Skip to content

Thin-strip energy release rate from the uniaxial-strain modulus - #349

Open
jameskermode wants to merge 1 commit into
masterfrom
thin-strip-modulus
Open

jameskermode wants to merge 1 commit into
masterfrom
thin-strip-modulus

Conversation

@jameskermode

Copy link
Copy Markdown
Member

Problem

strain_to_G / G_to_strain convert thin-strip strain to energy release rate with

G = E / (1 - nu^2) * strain^2 * h / 2

E / (1 - nu^2) is the plane-strain modulus with sigma_xx = 0. In a thin strip, though, the region far ahead of the crack is in uniaxial strain: the fixed grips and the strip's length along the crack direction pin u_x (eps_xx = 0), and the periodic front gives eps_zz = 0. The stored energy density is then M eps_yy^2 / 2 with M = C'_yyyy, the stiffness rotated into the crack frame, so

G = M * strain^2 * h / 2

The old form underestimates G. For Si(110)[1-10] with nu = PoissonRatio_yx (as the staging thin-strip script stores it), Stillinger-Weber gives E / (1 - nu^2) = 131.8 GPa against M = 170.4 GPa: G is about 23% too low. The strain energy of a uniaxially strained Si cell measures 170.1-170.2 GPa, confirming M. For an MH-1 fine-tuned MACE the gap is 146.5 vs 175.0 GPa.

Changes

  • thin_strip_modulus(C, crack_direction, cleavage_plane, crack_front) returns M = C'_yyyy.
  • thin_strip_G(strain, M, h) / thin_strip_strain(G, M, h).
  • get_energy_release_rate uses atoms.info['ThinStripModulus']. Without it, it falls back to the old form with a UserWarning.
  • strain_to_G / G_to_strain are kept, with a note pointing to the new functions.
  • staging/fracture_mechanics/make_crack_thin_strip.py computes, stores and uses the new modulus.
  • ChangeLog entry.

Not changed here: get_stress_intensity_factor still converts G to K with E / (1 - nu^2) using PoissonRatio_yx, which is a separate (near-tip, anisotropic) approximation.

Tests

tests/test_thin_strip.py:

  • isotropic C gives lambda + 2 mu;
  • cubic (110) gives (C11 + C12)/2 + C44;
  • thin_strip_G / thin_strip_strain round-trip;
  • the strain energy of uniaxially strained SW Si matches M to 0.5% (central difference), and the old modulus is more than 20% below it;
  • get_energy_release_rate uses ThinStripModulus, and warns without it.

test_crack.py and test_cubic_crystal_crack.py still pass.

🤖 Generated with Claude Code

Far ahead of a thin-strip crack the fixed grips and the strip length leave
only eps_yy nonzero, so G = M strain^2 h / 2 with M = C'_yyyy in the crack
frame. strain_to_G / G_to_strain used E / (1 - nu^2) (sigma_xx = 0), which
underestimates G by more than 20% for Si(110)[1-10]. Add thin_strip_modulus,
thin_strip_G and thin_strip_strain; get_energy_release_rate uses
ThinStripModulus and warns on the old fallback; the staging thin-strip
script stores and uses the new modulus.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

This branch has not been deployed

No deployments
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant