Presets
Playback
Parameters
Camera
Layers
Quality
Model check
The science
What is modelled
A steel (or aluminium, titanium) bar is clamped to a wall and flexed up and down by a crank. A small flaw sits at the edge of the clamp, where the bending stress is highest. Each cycle the crack tip opens a little and the crack grows by a tiny amount; because a longer crack concentrates stress more, the growth speeds up until the rest of the bar breaks at once. The page computes that whole history, then plays it back with the cycle counter on a logarithmic scale.
Equations
- Stress intensity of a single edge crack of depth a in a bar of depth W under bending:
K = Y(a/W) σ √(π a)withY = 1.122 − 1.40 α + 7.33 α2 − 13.08 α3 + 14.0 α4,α = a/W(handbook fit, valid up to α = 0.6). - Paris law:
da/dN = C (ΔK)mforΔKabove the thresholdΔKth, and zero below it. With the stress ratio R = σmin/σmax,ΔK = Y σmax(1 − max(R,0)) √(π a)(a crack is closed in compression). - Life:
Nf = ∫ da / (C (ΔK)m)from the starting flaw a0 to the critical depth ac. For constant Y it is[ac1−m/2 − a01−m/2] / (C (1 − m/2) (Y Δσ √π)m); the page uses Simpson integration with the real Y(a/W). - Final fracture happens at
Kmax = KIcor when the remaining ligament yields,σmax = 1.5 σy (1 − α)2, whichever comes first. - Beam: tip force from
σ = 6 F (L − xc) / (B W2), tip deflectionδ = F L3 / (3 E I),I = B W3/12.
Checks run in the page
- The numerical life integral is compared with the closed form for constant Y.
- The beam curvature is integrated twice and compared with F L3/(3EI).
- The critical crack depth from the life curve is compared with an independent bisection on the fracture criterion.
- Just below the threshold stress there is no growth, just above it there is; the life scaling with stress is compared with the rough Paris power law.
Honest simplifications
- Paris constants, thresholds, toughness and yield strengths are typical handbook-order values for each material class, not data for a specific alloy or heat treatment; real scatter in life is a factor of several.
- Only crack growth is modelled: a flaw of the chosen depth is assumed to exist from the start. The initiation stage, which can dominate the life of a polished bar, is ignored. Overloads, retardation, corrosion, the near-threshold and near-KIc regimes (Forman acceleration) and the 3D shape of the crack front are ignored, as is the loss of stiffness as the crack grows.
- The crack on screen is drawn deeper than the true depth when it would be invisible (minimum 1.2 mm), and the bending is exaggerated so the tip swing is visible.
- The crank runs at a few hertz for the picture; the time-to-failure readout assumes 50 Hz. Playback time is not real time: cycle counts are compressed on a logarithmic scale.
What to try
- Raise the stress range: life falls roughly as the stress to the power m.
- Choose the "Below the threshold" preset: the same flaw never grows.
- Compare steel with aluminium at the same stress and watch the Paris plot.
Loading model...
Loading Blender model...