting a lot: “Show me what the process looked like when this weld was made.” It is not a substitute for radiography or UT on critical applications, but it is the process evidence that justifies the inspection plan.

HeatCam IR-C is built for exactly this on a fixed mount: a stationary cell, machine or production line, watching every pool and logging every signature. HeatCam IR-S is the embedded variant, for WAAM, DED and robotic heads. Compare thermal and conventional weld monitoring camera approaches, and see real-time porosity detection with thermal imaging for a worked example. For the full platform view, see the welding camera hub.

From a Porosity Finding to an Audit Record

When a thermal monitor flags a potential defect signature during the pass, the next step depends on the application and the acceptance criteria. In high-reliability work—pressure vessels, aerospace, critical infrastructure—the part moves to volumetric NDT (radiography or ultrasound) for confirmation. The thermal record goes with it as process evidence, not as a final verdict.

In process-control applications, the same thermal data is the primary evidence. A fabrication shop working to ISO 3834 requirements on structural welds may use in-cycle thermal monitoring as the real-time process control, with periodic spot checks by conventional NDT to validate the monitoring. The shop logs every weld’s thermal signature and every control point (pool temperature, cooling time in a given window, whether any anomalies were flagged), and uses that log as proof that the welding process was under control during the production run.

ISO 17635, the standard for NDT method selection, requires that every weld inspection plan name which NDT methods apply at which testing extent. Thermal monitoring fits into that plan not as an NDT method itself (it is not listed in ISO 17635), but as the real-time process evidence that supports the inspection plan. Combined with periodic radiographic or UT spot checks, it creates a layered quality record: continuous process visibility during production, plus final acceptance testing on a planned sample.

The evidence chain works like this: the thermal baseline for your WPS is captured during procedure qualification; on production, every weld is monitored against that baseline; any excursion is flagged and either corrected immediately or marked for enhanced inspection; final results are recorded against the quality level (ISO 5817 B/C/D) and the code requirements. The auditor can then ask “show me a weld that hit porosity risk” and you can pull the thermal data, the correction action, and the final inspection result for that specific weld. That is process traceability. Without it, you have only inspection results and no evidence of process control.

The welding camera hub covers how this data is logged in real time and stored by weld ID, making it auditable and traceable. Every weld is connected to its thermal record, its parameters, and its outcome. That record is what makes an ISO 3834 or ISO 17635 quality audit verifiable—not guesswork or retrospective stories, but the actual behaviour of the process when the weld was made.

Porosity is one entry in the broader defect landscape — for the complete map of welding defects and their acceptance criteria, see the welding defects guide.

Therness solution: Welding monitoring system — Every weld inspected in the cycle, with defect flags and the ISO 3834 / ISO 17635 evidence kept per weld.

Written by Luca Santoro , founder