NavArch Handbook
Chapter 2 / Section 2.11
Repeated stress cycles well below yield initiate and grow cracks at stress concentrations — weld toes, cut-outs, bracket toes, pipe penetrations. Ship structures see – wave-induced cycles in 25 years plus machinery vibration cycles; cranes and winches see load cycles per lift. Fatigue, not static strength, governs most structural repairs in service.
The governing quantity is the stress range of each cycle (not the maximum stress). A variable-amplitude history is reduced to a histogram of ranges by rainflow counting; for ships, the long-term distribution of wave-induced stress ranges is often represented by a two-parameter Weibull distribution (DNV-CG-0129).
Fatigue life (cycles to failure) at constant stress range :
and depend on the detail category (geometry of the weld detail), the environment (air, seawater with cathodic protection, free corrosion) and the stress approach (nominal or hot-spot). DNV-RP-C203 gives bilinear curves: slope up to cycles, then (in air); in seawater with cathodic protection the first branch is steeper-lived (lower ) and in free corrosion there is no change of slope. The curves are mean-minus-two-standard-deviations (97.7 % survival).
From B1 (plain plate, machined edges) through C, D (transverse butt welds, well-made), E, F, F1, F3 (attachments, fillet welds of increasing severity), G, W1–W3 (load-carrying fillet welds, weld throat). The table gives the air-curve constants for the most used categories. The hot-spot approach uses the D curve with the structural hot-spot stress from a fine-mesh FE model ( elements, read-out at and — RP-C203 Sec. 4).
For plates thicker than the reference mm the fatigue strength is reduced: is multiplied by with from the category (0.20 for D, 0.25 for F–W).
Welded joints: residual stresses of yield magnitude mean that the full range is damaging regardless of the mean stress — no correction. Base material and post-weld-treated details: a reduction factor applies when part of the cycle is compressive (RP-C203 Sec. 2.5, factor : full range when fully tensile, 0.6 × range when fully compressive, linear in between).
For blocks of cycles at range with lives :
Failure is predicted at ; the design criterion uses with a design fatigue factor DFF between 1 and 3 (or more) depending on accessibility for inspection and consequence of failure (DNV-OS-C101 for offshore units; ship rules work with a 25-year design life at a specified probability level — DNV Rules Pt.3 Ch.9 and DNV-CG-0129).
| Symbol | Description | Unit |
|---|---|---|
| Number of cycles to failure | - | |
| Stress range | MPa | |
| S–N curve intercept (category and environment dependent) | MPa^m | |
| Negative inverse slope (3 or 5 for welded steel in air) | - |
Result unit cycles
| Symbol | Description | Unit |
|---|---|---|
| Stress range at the knee (10⁷ cycles), e.g. 52.63 MPa for D | MPa |
Result unit cycles
| Symbol | Description | Unit |
|---|---|---|
| Thickness exponent (0 for B1/B2; 0.20 for D, E; 0.25 for F–W) | - |
Result unit MPa
| Symbol | Description | Unit |
|---|---|---|
| Number of cycles in block i | - | |
| Cycles to failure at range Δσ_i | - | |
| Design fatigue factor (1–3 typical; up to 10 for non-inspectable critical offshore details) | - |
Result unit -
| Symbol | Description | Unit |
|---|---|---|
| Surface stresses read at 0.5t and 1.5t from the weld toe (linear extrapolation) | MPa |
Result unit MPa
DNV-RP-C203 Sec. 4; use the D curve with hot-spot stress.
| Symbol | Description | Unit |
|---|---|---|
| Tensile part of the stress range | MPa | |
| Compressive part of the stress range (absolute value) | MPa |
Result unit MPa
Not for as-welded joints (residual stresses). Confirm the current form in RP-C203 Sec. 2.5.
| Category | (N ≤ 10⁷) | (N > 10⁷) | Range at 10⁷ cycles [MPa] | Thickness exponent | Typical detail | ||
|---|---|---|---|---|---|---|---|
| B1 | 4 | 15.117 | 5 | 17.146 | 106.97 | 0 | Plain plate, machined edges, no welds |
| C | 3 | 12.592 | 5 | 16.320 | 73.10 | 0.05 | Transverse butt weld ground flush, NDT |
| C1 | 3 | 12.449 | 5 | 16.081 | 65.50 | 0.10 | Transverse butt weld, as welded, toe angle ≤ 30° |
| D | 3 | 12.164 | 5 | 15.606 | 52.63 | 0.20 | Transverse butt weld as welded; **hot-spot curve** |
| E | 3 | 12.010 | 5 | 15.350 | 46.78 | 0.20 | Butt weld on backing; fillet-welded attachment ≤ 150 mm along stress |
| F | 3 | 11.855 | 5 | 15.091 | 41.52 | 0.25 | Attachment > 150 mm; cruciform full-penetration |
| F1 | 3 | 11.699 | 5 | 14.832 | 36.84 | 0.25 | Fillet-welded cruciform / bracket toe with gap |
| F3 | 3 | 11.546 | 5 | 14.576 | 32.75 | 0.25 | Partial penetration cruciform; cover plate ends |
| G | 3 | 11.398 | 5 | 14.330 | 29.24 | 0.25 | Attachment on edge; load-carrying fillet toe |
| W3 | 3 | 10.970 | 5 | 13.617 | 21.05 | 0.25 | Fillet weld throat (shear on weld) |
| Situation | DFF | Source |
|---|---|---|
| Accessible for inspection and repair, in dry or above splash zone, failure not critical | 1.0 | DNV-OS-C101 (offshore) |
| Accessible, below splash zone / wet | 2.0 | DNV-OS-C101 |
| Not accessible for inspection, or failure critical | 3.0 | DNV-OS-C101 |
| Non-inspectable, substantial consequence (some offshore cases) | up to 10 | DNV-OS-C101 / project specification |
| Ship hull details (class rules) | Design life 25 years at the rule probability level; equivalent to DFF ≈ 1 with conservative load spectrum | DNV Rules Pt.3 Ch.9; DNV-CG-0129 |
Worked example
Problem
A deck crane pedestal bracket toe (as-welded, fillet, category F1, plate 20 mm) sees during 25 years: block 1 — 2.0 × 10⁶ cycles at a hot-spot/nominal stress range of 60 MPa (lifting operations); block 2 — 1.0 × 10⁷ cycles at 40 MPa (wave-induced hull girder plus deck vibration). Compute the fatigue damage with the F1 air curve and compare with DFF = 2 (accessible, splash-zone deck).
Solution
Plate 20 mm < 25 mm → no thickness penalty.
Block 1: MPa > 36.84 (knee) → : . .
Block 2: MPa > 36.84 → still : . .
Total — fails by a factor of more than 4.
With the toe ground and a full-penetration weld the detail might be classed E (, ): , ; → above the knee? No: at 40 MPa the E-curve knee is 46.78 MPa, so 40 MPa lies on the branch: , ; total — still above 0.5. The bracket must be redesigned (soft toe, larger radius, thicker insert to reduce the stress range by ~25 %: damage scales with , so a 25 % reduction gives ✓). This is the typical outcome: fatigue at pedestal and winch foundations is governed by detail design, not by plate thickness.
Free download (registration) from dnv.com. Confirm edition.
Referenced by DNV Rules Pt.3 Ch.9.