Drucker–Prager yield stress convergence issues

[Glenice-Wang]

Hi,

When using the Drucker–Prager yield criterion with full pressure (not just lithostatic) dependence, convergence of the nonlinear Stokes system becomes difficult — Picard iterations may stall at large residuals, and Newton methods can generate spurious shear bands (Spiegelman et al., 2016). Solutions often only reach relatively loose tolerances (~1e−2–1e−4). How does Underworld2 handle or mitigate such convergence issues?

Thank you very much.

Reference:
Spiegelman, M., May, D. A., & Wilson, C. R. (2016). On the solvability of incompressible Stokes with viscoplastic rheologies in geodynamics. Geochemistry, Geophysics, Geosystems, 17(6), 2213–2238. https://doi.org/10.1002/2015GC006228