Fracture density modeling

The fracture density approach in PISM is based on [92] and assumes a macroscopic measure for the abundance of (partly microscale) crevasses and rifts that form in ice (shelves) and that can be transported with the ice flow as represented in a continuum ice-flow model. This approach is similar to the Continuum Damage Mechanics (CDM) (e.g. [93] and [94]) introducing a damage state variable (\(\phi\) or \(D\)) that equals zero for fully intact ice and one for fully fractured ice, that can be interpreted as a loss of all load bearing capacity.

The feedback of damage to the ice flow (creep) works within the existing constitutive framework by introducing a linear mapping between the actual physical (damaged) state of the material and an effective state that is compatible with a homogeneous, continuum representation of the creep law (Eq. 6 in [95]).

Fractures form above a critical stress threshold \(\sigma_{\text{cr}}\) in the ice (e.g. von Mises criterion, maximum stress criterion or fracture toughness from Linear Elastic Fracture Mechanics) with a fracture growth rate proportional to \(\gamma\) (Eq. 2 in [95]), that is related to the strain rate (longitudinal spreading or effective strain rate; Eq. 9 in [92]). Fracture healing is assumed to occur with a defined healing rate below a strain rate threshold (scaled with the difference to the threshold or constant; Eq. 11 in [92]).

The fracture growth constant \(\gamma\) (fracture_density­.gamma) is ignored if fracture_density­.borstad_limit is set.

To enable this model, set fracture_density­.enabled.


Prefix: fracture_density.

  1. borstad_limit (no) Model fracture growth according to the constitutive law in [96] (Eq. 4), ignoring fracture_density­.gamma.

  2. constant_fd (no) Keep fracture density fields constant in time but include its softening effect.

  3. constant_healing (no) Use a constant healing rate \(-\gamma_h \dot{\epsilon}_h\) independent of the local strain rate.

  4. fd2d_scheme (no) Use an alternative transport scheme to reduce numerical diffusion (Eq. 10 in [95])

  5. fracture_weighted_healing (no) Multiply the healing rate by \(1 - D\), i.e. assume that highly damaged ice heals slower. This mechanism can be combined with fracture_density­.constant_healing.

  6. gamma (1) fracture growth constant \(\gamma\)

  7. gamma_h (0) fracture healing constant \(\gamma_{h}\)

  8. healing_threshold (2e-10 1/s) fracture healing strain rate threshold \(\dot \epsilon_{h}\)

  9. include_grounded_ice (no) Model fracture density in grounded areas (e.g. along ice stream shear zones) in addition to ice shelves

  10. initiation_threshold (70000 Pa) fracture initiation stress threshold \(\sigma_{\text{cr}}\)

  11. lefm (no) Use the mixed-mode fracture toughness stress criterion based on Linear Elastic Fracture Mechanics, Eqs. 8-9 in [95]

  12. max_shear_stress (no) Use the maximum shear stress criterion for fracture formation (Tresca or Guest criterion in literature), which is more stringent than the default von Mises criterion, see Eq. 7 in [95]

  13. phi0 (0) Fracture density value used at grid points where ice velocity is prescribed. This assumes that all ice entering a shelf at bc_mask locations has the same fracture density.

  14. softening_lower_limit (1) Parameter controlling the strength of the feedback of damage on the ice flow. If \(1\): no feedback, if \(0\): full feedback (\(\epsilon\) in Eq. 6 in [95])


See the scripts in example/ross/fracture for a way to test different damage options and parameter values. Build a setup for the Ross Ice Shelf and let the damage field evolve, with fracture bands reaching all the way from the inlets to the calving front.

Previous Up Next