4.5 Fatigue Life, Weld-Toe Stress Concentration & Paris-Erdogan Crack Growth

4.3 Mechanical Failure Modes: Fatigue, Brittle Fracture & Creep

Quick Answer: Welded structures fail primarily through three mechanisms: fatigue, brittle fracture, and creep. Fatigue progresses through initiation, stable propagation, and final overload. Because weld toes contain microscopic flaws and high tensile residual stresses, welded fatigue life is propagation-dominated, governed by the Paris Law $da/dN = C(\Delta K)^m$ and cyclic stress range $\Delta\sigma$ rather than base steel static strength. Brittle fracture occurs when stress intensity exceeds fracture toughness ($K_I \ge K_{Ic}$), triggered when service temperatures fall below the Charpy ductile-to-brittle transition temperature (DBTT). Creep occurs at elevated temperatures ($T > 0.4 T_m$), where steady-state strain rate follows Norton's law $\dot{\varepsilon}{ss} = A \sigma^n \exp(-Q/RT)$, and rupture life is extrapolated using the Larson-Miller parameter $LMP = T(C + \log{10} t_r)$.

---

1. Fatigue Lifecycle & S-N Behavior in Welded Structures

Fatigue is the progressive, localized structural damage that occurs when a member is subjected to cyclic, fluctuating loads at stresses well below nominal yield strength. Statistically, fatigue accounts for over 80% of all service failures in welded engineering structures.

   Cyclic Stress Parameter Definitions                 The Three Stages of Fatigue Failure

      Stress (σ)                                       Stage I: Microcrack Initiation
          ^                                            - Slip-band extrusion/intrusion
   σ_max -+--    _--_        _--_                      - In welds: bypassed by toe flaws
          |     /    \      /    \                     
     Δσ --+    |      |    |      |                    Stage II: Subcritical Propagation
          |    |  σ_m |    |  σ_m |                    - Mode I opening driven by ΔK
   σ_min -+----+-\----+----+-\----+---> Time (t)       - Paris Law: da/dN = C(ΔK)^m
          |       \__/        \__/                     - Micro-striations, macro-beach marks
          |       
          |    Stress Range: Δσ = σ_max - σ_min        Stage III: Unstable Catastrophic Fracture
          |    Stress Ratio: R = σ_min / σ_max         - K_I ≥ K_Ic or plastic collapse

The Three Stages of Fatigue

1. Stage I (Initiation): Cyclic shear stresses cause localized dislocation slip along crystallographic slip planes, forming microscopic slip-band extrusions and intrusions on the material surface. In un-notched, polished laboratory specimens, Stage I consumes $80\text{ to }90%$ of total fatigue life.
2. Stage II (Propagation): The microcrack reorients perpendicular to the maximum tensile stress axis, propagating stably cycle-by-cycle as a Mode I planar crack. Each stress cycle produces an incremental blunting and resharpening of the crack tip, leaving microscopic fatigue striations visible under scanning electron microscopy (SEM) and macroscopic beach marks (arrest lines) visible to the unaided eye during load variations.
3. Stage III (Final Fracture): As the crack extends, the uncracked cross-sectional ligament diminishes. Catastrophic fracture occurs when the crack-tip stress intensity reaches fracture toughness ($K_I \ge K_{Ic}$) or net-section tensile overload occurs.

The S-N Curve & Endurance Limit Paradox in Welded Joints

For smooth, unnotched wrought specimens of carbon and low-alloy ferritic steels, S-N curves (Wöhler curves) display a distinct horizontal plateau known as the fatigue limit (or endurance limit, $S_e$), typically occurring around $10^6\text{ to }10^7\text{ cycles}$ at a stress amplitude $S_e \approx 0.40\text{--}0.50 \sigma_{uts}$. At stress amplitudes below $S_e$, the material theoretically exhibits infinite life.

    Stress Range (Δσ)
       ^
       |  Unnotched Smooth Steel: S_e ≈ 0.5 σ_uts
       |   \ 
       |    \____...----------------------- Endurance Limit Plateau
       |     
       |  Welded Joint (AWS D1.1 Cat C/D/E):
       |   \ 
       |    \     NO Endurance Limit Plateau in as-welded joints!
       |     \    Governed by: log(N) = log(C) - m · log(Δσ)
       |      \ 
       +-------+-------------------------------------------> Cycles (N)
              10⁴          10⁶          10⁷

The Great Welding Engineering Paradox: In as-welded structures, there is NO true endurance limit, and increasing the tensile strength of the base steel provides ZERO improvement in fatigue life!

Why does this occur?

1. Elimination of Stage I Initiation: Fusion welding naturally produces microscopic non-metallic slag intrusions, undercut, cold laps, and sharp reentrant angles at weld toes, with effective notch radii often smaller than $0.05\text{ mm}$. These discontinuities act as pre-existing sharp cracks ($a_0 \approx 0.1\text{ to }0.25\text{ mm}$). Consequently, Stage I initiation life is zero; the weldment begins life in Stage II propagation.
2. Residual Stress Domination: Differential thermal shrinkage generates tensile residual stresses approaching the yield point of the base metal ($\sigma_{\text{res}} \approx \sigma_y$) at weld toes. When cyclic external loads are applied, the local effective stress fluctuates near the yield point regardless of nominal applied mean stress, forcing the effective stress ratio to $R_{\text{eff}} \approx 1.0$.

Consequently, major structural welding design standards (AWS D1.1 Clause 4 Part C and Table 4.5, IIW Recommendations, Eurocode 3 EN 1993-1-9) classify fatigue resistance purely by joint geometry categories (Category A through F2). The design curves follow Basquin's power-law formulation:

$$\Delta\sigma^m \cdot N = C_f \implies \log(N) = \log(C_f) - m \log(\Delta\sigma)$$

where $m = 3.0$ for structural steel welds, and the fatigue strength depends strictly on cyclic stress range $\Delta\sigma$, not on the yield strength or UTS of the steel.

---

2. Stress Concentrations & Notch Sensitivity at Weld Toes

The geometrical transition between the deposited weld reinforcement and the base plate creates an abrupt change in cross-section, concentrating nominal stresses.

                         Weld Reinforcement Crown
                                  _____
                                _/     \_
       Base Plate             _/         \_            Base Plate
  ---------------------------+             +---------------------------
                             |\ θ       θ /|
                             | \_       _/ |
                             |   \_____/   |
                                  Weld Toe
                                  Notch Radius: ρ
                                  K_t = 1 + α · (h/ρ)^0.5 · (tan θ)

Theoretical vs. Fatigue Notch Factors

• Theoretical Elastic Stress Concentration Factor ($K_t$): The ratio of peak elastic stress at the notch root to nominal stress:

  $$K_t = \frac{\sigma_{\max}}{\sigma_{\text{nom}}}$$

  For a transverse fillet or butt weld toe, $K_t$ is approximated by the Lawrence/Mattos equation:

  $$K_t = 1 + 0.35 \left(\frac{h}{\rho}\right)^{0.5} \tan(\theta)$$

  where $h$ is reinforcement height, $\rho$ is weld toe radius, and $\theta$ is weld flank angle. Steep flank angles ($\theta > 45^\circ$) and sharp toe radii ($\rho < 0.5\text{ mm}$) drive $K_t$ above $3.0\text{ to }5.0$.

• Fatigue Notch Factor ($K_f$): The actual ratio of unnotched to notched fatigue strength at a given life:

  $$K_f = \frac{S_{e,\text{unnotched}}}{S_{e,\text{notched}}}$$

• Peterson's Notch Sensitivity Index ($q$): $q = \frac{K_f - 1}{K_t - 1} \implies K_f = 1 + q(K_t - 1)$ where $q = \frac{1}{1 + a_P / \rho}$, and $a_P$ is Peterson's material parameter related to grain size and tensile strength. In high-strength steels, $a_P$ is very small, driving $q \to 1.0$ and $K_f \to K_t$ (extremely notch sensitive).

Post-Weld Fatigue Improvement Methods

Because fatigue crack growth initiates at weld toes, modifying the toe geometry or residual stress field provides massive fatigue life extensions:

1. Burr Grinding / Profiling: Rotary carbide burrs remove toe micro-flaws and grind a smooth, gradual radius ($\rho \ge 3\text{ mm}$, flank angle $\theta \le 30^\circ$), lowering $K_t$ and increasing fatigue strength by $30\text{--}50%$.
2. TIG / Plasma Dressing: A non-consumable TIG arc remelts the toe without filler metal, washing the abrupt bead contour into the plate and eliminating micro-inclusions, boosting fatigue strength by $40\text{--}60%$.
3. High-Frequency Mechanical Impact (HFMI) / Ultrasonic Impact Treatment (UIT): High-frequency peening needles plastically deform the toe metal. This imparts three simultaneous benefits: (a) eliminates sharp reentrant notches, (b) severely work-hardens the surface, and (c) introduces deep compressive residual stresses (up to $-400\text{ MPa}$), increasing fatigue limit by $80\text{--}150%$ and upgrading the joint by several AWS detail categories.

---

3. Subcritical Crack Propagation & The Paris-Erdogan Law

Linear Elastic Fracture Mechanics (LEFM) models Stage II crack growth through the relationship between cyclic crack extension per cycle ($da/dN$) and the stress intensity factor range $\Delta K$.

  log(da/dN) ^
             |                                 Regime III: Rapid Tearing
             |                                 (Approaching K_c)
             |                                     /
             |               Regime II:           /
             |               Paris Power-Law     /
             |               da/dN = C(ΔK)^m    /
             |                                 /
             |                                /
             |            /-------------------
             |           /  Regime I: Near-Threshold (Dormant below ΔK_th)
             |          |
             +----------+---------------------------------------> log(ΔK)
                       ΔK_th                           K_c

The Three Regimes of Crack Growth

1. Regime I (Threshold): Below the fatigue threshold $\Delta K_{\text{th}}$ (typically $2\text{--}4\text{ MPa}\sqrt{\text{m}}$ for structural steel), crack growth is microstructurally sensitive, and crack propagation ceases for all practical engineering purposes ($da/dN < 10^{-10}\text{ m/cycle}$).
2. Regime II (Paris Law): Stable, continuous macrocrack growth governed by the Paris-Erdogan Law: $$\frac{da}{dN} = C (\Delta K)^m$$ where $\Delta K = K_{\max} - K_{\min} = Y \Delta\sigma \sqrt{\pi a}$, $Y$ is the dimensionless geometric correction factor, and $C, m$ are empirical material constants. For structural ferritic steels conforming to BS 7910 or IIW: $m \approx 3.0, \qquad C \approx 3.0 \times 10^{-13} \, \frac{\text{m/cycle}}{(\text{MPa}\sqrt{\text{m}})^3}$
3. Regime III (Unstable Tearing): As $K_{\max}$ approaches the fracture toughness $K_{Ic}$, crack growth accelerates asymptotically until rapid, final rupture occurs.

Analytical Life Integration

Integrating the Paris Law from initial detectable flaw depth $a_0$ to critical terminal flaw depth $a_f$:

$$\int_0^{N_f} dN = \frac{1}{C (Y \Delta\sigma \sqrt{\pi})^m} \int_{a_0}^{a_f} a^{-m/2} \, da$$

For structural steels with $m = 3.0$:

$$N_f = \frac{2}{C (Y \Delta\sigma \sqrt{\pi})^3} \left( \frac{1}{\sqrt{a_0}} - \frac{1}{\sqrt{a_f}} \right)$$

Key Physical Law: Because $a_0 \ll a_f$, $1/\sqrt{a_0}$ overwhelmingly dominates the equation. Over 90% of total structural fatigue life is spent growing the crack from its initial microscopic size $a_0$ to just $2\text{--}3\text{ mm}$. Detection and remediation of small toe cracks is paramount.

---