Porosity
Gas that got into the puddle and didn't get out before it froze.
The short answer. Porosity is contamination or lost shielding — those are effectively the only two categories. Something on the plate boiled off into the puddle, or the shielding gas or flux failed to keep the atmosphere out. Clean the joint properly and check your shielding before you touch any machine settings.
What it looks like
Round holes in the surface of the bead, from pinpricks to several millimetres across. Sometimes the surface looks fine and the porosity only appears once the bead is ground or cut — which is the whole problem with this particular defect.
The pattern tells you a great deal, so look at the distribution before the holes:
- Scattered — spread evenly along the bead. Usually general contamination or a marginal shielding problem present the whole time.
- Clustered — a group in one spot with clean weld either side. Something localised: a patch of paint, oil, or rust you missed, or a momentary gas interruption.
- Linear — a row following the joint line. Often contamination sitting in the root, or trapped in a gap.
- Wormholes (piping) — elongated, tapered holes, often at the start or the crater. Gas escaping through metal that was already freezing around it.
- At the start only — pre-flow. Your gas hadn't arrived yet.
- At the end only — post-flow, or you pulled the torch away while the crater was still molten and let air straight into it.
What causes it
Contamination on the plate
Paint, oil, grease, cutting fluid, rust, mill scale, galvanising, and above all moisture. All of them decompose in the arc and release gas directly into the puddle. Wire-brushing a rusty plate until it looks grey is not the same as cleaning it — grinding back to bright metal either side of the joint is. Galvanised coating is a special case: it must be ground off the weld area entirely, and the fumes are genuinely dangerous, so ventilate.
Lost gas shielding (MIG/TIG)
Any of these will do it, and they're worth checking in this order because it's roughly the order of likelihood:
- Draught. The most common cause outdoors and the most commonly missed one indoors. A fan, an open roller door, or a mate's grinder blowing across your work is enough to strip the shield. Solid-wire MIG in any real wind is a losing battle — that's exactly what flux-core exists for.
- Flow rate wrong. Too low leaves gaps in the shield; too high causes turbulence at the nozzle that drags air in. More gas is not more protection.
- Spatter blocking the nozzle, giving you a shield full of holes.
- Leaks anywhere from bottle to torch — a split liner or a loose fitting will suck air in on the low-pressure side while still showing normal flow at the gauge.
- Nozzle too far from the work, or excessive stickout, so the gas has dispersed before it reaches the puddle.
Damp or damaged electrodes
Low-hydrogen rods — 7018 and family — absorb moisture from the air, and that moisture goes straight into the weld as hydrogen. This is why they live in a rod oven and not in the boot of a car. A 7018 that's been out on a humid bench all weekend will give you porosity, and worse, it can give you hydrogen cracking days later in something that looked perfect when you finished it. Flux-core wire picks up moisture the same way once the spool is open.
Arc length and technique
Too long an arc pulls atmosphere into the puddle regardless of how good your gas is. On stick, whipping too far ahead can let the puddle pick up air behind the shield of the flux. Both of these are worth checking once you've eliminated cleanliness and gas.
How to work out which one is yours
- Only at the start — increase pre-flow, or hold the torch on the joint a moment before striking.
- Only at the end/crater — increase post-flow and hold the torch over the crater until it solidifies.
- Clustered in patches — it's the plate. Clean further back than you think you need to.
- Evenly scattered, indoors, no draught — flow rate, a blocked nozzle, or a leak. Check the nozzle first, it's free.
- Evenly scattered, outdoors or near a door — draught. Screen the joint or switch to flux-core.
- Stick welding, no gas involved — rod condition and arc length.
Why this one is different from the rest
Surface porosity is the visible part of an invisible problem. Every other defect in these guides is a surface feature that a photograph can honestly assess. Porosity is not. The holes that broke through to the surface are a sample, not a count — and whatever caused them was acting on the entire weld pool, including all the parts buried inside the joint.
A bead showing three pinholes may be riddled underneath. A bead showing none at all may also be riddled underneath. This is the defect where "it looks fine" means least, and it is exactly why radiographic or ultrasonic testing exists. If the joint matters, it needs RT or UT and a qualified inspector — no photo, no app, and no amount of squinting at the cap will substitute.
The practical consequence for practice: treat any visible porosity as a signal to fix your process, not as a measurement of how bad the weld is. The correct response is never "only a few holes, that'll do" — it's to find the contamination or the leak, because the same cause is producing far more damage where you can't see it.
Related
- Running too cold — the other defect that hides below the surface.
- Undercut — by contrast, entirely readable from a photo.
- All defect guides