the critical crack length
A crack in a loaded part is not always a death sentence — a small one can sit there for years. But there is a size beyond which it suddenly runs away and the part fails in an instant. That threshold is the critical crack length, a_c: the largest crack the part can carry at its working stress before fast fracture takes over.
It comes straight from setting the crack's driving force equal to the material's resistance: K = K_IC. Since K = Y sigma sqrt(pi a), solving for a gives a_c = (1/pi) times (K_IC / (Y sigma))^2. Read it and you see the levers: raise the stress and the tolerable crack shrinks (as 1/sigma^2, fast); choose a tougher material (higher K_IC) and it grows. Example: steel with K_IC = 60 MPa-m^0.5 at a stress of 300 MPa (Y=1) tolerates a_c = (1/pi)(60/300)^2 = about 13 mm; double the stress to 600 MPa and a_c collapses to about 3 mm.
This number is the beating heart of damage-tolerant design. Engineers compute a_c, then demand that inspection (ultrasound, dye-penetrant, radiography, eddy-current) can reliably find cracks much smaller than a_c, and schedule re-inspection so a growing fatigue crack is caught long before it reaches a_c. If a_c comes out smaller than the smallest flaw your inspection can detect, the design is unsafe no matter how strong the steel is — you would never see the crack coming.
An aircraft skin panel is designed so its critical crack length is, say, 50 mm — long enough to be spotted at a routine check, or even to be visible as it grows across a rivet line (leak-before-break in a fuselage). Inspection intervals are set so a crack is found while still far below 50 mm.
a_c = (1/pi)(K_IC/(Y sigma))^2 — the crack size at which stable becomes catastrophic.
Critical crack length depends on the working stress, not just the material: the same steel tolerates a big crack at low stress and only a tiny one when stressed hard. It only means safety if your inspection can reliably detect cracks well below a_c.