Fillet Weld Size & Strength
Throat, capacity and weld metal for fillet welds — and why rounding a size up buys 33% more strength for 78% more metal.
The weld
Materials and method
What each size actually costs
Strength climbs in a straight line and metal climbs as a square, so the value per kilogram deposited falls away steadily. The last column is the one to read.
Weld metal is quadratic, strength is linear
A fillet's cross-section is a triangle, so its area follows leg squared. Its capacity comes from the throat, which is 0.707 of the leg — so capacity follows leg. Step a 6 mm fillet up to 8 mm and you buy 33% more strength for 78% more weld metal, and you pay for that metal three times: in consumable, in arc time, and in the distortion the extra heat puts into the part. Rounding a fillet size up "to be safe" is the most expensive reflex in fabrication.
Which is why two small fillets beat one big one whenever you can reach both sides. Two 6 mm fillets give half again the throat of a single 8 mm for barely more metal, and they load the joint symmetrically instead of pulling it into a bend the way a single-sided fillet always does. If the joint still needs more after that, lengthening the weld is cheaper than deepening it — length is linear in both metal and strength.
- The throat of an unequal fillet is below both legs. It is (a × b) ÷ √(a² + b²) — the altitude of the triangle, not the average and not the smaller leg. A 6 × 12 mm fillet has a 5.37 mm throat, less than the 4.24 you'd get assuming a 6 mm equal-leg but also nothing like the 12.
- The minimum size in the code is a cooling rule, not a strength rule. Thick plate draws heat out of a small weld fast enough to crack it — the same argument that drives preheat. That's why the minimum is set by the thicker part joined, and why "but the load is tiny" is not a defence.
- The cost steps as well as curving. Most processes lay about 8 mm in a single horizontal pass. Past that you take on repositioning, interpass cleaning and inspection for every extra run, so the real cost curve has a step in it right where the square law is already biting.
- The base metal caps the joint. Once weld capacity passes the plate's shear rupture strength, more weld buys nothing at all — and neither does a stronger electrode. At that point the answer is a thicker plate or a longer weld.
- These are welds loaded in shear along their length, which is the common case and the conservative one. A fillet loaded transversely is roughly half again stronger, and both AISC and Eurocode have directional methods that take credit for it. Fatigue is a separate calculation and usually governs anything that moves.
How to use
- Enter the leg size, weld length and whether both sides are welded.
- Pick the electrode, base steel and design method.
- Enter both part thicknesses — they set the minimum and maximum size.
- Read the cost of the next size up before rounding anything.
Frequently asked questions
Why is oversizing a fillet weld so expensive?
Because weld metal follows the square of the leg while strength follows the leg. The cross-section is a triangle, so its area is leg squared over two, but capacity comes from the throat, which is only 0.707 of the leg. Stepping from a 6 mm fillet to an 8 mm one therefore buys 33 per cent more strength for 78 per cent more weld metal. You pay for that metal three times over: in consumable, in arc time, and in the distortion the extra heat puts into the part.
What is the throat and why does it matter more than the leg?
The throat is the shortest distance across the weld from the root to the face, and it is the plane the weld actually fails on. Drawings call out legs because that is what a gauge measures, but every strength calculation uses the throat. For an equal-leg fillet it is 0.707 times the leg. Getting this backwards overstates a joint by about 40 per cent, which is the difference between a design that works and one that does not.
How do I find the throat of an unequal fillet?
It is the two legs multiplied together and divided by the square root of the sum of their squares — the altitude of the right triangle. Crucially it is neither the average of the legs nor the smaller leg: it sits below both. A 6 by 12 millimetre fillet has a throat of 5.37 mm, which is less than the smaller leg itself. Assuming the smaller leg with the usual 0.707 factor would have given 4.24 and understated it instead, so the error goes both ways depending on which shortcut you reach for.
Is one big fillet or two small ones better?
Two small ones, whenever you can reach both sides. Two 6 mm fillets give half again the total throat of a single 8 mm for barely more weld metal, because you are buying throat linearly instead of paying for metal quadratically. They also load the joint symmetrically rather than pulling it into a bend, which a single-sided fillet always does. If more capacity is still needed after that, lengthening the weld is cheaper than deepening it.
Why is there a minimum fillet size if my load is tiny?
Because the minimum is a cooling-rate rule rather than a strength rule. Thick plate draws heat out of a small weld fast enough to crack it, which is the same argument that drives preheat requirements. That is why the governing dimension is the THICKER part being joined and why a light load is no defence — you are not being protected from overload, you are being protected from a weld that quenches itself against the parent metal.
Is there a maximum fillet size too?
Along a plate edge, yes. On material below about 6 mm the limit is the thickness itself, and above that it is the thickness less roughly 1.5 mm. The reason is simply that you cannot melt the edge away and still have an edge to weld to — a fillet built right up to a corner rounds it over and the actual throat becomes impossible to verify. Away from an edge the practical limit is what the process can deposit rather than the geometry.
What happens above the single-pass limit?
The cost gains a step on top of the square law. Most processes will lay about 8 mm in a single horizontal run, and past that every extra pass brings repositioning, interpass cleaning and inspection with it. So the real cost curve is not just steepening, it jumps right where the square law is already biting. Designers who know this stop at 8 mm and go double-sided rather than pushing a single fillet larger.
When does adding more weld stop helping?
When the base metal becomes the limit. Once the weld capacity exceeds the shear rupture strength of the thinner plate, more weld metal buys nothing at all, and neither does a stronger electrode — the plate simply fails first. The tool shows both numbers so you can see which one is governing. Past that point the only real options are a thicker plate or a longer weld.
🔒 This tool runs entirely in your browser. Nothing you enter is uploaded, logged, or stored.