High performance welds - non-welding, non-fatigue explanation

Non-explanation of theory...

Metal fatigue

For this objective, here is sufficient knowledge of metal fatigue [Wikipedia].
Only the overall nature of fatigue and its characteristics are needed here - the Introduction, Stages of fatigue and Characteristics of fatigue
Significant in the following explanation will be these two topics from section "Stages of fatigue"; Crack initiation and Crack growth .

Welding

No overall treatment given. Sufficient details and basis to discuss...

Overall form of a weld : "Parts of a Weld" [Lincoln Electric]
Note the two weld toes in both main types of weld, which are central to the following discussion.
The American Welding Society definition [ > ] of Weld Toe is "The junction of the weld face and the base metal."

This explanation - "the weld SCF = 3" case

SCF == Stress Concentration Factor, and is "the crux of matter".

My approach here is dismissive of previous treatment of fatigue design.

For a "perfect" length of any material in uniaxial tension, Stress=Force/Area, which is uniform across the cross section. That SCF=1. That is unattainable in general welding. Moving on...

Welds should have an SCF of around 3.

This aligns with the SCF of a smooth hole in infinite plate being SCF=3. This is well-known in engineering - original derivation cited as being E.G.Kirsch in 1897. That article providing the familiar illustration of the effect.

Finite Element Analysis modelling of a "good" weld (blended weld toe cross-sectional shape) concurs.

(Model before mesh refinement),

(very approximate stress result before mesh refinement)

with result (see the red-brown "toegroove" line) converging at SCF~=2.7

You cannot easily beat "SCF = 3", as the steel will have its own flaws likely to be around 3 in the best of circumstances.

Take-aways

• SCF=3 means local stress concentration to 3X mean stress - which might sound alarming - but abundant experience is smooth round holes in, eg the structure of aircraft, is not a problem
• problematic sharp flaws have SCF's in 10's.

The outcome - welding steel structure to the "SCF~=3" strategy

SCF~=3 == SCF is approximately equal to 3; it is around 3

The "to this strategy" outcome is arrowed with the "!!"

Note both scales are logarithmic. That's 5.9X current expected endurance to break, unbroken.

That is along at the fatigue performance of the steel plate itself - the "Class B" line.
ie the weld and the steel being joined have about the same fatigue endurance.

Which is what you'd expect, implementing the strategy explained.

Take-aways

• The effect of the "SCF~=3" strategy on fatigue endurance is astonishing; compared to current very consistent familiar performances

Historic performance welds

The historic performance of welded structure in fatigue loading (cyclic loading), made according to current orthodoxies, is much lower. Perfectly illustrated in the "S-N curve" image just shown. The otherwise identical but to current Welding Procedure Specification weld labeled "Class F" broke at exactly the expected fatigue endurance according to current orthodoxies.

[Note : consider only the "S355MIG" and "DH36FCW" "commercial weld" points - the other points are commercially infeasible exotic trial welds]

This is the performance associated with all current Standards on welds (listed without explanation - ISO15614; ISO9606; ISO5817 & ISO 6520) and fatigue design (also without explanation - the EN1993 "Euronorms").

Why?

Historic welds in cyclic-load performance - non-welding, non-fatigue explanation

The elegant explanation continues the "Stress Concentration Factor" approach.

These historic SCF's >> 3 (are much greater than 3). Acting at the weld toes, previously noted as being where attention is to be directed.

Earlier, in previously seen image for a new-style "good" weld

is a curvaceous blended weld toe with an SCF~=2.7

Welds to current orthodoxies have the "dreaded" toe intrusions, with very high SCF - example

The horizontal toe-intrusion is about 0.25mm length. These tiny crack-like features at the weld toes, with a very high Stress Concentration Factor, do most of the harm.
An early-stage fatigue crack is the vertical feature from its inner tip...
That SCF cannot be analysed by Finite Element Analysis modelling (?), as the model is "divergent" on mesh refining chasing an answer. Already to be seen in FEA results - the green "toeintrusion" line.
Analytical treatments are subjects like Fracture Mechanics.

Take-aways

• Toe-intrusions are a "pessimum" (opposite of optimum) - it infeasible to visualise a consistent strategy with worse outcome
• Plate steels have good fatigue resistance; weld metals have good fatigue resistance - it is only the engineering stress-concentration effect of these geometric features at the weld toes which cause the very low fatigue performance of current welded steel structure

Developing explanations and a prognosis

Most explanation can be constructed from the "Stages of fatigue"; Crack initiation and Crack growth .

The duration of the crack initiation stage can be much longer than the duration of the crack growth stage. So for there to be an extensive fatigue crack initiation stage is highly advantageous to structural performance in cyclic loading.

Furthermore, crucial to the following arguments:

• the fatigue crack growth stage duration is independent of yield stress of the steel (akin to how Young's Modulus is a nearly constant property of steels)
• the fatigue crack initiation stage duration can have a favourable relation to the steel's yield strength (the higher the yield stress, the higher the endurance)

With toe intrusions case

The very high SCF from toe-intrusions results in immediate crack growth, with no crack initiation stage.

This is axiomatic in current weld fatigue orthodoxy, including Standards.

Consequences:

• fatigue performance is independent of yield stress of the steel
• fatigue endurance to fracture is very consistent and predictable
• fatigue performance is poor

Blended weld toes case

The only conceivable (?) explanation for the "to this strategy" / "SCF~=3" outcome ("!!"), recently presented as the S-N curve image, is that a fatigue crack initiation stage has been created.

Evidence supporting the suggestion : that sample has done 5.9X the conventional entire endurance to fatigue fracture with no detectable fatigue crack yet
(detecting >=0.5mm crack; true detection ability not better known).

If a fatigue crack initiation stage has been introduced, the consequences:

• fatigue performance is good
• potentially : fatigue performance could increase with increasing yield stress of the steel (as is seen for forged-and-machined components)
• that fatigue endurance to fracture will tend to show scatter in outcomes

The caution around "potential" is that favourable tendencies could be defeated by other factors.

Were fatigue endurance to become a favourable function of yield stress, the commercially valuable consequence would be advantage in using high-strength steels in cyclically loaded structures like bridges.

(R. Smith, 04Jul2021, 05Jul2021 (ref Intro, S1, S2))