Skip to content

August 2023

Warning

The Release notes have yet not been fully transferred to the HTML documentation. Contact the developers if you have specific questions.

The release notes for August 2023 are:

Features:

  • Pre-built Windows version

  • Procedures

    • Explicit dynamics (yet not available in the public version), see the Reference Manual

  • Constitutive models

    • Fredlund & Xing hydraulic model, see Section 5.5.5.4{reference-type="ref" reference="sec:swrc_fredlundxing"}

    • The Modified-Cam-Clay implementation has been revised to match the generalized MCC model implemented in Abaqus, see Section 5.5.2.11{reference-type="ref" reference="sec:mat_modified_cam_clay"}

    • Sanisand-2: A new implementation of the well-known Sanisand model, created in collaboration with Mahdi Taiebat and Sheng Zheng. See Section 5.5.2.8{reference-type="ref" reference="sec:mat_sanisand2"}.

    • Hypo-ISA: we revised the implementation and the interface to the model. See Section 5.5.2.6{reference-type="ref" reference="sec:mat_hypoplasticity_isa"}.

  • Contact

    • General contact: Surface pairs don't need to be specified but are searched for automatically

    • Stiffness-dependent tangential penalty factor

    • User can choose if the normal penalty factor may change during the analysis

  • Solver

    • New solver: *Solver, simple: A simple implementation of a direct solver using serial forward substitution. Should only be used for very small systems, e.g. single-element tests.

    • New solver: *Solver, Diagonal: A simple implementation of a direct solver for diagonal matrices Should only be used for explicit dynamics with mass lumping.

    • Extrapolation keyword no longer part of *Solver

  • Spatially fluctuations for initial statevs using auto-correlation and Markov's approach

  • Multiple *Initial conditions, type = statevariables of different type now possible

  • User-defined load: Non-linear (with respect to coordinate) load is now possible

  • New option to control the automatic time stepping based on a maximum allowed solution increment, see Section 5.11.3{reference-type="ref" reference="sec:ctrl_max_solution_increment"}

  • Amplitudes, see Section 5.6{reference-type="ref" reference="input_amplitude"}

    • lab-cyclic-stress-strain-control: New amplitude designed for undrained cyclic loading in laboratory tests.

    • lab-triaxial-constant-pressure: New amplitude designed for axially loaded triaxial tests with constant mean effective stress.

Resolved bugs:

  • The scaling factor of the line search algorithm is now properly documented in the message file (*.msg)

  • Pore water/air pressure from *Restart no longer overwritten by initial pore water/air pressure

  • Fixed missing nodes in the master node set for *Equation in special cases where the slave node sets contain more nodes than the master node set

  • Rodriguez rotation now possible in 2D

  • Bugfix of string passing through interfaces to the user routines for initial stress and initial state variables. Note that the interface and example coding changed slightly.

General modifications:

  • Improved numerical stability in infiltration processes

    • Added option to calculate the Jacobian of the (effective) degree of saturation \(\partial S^e / \partial p^c\) as a smoothed representative of the Jacobian evaluated over 10 substeps. This has proven to enhance convergence and numerical stability during infiltration processes, especially when transitioning from nearly saturated (e.g. \(S^e=0.9\)) to fully saturated (\(S^e=1\)). Evaluation of the Jacobian is further available via the analytic derivation or approximated by a finite divided difference approximation (chord slope).

    • Same options are available for the Jacobian of the relative permeability \(\partial k^r / \partial S^e\) although the performance was not significantly enhanced.

  • Constitutive models

    • Substepping scheme for integration of intergranular strain leads to somewhat deterministic results (non-deterministic results for multiple processors prior to the substepping scheme caused by round-off errors)

    • Tolerated error in integration of stress rate of hypoplastic model decreased to \(10^{-4}\) (used to be \(10^{-3}\))

  • Contact

    • Default stiffness multiplicator for normal penalty factor increased to 20 (formerly 12)

    • Contact sparsity is only evaluated for the ten-closest points if the user does not use nlgeom. This makes the evaluation much faster but is not valid for large deformation problems. The user gets a message if he should be using nlgeom since otherwise the stiffness of the contact can not be added to the SOE resulting in inferior convergence.

    • Faster contact since the allocation of surfaces is now only done once, implying that the number of surfaces stays constant during the calculation

  • Changed output name in Paraview files of stresses resulting from phantom elasticity, see Section 5.5.12{reference-type="ref" reference="sec:mat_phantom_elasticity"}

  • Resolved bug occurring when requesting statevariables as output in simulations using user-defined materials.

  • User files are now automatically compiled at the beginning of an analysis

  • Non-local smoothing only for specified element set, not all elements of the model

Corrections Keywords Manual:

  • Corrected minor mistake in Equations for relative permeability in Nguyen and Brooks & Corey model, see Sections 5.5.8.1{reference-type="ref" reference="sec:krel_nguyen"} and 5.5.8.2{reference-type="ref" reference="sec:krel_brookscorey"}

  • Corrected mistake in the user-defined material interface

  • Relative permeability "Logistic-fit" (see Section 5.5.8.4{reference-type="ref" reference="sec:krel_logistic"}) had a type in the models name

  • More information about using (see Section 5.19.2{reference-type="ref" reference="sec_out_print"}) and (see Section 5.19.3{reference-type="ref" reference="sec_user_print"}).