Weld Defect Identifier
Tick what you can see and get the likely defects — with the pairs that look alike and need opposite fixes flagged before you make it worse.
Tick what you can actually see. Nothing is listed unless one of its characteristic symptoms matches — a diagnosis built from weak signals alone is worse than none.
On the surface
Cracking
Found on cutting, X-ray or ultrasonics
While welding
Most likely
These two look alike and want opposite things
Both of these match what you described. Getting the wrong one does not simply fail to help — the cure for each makes the other worse.
The pairs worth learning as pairs
What makes weld diagnosis hard is not that the defects are obscure. It is that the obvious response to what you can see is as likely to be backwards as right, and three pairs account for most of it.
- Undercut and overlap. Both are toe defects and both read as "a bad edge". Undercut is too much heat — metal melted out of the base and never replaced. Overlap is too little — weld metal rolled over base metal it never fused to. And overlap is the more dangerous, precisely because it looks stronger: undercut announces itself as a groove, while overlap hides a completely unfused interface under a fat, healthy-looking bead.
- Porosity from too little gas and from too much. Identical holes. Past a certain flow rate the shield goes turbulent and starts drawing the surrounding air into the pool, so cranking the flowmeter is as likely to deepen the problem as fix it. The tell is where: starved shielding is worst at the start of a run and in a draught; turbulent shielding is uniform the whole way down and improves when you turn the flow down.
- Hot and cold cracking. The diagnostic is when it appeared. Solidification cracking is there as soon as the weld cools, usually down the centreline, and is cured by a wider, shallower bead — more penetration makes it worse. Hydrogen cracking turns up hours or days later, at the toe or in the heat-affected zone, and is cured by preheat and dry consumables. A crack you didn't see yesterday is hydrogen.
And the one rule with no exceptions: any crack is rejectable. AWS D1.1 puts it plainly — any crack is unacceptable regardless of size or location. Almost every other defect here has an allowable limit you can measure against. Cracks have none.
How to use
- Tick the symptoms you can actually see on the weld.
- Read the ranked defects, most likely and most serious first.
- Check the conflict panel before adjusting anything.
- Change one variable at a time on a test coupon rather than on the job.
Frequently asked questions
What is the difference between undercut and overlap?
They are opposites with opposite cures, which is why they are worth learning as a pair. Undercut is a groove along the weld toe caused by too much heat — base metal melted out and never replaced — so the fix is less current or more filler. Overlap is weld metal rolled over the toe caused by too little heat, a pool never hot enough to wet into the base, so the fix is more current. Both look like a bad edge, and applying either fix to the other deepens the problem.
Which is worse, undercut or overlap?
Overlap, and precisely because it looks stronger. Undercut announces itself as a visible groove and has measurable allowable limits. Overlap presents as a fat, generous-looking bead while hiding a completely unfused interface underneath, so it passes a glance and fails a load. It is a fusion defect rather than a shape one, which puts it in a different seriousness category altogether despite being the one that looks healthier.
Will more shielding gas fix my porosity?
Possibly, and possibly the opposite. Too little gas and too much gas produce the same holes: past a certain flow rate the shield stops being laminar and starts drawing the surrounding air into the pool. The tell is where the porosity appears. Starved shielding is worst at the start of a run before the gas establishes, and worst in a draught. Turbulent shielding is uniform down the whole weld and improves when you turn the flow down. If in doubt, turn it down and run a test piece.
How do I tell hot cracking from cold cracking?
By when it appeared, which is the single most useful discrimination in the subject. Solidification or hot cracking is there as soon as the weld cools and usually runs down the centreline of the bead. Hydrogen or cold cracking turns up hours or days later, usually at the toe or in the heat-affected zone. If you did not see it and it is there the next morning, it is hydrogen. The two want opposite things, so getting this wrong sends you in exactly the wrong direction.
Why does more penetration make centreline cracking worse?
Because a deep narrow bead solidifies from both sidewalls toward the middle, and the two fronts meet in a plane down the centre. The last liquid to freeze carries the low-melting impurities with it and ends up trapped exactly there, which is a ready-made crack. The cure is a wider, shallower bead — a width-to-depth ratio comfortably above one — which feels wrong because it means deliberately backing off the penetration.
Is a small crack acceptable?
No. AWS D1.1 states it without qualification: any crack is unacceptable regardless of size or location. Almost every other defect has an allowable limit you can measure against — porosity has generous allowances, undercut has a permitted depth — but cracks have none at all. A crack found in a weld is a crack that comes out, and the reason is that a crack is a ready-made stress concentration that will grow rather than sit still.
What is the difference between lack of fusion and lack of penetration?
They get used interchangeably and should not be. Lack of fusion is weld metal touching the sidewall or the previous pass without bonding to it, and it can happen in a joint with perfect penetration. Lack of penetration is the root never being reached at all. One is a bonding problem and the other a depth problem, so opening the joint preparation fixes the second and does nothing whatsoever for the first.
Why is a flat defect worse than a round one of the same size?
Because it concentrates stress the way a crack does. A round pore has the load flow around it smoothly; a flat unfused interface across the stress direction behaves like a pre-existing crack tip. That is why lack of fusion and lack of penetration are treated far more seriously than porosity of comparable size, and why ultrasonics matters more than radiography for finding them — X-rays can miss a planar flaw edge-on almost entirely.
🔒 This tool runs entirely in your browser. Nothing you enter is uploaded, logged, or stored.