ahartloper / uvc_matmod Goto Github PK

It would be nice to have the return-mapping in stress spaces with constraints other than plane-stress. This has an easy implementation by creating an abstract method that defines the C and Q matrices, and their eigen-decompositions. Then various constrained stress spaces can be implemented by just defining the matrices and decompositions.

There are no testing files for the material at the moment, please add some input files to validate the material model!

See line 57 in RESSForLabMA.for.

When I try to run the OpenSees test file for the 3b_brick using the following material parameters: nDMaterial UVCmultiaxial 1 185970.0 0.3 332.18 120.48 8.14 93.15 261.75 2 21102.0 173.6 2300.6 10.42, the code diverges the first time the reverse mapping procedure is used.

When return mapping does not converge XIT is called. When the strain increment is very large the return mapping should be by-passed and the time step should be reduced.

The following initial state does not converge using UVCuniaxial:

Strain increment = -5.77349738E-06
Trial stress = -41610.0540172518
Initial state variables = 1.18028336E-04 -298.243073 -11.1183214

With parameters
PROPS(1:10) = [ 1.90481E7, 41500., 14147., 15.2, 13595., 211.16, 2.5621E6, 235.12, 94218., 3.16 ]

Probable cause is using single precision for the backstress components, i.e., lines

The Abaqus versions are not compatible when an initial stress is defined in the elements.