Fillet-weld test evaluation


Abbreviations used

January2021 - the new data and interpretations presented here come from the series of four mini-projects, three in November 2020 and one in January 2021, abbreviated as follows:

Butt-welds - not part of this topic:

For the fillet welds tests:

See Structures "index" for context to the tests and this article.

Other abbreviations used:


This page collates

from in serving investigation and evaluation of fillet-weld [definition] performance.

At time of creating this page, late January2021, own content may include


At time of writing, January2021, the major interests are

Previous investigations and findings

Fillet-weld fatigue-resistance investigation 2011

From 2011 - fatigue-resistant fillet welds test programme -"index" page to the methods, tests and interpretations of a programme investigating the possibilities for fatigue-resistant fillet welds for warships and bridges.
Also as report [PDF, 3.2MB]. Interesting result - FCAW T-fillet 1.4Million cycles unbroken no cracks detectable, for 250thousand cycles expected to break. Which could now be progressed if the "bcfwtt" could be made cyclic loading for fatigue testing. Considered in Proposal : research path to high-strength steel fatigue-resistant welded structures

The beam-configuration fillet-weld tensile test findings

These are mp3 and mp4 of the Nov2020-to-Jan2021 test programme.

Necessary data from the beam-configuration fillet-weld tensile test to enable calculating the force on the test-weld are

Proceeding onwards : to then obtain the stress(es) in the test weld, necessary data are Proceeding further onwards : to evaluate the calculated values and arrive at interpretations and conclusions, the following is among the information needed

In the "bcfwtt", to obtain the force in the weld at breaking F_w (which we want) from the force applied by the hydraulic piston F_p (which we know) uses the formula
which is probably what one would intuitively guess.
Whose derivation is presented in mp3 see section Stresses in weld analysed .

Data and its processing

"Fillet welds tensile tested in beam test" [mp3]

The measurements and analysis are presented in the web-page for mp3

For the [ISO14341-A] "G3Si1" weld-metal, the calculated breaking strength is 567MPa - for expected typical value for "G3Si1" weld-metal of 560MPa.

"Tensile-test rig for beam-configuration fillet-weld samples" [mp4]

The hydraulic pump-and-cylinder used in mp4 had no hydraulic-fluid pressure-gauge, so an accurate citable "F=P.A" estimate of cylinder force applied to the test sample is not available.

An estimate of force applied was made by a very inexact method described. But where the resulting deduced weld breaking stress could not be known at the time of the test. So there could be no human bias in the applied force estimate.

The resulting fillet-weld breaking strength estimate of 572MPa, for the expected 560MPa mean value for "G3Si1" GMAW/MIG weld metal, is surprising for its apparent accuracy.

Evaluation and interpretation of tests

Writing Thursday 28January2021:

Added on Sunday 07February2021 - FEA of the "bcfwtt":

More detailed interpretation of fillet-weld tensile break measurements

The findings can now be considered in relation to established knowledge and theory.

Concepts and terminology:

- as I visualise them, from mp3

- as visualised by Hicks (ref.) on pg85

where Sigma=linear-stress and Tau=shear-stress.

Hicks (ref.) on pg85 presents the known established formula which should connect
for a *double-sided* fillet weld

In this equation he presents as [6.3],
with symbols here given text name
P_perp=2.t.L/sqrt2 * (sigma_perp+tau_perp)
and introducing t=fillet throat thickness ("a" in common welding terminology) and L=length of the weld

Taking [6.3] and [6.4]
P_perp=2.t.L/sqrt2 * (sigma_perp+tau_perp)
sigma_perp = tau_perp {for this geometry}
we recognise for a 45-45-90 triangular cross section fillet that "t", the weld throat, which is "a" in conventional weld terminology - that sqrt2.a=z - the fillet leg-length

This is for a *double-sided* fillet weld.
Mine has two welds, but in series, not in parallel.
Each weld must take the full stress - so it is in stress terms a single-sided fillet weld ("ssfw").
So in my case of applying "Hicks"
which is the same as what I found empirically and in forming mathematics which matches the physical situation. The observation that fracture is along the longitudinal leg (length "z") of the fillet weld.

"My" derivation for fillet weld strength and the established arithmetic expression for fillet weld strength give the same answer.

So, to that extent, my "finding" is "expected".

Whereas the mode of breakage differs, with my derivation expecting fracture along the longitudinal fillet leg, and the established expression implying fracture is expected across the weld throat.

Hicks qualitatively comments, on pg82, that fillet weld overload fracture has the form which I found - close to the longitudinal fillet leg. So he's derived and presented "the established" analysis in that knowledge.
So it "doubly" affirms that the derivation applies.

I invite comment on these following things which I find surprising:

"Watch this space.." :-)


John Hicks, Welded joint design [3rd-edn.], Abington Publishing, 1999

(R. Smith, 28Jan2021, 29Jan2021 (mp4 str.calc. retn), 31Jan2021 (break tens), 03Feb2021 (560MPa 355MPa), 07Feb2021 (aname, FEA, FEA img preview), 08Feb2021 (aname fea bcfwtt))