Skip to content

Differentiation interface wrt. prev state #15

Description

@KnutAM

Currently, following Meyer and Ekre (2023), the interface assumes that the dsdp field in MaterialDerivatives is calculated neglecting the influence of dⁿsdp, and this should be added in derivatives working with stress iterations. This is not documented, and is not logical, and should be changed such that the MaterialDerivatives contain this history dependence.

This will require changes to,

  • All material implementations (adding the term ds/dⁿs * dⁿs/dp to the dsdp calculation)
  • MMB: Need interface for reset_derivatives!(::MaterialDerivatives, ::AbstractMaterial) (defaults to setting all to zero, but can be overloaded in case initial state variables depend on material parameters) and default reset_derivatives!(::MaterialDerivatives, previous::MaterialDerivatives) to transfer dsdp if the simulation continues from a previous simulation.
  • MatModFit: Need to reset MaterialDerivatives at beginning of new simulations
  • MatModFit/MMB: StressIterDerivatives must be updated removing the term ds/dⁿs * dⁿs/dp. Potentially this should be moved to MMB to better handle different stress states.

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions