FAT 80, FAT 100, FAT 125: What Offshore Engineers Get Wrong About Fatigue Design

A structural detail is subjected to a stress range of 100 MPa at N = 10⁶ cycles. Is this acceptable?

The answer most engineers give — if they have some familiarity with fatigue design — is to check whether 100 MPa is below the allowable stress for the applicable FAT class. If this is a fillet weld toe on a plate edge, the allowable stress range at 10⁶ cycles is approximately 107 MPa for FAT 80. The detail passes.

But the question is poorly framed — and the fact that it is poorly framed is itself the signal. In offshore fatigue design, you almost never assess a detail against a single stress range at a single cycle count. You assess it against a loading spectrum — a distribution of stress ranges, each occurring a different number of times, over the design life. The question is not “does this detail survive 10⁶ cycles at 100 MPa?” The question is “does this detail survive the cumulative damage from 10⁶ cycles at 80 MPa, plus 5×10⁵ cycles at 120 MPa, plus 2×10⁵ cycles at 150 MPa, plus all the other stress events in the spectrum, without the cumulative damage exceeding 1.0?”

Getting to that question — and knowing how to answer it — requires understanding how FAT classes are defined, how S-N curves are constructed, what the slope change at high cycle counts means, and how Miner’s rule works in practice. This post covers the things that are most commonly misapplied.

What FAT Actually Means

FAT is not a material property. It is not a steel grade. It is a detail classification — a number that indexes an S-N curve to a specific welded joint geometry.

The FAT number is defined as the allowable stress range in megapascals at N = 2×10⁶ cycles. This is important: the reference cycle count is 2×10⁶, not 10⁶. Confusion between these two cycle counts is one of the most common errors in fatigue design. The fatigue limit sometimes cited at 10⁶ cycles in undergraduate texts is a different concept — it refers to the approximate knee of the S-N curve for unwelded parent material. The FAT class is defined at 2×10⁶ because that is the reference cycle count used in DNV-RP-C203, and that is the standard that governs offshore fatigue design in the North Sea and most international offshore jurisdictions.

FAT 80 means: at N = 2×10⁶ cycles, the allowable stress range is 80 MPa. FAT 100 means: at the same cycle count, the allowable stress range is 100 MPa. The higher the FAT number, the more fatigue-resistant the detail category.

The FAT class hierarchy in DNV-RP-C203 runs from FAT 71 (most severe — fillet weld toe on structural tee, unsupported) to FAT 125 (least severe — unwelded parent material). The progression is not arbitrary. It reflects the severity of the stress concentration at the crack-prone location in each detail type.

Why FAT Class Is Determined by Geometry, Not by Steel Grade

A question that arises regularly: does higher-strength steel give a better FAT class?

No. S355 and S690 steel, welded into the same joint geometry, receive the same FAT class. FAT class is a function of stress concentration geometry — the local stiffness and geometry of the welded connection — not the parent material yield strength. A FAT 80 joint in S355 and a FAT 80 joint in S690 have the same fatigue performance in terms of allowable stress range at a given cycle count. The higher-strength steel does not compensate for a poor joint geometry. The geometry is what matters.

This is why offshore fabrication specifications require specific weld toe profiles, minimum leg sizes, and in some cases weld toe grinding and MPI inspection — not because the weld needs to be stronger in a yield sense, but because the weld toe geometry determines the FAT class, which determines the allowable stress range, which determines the fatigue life.

The S-N Curve: How the FAT Class Defines the Curve

The FAT class defines the intercept of the S-N curve on the stress axis. The slope is m = 3 for N < approximately 5×10⁶ cycles (the high-cycle regime). The governing equation is:

N = C / Δσ^m

where C is a constant derived from the FAT class. The relationship between FAT class and C is:

Using the correct C value matters. If you use the wrong C — for example, applying C = 10¹²·⁰⁰⁰ (FAT 71’s value) to a FAT 100 detail — you will under-predict the allowable stress range for FAT 80, FAT 100, and FAT 125 details and produce an unconservative fatigue assessment.

For a FAT 100 detail at N = 10⁶ cycles: Δσ_all = (C/10⁶)^(1/3) = (10¹²·⁴⁸¹/10⁶)^(1/3) = 10²·¹⁶⁰ ≈ 144 MPa. If you had mistakenly used C = 10¹²·⁰⁰⁰, you would get Δσ_all ≈ 126 MPa — a 13% error in the allowable stress range, in the conservative direction for this case. At higher cycle counts, the error grows.

The Slope Change at 5×10⁶ Cycles

Below approximately 5×10⁶ cycles, the S-N curve for most offshore detail categories follows m = 3. Above 5×10⁶ cycles, the slope steepens to approximately m = 5. This is not a minor technical footnote — it is the difference between a fatigue assessment that passes and one that significantly underestimates fatigue damage.

The reason for the slope change is metallurgical: at very high cycle counts, the fatigue mechanism shifts from the formation and propagation of a dominant crack to the simultaneous initiation of multiple micro-cracks. The effective stress intensity at the crack tip reduces, and the growth rate per cycle drops. The m = 5 slope above 5×10⁶ reflects this different mechanism.

In cumulative damage calculations using Miner’s rule, this slope change means that stress ranges in the very high-cycle regime contribute less damage per cycle than the m = 3 extrapolation would predict. But it also means that if your loading spectrum includes a large number of cycles in the 10⁶ to 10⁷ range — as many offshore structures do — the fatigue assessment must use the correct slope for each portion of the spectrum.