Free tool · no sign-up
Fillet weld size calculator
Convert a fillet leg into its design throat, or a required throat into the leg that delivers it, and see immediately what that size costs: cross-section, weld metal volume and mass per metre, and the consumable for the whole run. Unequal legs and a qualified deep-penetration allowance are both handled.
No minimum size is built in. The smallest fillet a joint may carry follows from the thicker part being joined and is set by the code the job is built to, so you type that figure and the tool reports the margin. Everything runs in your browser.
One triangle, two sizes and a mass
Design throat, equal legs a = z · cos 45° = z / √2 ≈ 0.707 · z
Leg from a required throat z = a · √2 ≈ 1.414 · a
Design throat, unequal legs a = z1 · z2 / √(z1² + z2²)
With counted penetration adesign = ageometric + s
Cross-section A = (z1 · z2 / 2) · (1 + k), k = convexity allowance
Mass per metre m = A [mm²] · ρ [g/cm³] / 1000 [kg/m]
Lengths in mm, density in g/cm³. The volume in cm³ per metre is numerically the same figure as the cross-section in mm².
All of it comes from one triangle. The throat is the altitude of the largest triangle that fits between the fusion faces, so it depends on both legs; the deposited metal is the area of that triangle plus whatever convexity the face carries. The throat is what carries load, the area is what costs money, and the two do not scale together — doubling the throat quadruples the weld metal.
How to calculate a fillet weld size
From the size on the drawing to the throat, the weld metal per metre and the wire for the job.
- Decide which size the drawing states. Under EN ISO 2553 a figure after z is a leg and a figure after a is a throat. Pick the matching option so the tool converts in the right direction: mixing them up either sends the corrected leg 41 % over with double the weld metal, or leaves the throat 29 % short of what the designer asked for.
- Enter the leg or the throat. For an equal-leg fillet the second leg stays empty. Fill it only when the joint deliberately uses unequal legs, in which case the throat is no longer 0.707 times either one.
- Add a penetration allowance only if you have earned it. Leave it at zero unless the governing code permits the credit and your procedure qualification demonstrated that penetration on a macro-section with these parameters.
- Set the convexity, density and run length. Convexity is the weld metal above the theoretical triangle; density belongs to the alloy, not to steel by habit; the length is every fillet of this size on the assembly.
- Check the size against your code and cost the run. Type the minimum your code imposes for the thicker part and read the margin, then take the consumable figure to the purchase order and the mass per metre to the deposition rate calculator for the arc time.
- Minimum sizes belong to the code, not to a calculator. They are cooling-rate floors, not fusion floors: they rise with the thickness of the thicker part joined because a thick section chills a small bead fast enough to harden the heat-affected zone and risk restraint cracking, and they differ between structural, pressure and pipeline rules. Read the figure in the code that governs the contract and type it into the comparison field; this page holds no copy of anyone's table, deliberately.
- Penetration is a credit you have to have earned. Where a code permits it, the allowance is what the procedure qualification demonstrated on a macro-section with those exact parameters, position and joint. Entering a number here does not create the evidence, and an audit will ask for the macro.
- Over-welding is the quiet cost. Section area grows with the square of the leg while the throat grows linearly, so a leg drawn one size too large buys a little strength and a lot of consumable, arc time and distortion — take the mass per metre to the deposition rate calculator to see the hours it adds.
- A bigger fillet is also a thermal decision. Laying the whole section in one pass raises the heat input and slows the cooling; splitting it into two changes both. The heat input calculator and the t8/5 cooling time calculator are where that side of the choice gets a number.
- Density is an alloy property, not a habit. The 7.85 g/cm³ default is carbon and low-alloy steel. Leaving it there on an aluminium assembly overstates the weld metal by nearly three times, which is enough to wreck a quotation on its own.
- The size on the drawing is not the size in the joint. What the arc actually deposited on the four-thousandth fillet is a measurement, not a calculation — see HeatCore AI for how the passes get watched, and the bead geometry measurement guide for how the finished profile gets checked.
For the leg-versus-throat chart, the geometry behind it and how codes set a minimum from the thicker part, read the fillet weld size chart guide.
Questions engineers ask about this calculation
Why is a fillet weld throat 0.707 times the leg length?
Because the design throat is the height of the largest triangle that fits between the fusion faces, measured from the root, and on an equal-leg fillet in a right-angled joint that height is the leg multiplied by the cosine of 45 degrees — a = z · cos 45° = z / √2 ≈ 0.707 z. Read the other way, a required throat needs a leg of a × √2 ≈ 1.414 a. On a deliberately unequal fillet the shortcut no longer applies and the throat is z1 · z2 / √(z1² + z2²). The drawing convention matters as much as the arithmetic: under the EN ISO 2553 symbol rules a size written after z is a leg and a size written after a is a throat, and confusing the two is not a rounding error — reading a leg value as if it were a throat sends the corrected leg 41 % over and doubles the weld metal, and reading a throat value as if it were a leg leaves the achieved throat 29 % short of what was specified.
Can deep penetration be counted towards the fillet weld throat?
Sometimes, and never by default. Structural codes on both sides of the Atlantic — AWS D1.1 in its clauses on effective weld size, and EN 1993-1-8 in its treatment of fillet weld throat with deep penetration — allow penetration beyond the root to be added to the throat only where the welding procedure qualification demonstrates that the deeper penetration is achieved consistently with those exact parameters, position and joint. The amount is whatever your qualification test showed on a macro-section, not a figure a calculator can supply, which is why this page leaves the allowance at zero and asks you to type it. Counting penetration that the procedure has not proved is the same as designing the joint smaller than the drawing says.
Why does one extra millimetre on a fillet leg cost so much weld metal?
Because the cross-section grows with the square of the leg while the strength grows only with the throat, which is linear in the leg. A 6 mm equal-leg fillet has a theoretical section of 18 mm², an 8 mm one has 32 mm² — 78 % more metal, 78 % more arc time and 78 % more heat into the joint for 33 % more throat. Over-welding is therefore expensive twice: once at the consumable store and once in distortion, and it is the single most common reason a fabricated assembly consumes more wire than the estimate said it would.
What decides the minimum fillet weld size on a joint?
The thicker of the two parts being joined, and the code the job is built to. Minimum sizes exist because of cooling rate, not fusion: a thick part conducts heat away from the arc so fast that a small bead deposited on it cools too quickly, hardening the heat-affected zone, cutting its ductility and risking restraint cracking — so the code sets a floor that rises with material thickness. AWS D1.1 states it as a minimum leg tied to the thickness of the thicker part joined. EN 1993-1-8 sets no equivalent thickness-tied minimum fillet size; the European route controls the same cooling-rate risk through preheat and the welding procedure qualified under EN 1090-2 and EN 1011-2. Read the actual figure in the code that governs your contract and type it into the comparison field above: this page reports the margin and deliberately holds no copy of any acceptance table.
The drawing states a size. Only a measurement says what the cell deposited.
A fillet size is a decision made once and repeated thousands of times by an arc that nobody watches. HeatCore AI monitors the thermal signature of every pass in a robotic cell and flags process drift as it happens, so a change that would show up on the finished bead is visible on the pass that produced it.
