A crack can grow when extending it releases more elastic energy than it costs to create new fracture surfaces.

For a brittle elastic body, this idea is expressed through the energy-release rate $G$: the decrease in stored potential energy per unit new crack area. Crack growth becomes energetically possible when

$$G\ge G_c,$$

where $G_c$ is the material's critical fracture energy.

For an ideal central crack of half-length $a$ in a large elastic plate under tensile stress $\sigma$, the scaling is

$$G\propto \frac{\sigma^2 a}{E}.$$

Thus a longer flaw releases more energy at the same applied stress. Solving the Griffith balance gives the characteristic brittle-fracture scaling

$$\sigma_f\propto \sqrt{\frac{E\gamma}{a}},$$

where $\gamma$ represents surface-energy cost in the ideal brittle limit.

This explains a central fact of fracture: strength is flaw sensitive. Two specimens made from the same material can fail at very different nominal stresses if their largest cracks differ.

Real structural materials often dissipate additional energy through plasticity, microcracking or other toughening mechanisms, so $G_c$ is generally larger than the energy needed merely to make two atomically clean surfaces. The energy framework nevertheless remains valid: fracture occurs when the crack-driving energy reaches the material's resistance.