Fillet Weld Strength Calculator

Why a fillet weld is checked in shear across its throat

A fillet weld is a triangle of deposited metal filling the corner between two parts. Load it any way you like — pull the parts apart, push them along each other, peel them — and testing shows the weld consistently fails by shearing across the narrowest section through the triangle. That section is the effective throat, and for the equal-leg fillet used in almost all fabrication it is the leg size multiplied by the sine of 45°, or 0.707 × w.

Design therefore ignores the leg you can see and works with the throat you cannot. A 1/4 in fillet is checked on a throat of 0.1768 in. This is why doubling the leg size doubles the capacity exactly — throat scales linearly with leg — and why a weld larger than the drawing calls for buys strength at a cost that goes up with the square of the leg, since the cross-sectional area of deposited metal is w²/2. A 3/8 in fillet costs 2.25 times as much filler as a 1/4 in fillet to gain 50 percent more strength.

The nominal shear strength of weld metal is taken as 0.60 times its tensile classification. An E70XX electrode deposits metal with 70 ksi tensile strength, so its nominal shear strength is 42 ksi acting on the throat area. Multiply by throat and by length and you have the joint capacity, before any safety factor.

The formula, the safety factors, and the base metal check

The nominal capacity of a fillet weld group is Rn = 0.60 × FEXX × 0.707w × L, where L is the total effective length. Two reduction conventions convert it to something you can design with. Under LRFD you multiply by φ = 0.75 and compare against factored loads. Under ASD you divide by Ω = 2.00 and compare against service loads. Do not mix them: an ASD comparison against factored loads is roughly a 50 percent error in the unsafe direction.

This produces the number every steel detailer carries in their head. A 1/16 in fillet in E70 metal gives 0.60 × 70 × 0.0442 = 1.856 kips/in nominal, which is 1.392 kips/in factored and 0.928 kips/in allowable. Because throat is linear in leg, you multiply by the number of sixteenths: a 5/16 fillet is 5 × 1.392 = 6.96 kips/in under LRFD. Checking a calculator against 1.392 per sixteenth is the fastest way to confirm you are using the right convention.

The weld is only half the check. Weld metal deposited on a thin or soft plate can be stronger than the plate it is attached to, and the joint then fails by tearing the base metal along the weld line. AISC handles this as base metal shear rupture: Rn = 0.60 × Fu × t × L, with the same φ and Ω. The governing capacity is the smaller of the two. When base metal governs, adding weld size does nothing at all — a point this calculator flags explicitly, because it is the most common way a designer wastes filler.

Two geometry rules constrain the size you may specify. The maximum fillet along the edge of a part 1/4 in thick or thicker is the thickness less 1/16 in, so that the edge remains visible for inspection; below 1/4 in it is the full thickness. Minimum sizes are tabulated by the thickness of the thinner part joined, and exist to control cooling rate rather than strength — a small weld on heavy plate quenches into hard, crack-prone metal.

Reading the numbers, and where they stop being conservative

Utilisation below about 90 percent is a healthy detail; at 100 percent you have no room for the weld being placed slightly small or slightly short. Effective length is not the length of the drawn weld — craters at each end, gaps and areas of poor fusion do not count, and end returns are treated conservatively.

Three limits move the answer away from the simple formula and are worth knowing.

Short end-loaded welds. Where the effective length of an end-loaded fillet is less than four times the leg size, AISC requires the effective size to be taken as one quarter of the length instead, because such a short weld cannot develop uniform stress. This calculator warns when you cross that threshold rather than silently reporting an unconservative figure.

Long end-loaded welds. Beyond 100 times the leg size, stress is far from uniform along the weld and a reduction factor applies; beyond 300 times, the factor is fixed at 0.60. Long lap-splice welds are the usual case.

Load direction. AISC permits an increase for loads not parallel to the weld axis, the familiar 1.0 + 0.50 sin^1.5θ factor, which raises capacity by 50 percent for a weld loaded perpendicular to its axis. This calculator does not apply it, so the figures here are the longitudinal-loading values and are conservative for transverse loading. That is deliberate: the increase requires that you know the direction of load on every weld element, and in weld groups loaded eccentrically the direction varies along the weld.

None of this covers fatigue. Cyclically loaded welds are governed by stress range and detail category, not by static capacity, and a weld with 40 percent static utilisation can still fail in fatigue. If the load cycles, look up the detail category rather than relying on this page.

Fillet weld capacity per inch, E70XX electrode

Multiply by the total effective weld length in inches. For other electrodes, scale by F_EXX / 70 — an E80 weld is 80/70 = 1.143 times these values.

Mistakes that put a welded connection at risk

• Comparing ASD capacity against factored loads. The two conventions differ by a factor of 1.5, and the error is in the unsafe direction. Pick one method for the whole connection.
• Forgetting the base metal check. A 1/2 in fillet on 1/4 in plate cannot deliver its capacity, because the plate tears first. When base metal governs, more weld buys nothing.
• Using drawn length rather than effective length. Craters, stop-starts and unfused areas do not carry load. Effective length is what a weld inspector can certify.
• Specifying a fillet larger than the edge allows. On material 1/4 in and thicker the maximum edge fillet is thickness less 1/16 in, so the plate edge remains visible for inspection.
• Ignoring minimum size rules. The minimum fillet on heavy plate exists to slow cooling, not to provide strength. Undersizing it risks HAZ cracking regardless of the load.
• Applying static capacity to a cyclic load. Fatigue is governed by stress range and detail category. A statically comfortable weld can still fail after a few million cycles.
• Assuming an intermittent weld equals its total length. Intermittent fillets carry the sum of their segment lengths, but the discontinuities become stress raisers and are barred from many fatigue-sensitive and corrosive applications.

Fillet welds among the alternatives

Fillet welds dominate fabrication because they need no joint preparation: bring two parts together and weld the corner. Complete-joint-penetration groove welds, by contrast, are checked against the base metal rather than the weld, because a properly executed CJP weld in matching filler develops the full strength of the connected material. That makes CJP welds stronger, but they need edge preparation, backing or back-gouging, and usually ultrasonic or radiographic inspection, so they cost several times as much per inch. Choose a fillet unless the joint genuinely needs to develop the section.

Once the size is fixed, the shop questions follow. Metallurgy in the heat-affected zone is governed by energy per unit length, which the welding heat input calculator quantifies; on hardenable steels the same weld that passes this strength check can crack in the HAZ if it is run too fast and cold. If you are working with unlabelled material and no mill certificate, a hardness reading converted through the hardness conversion calculator gives an approximate tensile strength, which is the F_u this page needs for the base metal check.

Filler consumption follows from the geometry directly. The deposited area of an equal-leg fillet is w²/2 plus reinforcement, so a 5/16 in fillet deposits about 0.049 in² per inch of length. Multiply by length and by the density of steel to estimate electrode weight, the same arithmetic the metal weight calculator performs for stock shapes. That figure drives both cost and, through heat input, distortion.

For members rather than connections, the governing checks move to the section itself — axial stress and elongation from the stress and strain calculator, and bending from a section modulus check. A connection designed to 70 percent utilisation on a member already at 95 percent has not helped anybody.