Version 3.0.2 of TFEL, MFront and MTest has been released on the 25th October, 2017.

This is mainly a bug fix version of the 3.0 series. All tickets solved are described below. There are no known regressions.

Tickets fixed

Ticket #93: Error declaring local variable

In certain circumstances, parsing a variable type was not consistent with C++ rules. For example, the following statement was not properly parsed:

@LocalVariable tfel::math::tvector<Nss,tfel::math::tensor<3u,real> >  mu;

For more details, see: https://sourceforge.net/p/tfel/tickets/93/

Ticket #92: Reading boolean values in Data::read_value is broken

For more details, see: https://sourceforge.net/p/tfel/tickets/92/

Ticket #91: Parts of the implicit systems and jacobian matrix blocks by the StandardElasticity brick shall be inserted at the beginning of the @Integrator block, not at the end

For more details, see: https://sourceforge.net/p/tfel/tickets/91/

Ticket #90: Wrong test on BOOST_VERSION version in bindings/python/include/TFEL/Python/SharedPtr.hxx

For more details, see: https://sourceforge.net/p/tfel/tickets/90/

Ticket #89: StandardElasticity: bug when only one the plane stress modelling hypotheses is selected

For more details, see: https://sourceforge.net/p/tfel/tickets/89/

Ticket #88: Error when treating axial growth: the axis along the growth is wrong

The main axis for the axial growth shall be the second direction of orthotropy, as described in the documentation, not the third one as implemented in versions \(3.0\) and \(3.0.1\).

For more details, see: https://sourceforge.net/p/tfel/tickets/88/

Ticket #87: The correction of the axial growth modifies all the stress free expansion treated so far

For more details, see: https://sourceforge.net/p/tfel/tickets/87/

Ticket #86: The material laws are not available in the ComputeStress block in the RungeKutta DSL

The material laws (declared for example via the @MaterialLaw keyword) were not available without extra qualification in the ComputeStress block when using the RungeKutta DSL: one have to add the mfront:: qualifier in front of the material law.

The problem is a missing call to writeMaterialLaws in the writeBehaviourParserSpecificMembers member function of the RungeKuttaDSLBase class.

For more details, see: https://sourceforge.net/p/tfel/tickets/86/

Ticket #85 Add support for an in plane “relocalisation strain” in behaviours laws

The @Relocation keyword can be used to impose a boundary condition specific the fuel performances describing the rigid body translation of fuel pellet fragment as an additional strain.

• an external state variable name.
• a string giving an mfront file implementing a relocation model.

Modelling hypotheses affected

This keyword is only effective in:

• axisymmetrical generalised plane strain.
• axisymmetrical generalised plane stress.
• generalised plane strain.
• plane strain.
• plane stress.

No expansion are added in the tridimensional and axisymmetrical modelling hypotheses.

Physical description

The value of the relocation expansion \(r\) is converted in an additional expansion in the radial and orthoradial directions:

\[ \left\{ \begin{aligned} \frac{\Delta\,l_{r}}{l_{r}} &= \frac{r}{2} \\ \frac{\Delta\,l_{\theta}}{l_{\theta}} &= \frac{r}{2} \end{aligned} \right. \].

This additional expansion is such that it does not create any stress in the fuel pellet for an isotropic elastic material with constant material properties.

The axial axis is the second direction in \(1D\) hypotheses and the third direction in \(2D\) hypotheses.

Examples

// relocation defined by an external model
@Relocation 'UO2Relocation.mfront';

// relocation defined by an external state variable
@ExternalStateVariable real r;
r.setEntryName("Relocation");
@Relocation r;

For more details, see: https://sourceforge.net/p/tfel/tickets/85/

Ticket #84: Check the NTENS value in the Abaqus interface

The value of the NTENS value was not used to check that the function called was consistent with the modelling hypothesis used.

For more details, see: https://sourceforge.net/p/tfel/tickets/84/

Ticket #82: Python interface is broken if no material name if not defined

For more details, see: https://sourceforge.net/p/tfel/tickets/82/

Ticket #81: Have two versions of MFront in the same environment

In cmake, the TFEL_APPEND_VERSION option will append the version number to the names of:

• The executables.
• The libraries.
• The python modules. Note that, to comply with python restriction on module’ names, the characters . and - are replace by _ and that only the first level modules are affected.
• The directories in the share folder.

The headers are installed in a subforder named TFEL-${TVEL_VERSION}.

For example, if the TFEL version is 3.0.2-dev, using TFEL_APPEND_VERSION opion will generate:

• The mfront-3.0.2-dev executable.
• The libTFELMaterial-3.0.2-dev.so library.
• The mtest_3_0_2_dev python module.
• The tfel_3_0_2_dev.material python module. In this case, the second level (material) is not affected.

This allows multiple executables to be installed in the same directory.

The tfel-config tool

The tfel-config tool has been modified to take this option appropriately. For example:

$ tfel-config-3.0.2-dev --includes
-I/tmp/install/include/TFEL-3.0.2-dev
$ tfel-config-3.0.2-dev --libs --material
-L/tmp/install/lib -lTFELMaterial-3.0.2-dev -lTFELMath-3.0.2-dev -lTFELUtilities-3.0.2-dev -lTFELException-3.0.2-dev

Modifications to mfront

mfront has been modified to call the appropriate tfel-config executable.

For more details, see: https://sourceforge.net/p/tfel/tickets/81/

An helper function for importing TFEL modules

import os, importlib

def python_module_version(v):
    """Return the python module extension from the TFEL version"""
    return v.replace('.','_').replace('-','_')

def import_tfel_module(n,*v):
    """Try to import a tfel module, taking into account the TFEL_VERSION environment variable."""
    if not n:
        ImportError('Empty module name')
    m   = n.split('.')
    if not v:
        env = os.environ;
        if('TFEL_VERSION' in env):
            v = env.get('TFEL_VERSION')
            m[0] = m[0]+'_'+python_module_version(v)
    elif len(v)==1:
        m[0] = m[0]+'_'+python_module_version(v[0])
    else:
        ImportError('Invalid number of arguments')
    return importlib.import_module('.'.join(m))

Ticket #80: Adding the possiblity to compile a law using python module

MFront relies on external tools, called generators, for this step (this semantic is the same than the one used by cmake).

The only available generator is currently the Makefile generator, handled by the MakefileGenerator class. This generator can be used as follows:

import mfront
dsl = mfront.getDSL("Chaboche.mfront")

dsl.setInterfaces(['castem'])
dsl.analyseFile("Chaboche.mfront")
dsl.generateOutputFiles()

m = mfront.MakefileGenerator()
m.exe(dsl.getTargetsDescription())

The exe methods provides several overloads and may accept an additional argument of type GeneratorOptions which let the user specifiy the optimisation level to be used.

For more details, see: https://sourceforge.net/p/tfel/tickets/80/

Ticket #79: Cyrano interface: bug in CyranoBehaviourHandler class. Missing test.

The problem appeared when local substepping is enabled.

This feature was not tested for the Cyrano interface, which explains why this regression has not been detected so far.

The problem was fixed and appropriate test were added.

For more details, see: https://sourceforge.net/p/tfel/tickets/79/

Ticket #78: Fix transpose of st2tost2 when using rvalue reference

The definition of the transpose of st2tost2 was broken when the argument was an rvalue reference.

For more details, see: https://sourceforge.net/p/tfel/tickets/78/

Ticket #77: operators *= and /= are broken for fourth order tensors

The operators *= and /= are broken for fourth order tensors: the std::is_same metafunction return value must be accessed with the value member, not the cond member. This was a reminiscence an old remove class called tfel::meta::IsSame which was removed in TFEL-3.0.

For more details, see: https://sourceforge.net/p/tfel/tickets/77/

Ticket #75: Support for python3 in the python interface

The code generated by mfront with the python interface is not compatible with python3. The trouble is linked to changes in the module declaration. See https://docs.python.org/3/howto/cporting.html for details.

For more details, see: https://sourceforge.net/p/tfel/tickets/75/

Ticket #73 AbaqusExplicit: wait for tasks termination when the ThreadPool threading model is used

The ThreadPool class now has a wait method that stops the main thread up to tasks completion. This method is now called at the end of the integration in the AbaqusExplicit interface when the ThreadPool threading model is used.

For more details, see: https://sourceforge.net/p/tfel/tickets/73/

Ticket #72: Error at compilation of documentation

The latex documents declared a unnecessary dependency to the stmaryrd package. This package became outdated in recent LaTeX distributions causing some portability issues. This dependency has been removed.

The inclusion of the babel package was missing in the docs/presentations/mfront-formation-tp/tp.tex.in file.

For more details, see: https://sourceforge.net/p/tfel/tickets/72/

Ticket #69: @Profiling true; Error in cygwin

Under Windows, the __declspec(dllexport) attribute does not propagate to inner structures, wheres the visibility flags used under POSIX systems does. This attribute is now set also on the BehaviourProfiler::Timer structure.

For more details, see: https://sourceforge.net/p/tfel/tickets/69/

Ticket #70: PLASTIC + MFRONT in CalculiX

CalculiX provides two interfaces to ABAQUS umat behaviours. In versions prior to 2.12, those interfaces are not fully compatible with ABAQUS concerning various checks. The issue has been reported to CalculiX developers.

To circumvent those issues, the MFRONT_ABAQUS_NORUNTIMECHECKS flag has been added. This flag desactivates:

• checks related to the number of material properties (value of NPROPS)
• checks related to the number of state variables (value of NSTATV)
• checks related to usage of the behaviour in finite strain analysis (KSTEP)

All those checks may fail with the ABAQUS and ABAQUSNL interfaces in CalculiX for reasons described in the previous posts. This flag can be set as follows:

mfront --obuild --interface=abaqus -D MFRONT_ABAQUS_NORUNTIMECHECKS SaintVenantKirchhoffElasticityTotalLagrangian.mfront

For more details, see: https://sourceforge.net/p/tfel/tickets/70/

Ticket #67 A suffix to the material name is required (abaqus interface)

The material name in Abaqus/Standard and Abaqus/Explicit are now expected to have the following form: LIBRARY_FUNCTION_HYPOTHESIS_IDENTIFIER.

The last part (_IDENTIFIER) is optional.

For more details, see: https://sourceforge.net/p/tfel/tickets/67/

Ticket #64: Change the library prefix to “cyg” under cygwin

The default library prefix under cygwin has been changed to cyg (rather than lib) to follow the standard conventions of this environment.

For more details, see: https://sourceforge.net/p/tfel/tickets/84/

Ticket #63: Error cygwin (make)

Compiling MFront under Cygwin was broken because the _GNU_SOURCE macro was not defined in the following headers:

• mfront\tests\properties\test-cxx.cxx
• src\System\ProcessManager.cxx
• src\System\SignalManager.cxx

For more details, see: https://sourceforge.net/p/tfel/tickets/83/

Ticket #58: Fix initialisation of an array of parameters

For more details, see: https://sourceforge.net/p/tfel/tickets/58/

Ticket #57: More robust detection of python versions: avoid mismatch between the python libraries, the python interpreter and the boost python library

For more details, see: https://sourceforge.net/p/tfel/tickets/57/

Ticket #54: Bug in StensorFromTinyVectorView: standard assignement fails

For more details, see: https://sourceforge.net/p/tfel/tickets/54/

Ticket #53: Add comments handling in parameters txt files

For more details, see: https://sourceforge.net/p/tfel/tickets/53/

Ticket #52: More robust handling of parameters txt file

For more details, see: https://sourceforge.net/p/tfel/tickets/52/

Ticket #20 Allow MFront to work under Windows without requiring MSYS

The documentation describing how to install TFEL on Windows has been updated: The main difficulty associated with building TFEL/MFront on Windows plateform is the lack of an unified build environment which must be set up by the user.

The installation on Windows plateform is thus described in the following pages:

• Installation based on cmake and Visual Studio.
• Installation based on cmake and MinGW (as packaged with Cast3M 2017). This tutorial can easily be adapted to other versions of MinGW, without requiring Cast3M to be installed.
• Installation based on MSYS.

The two first solutions does not depend on MSYS.

For more details, see: https://sourceforge.net/p/tfel/tickets/20/

Ticket #17: Add physical constants from CODATA

Various physical constants are now available. The values are extracted from: http://physics.nist.gov/cuu/Constants/index.htm

For more details, see: https://sourceforge.net/p/tfel/tickets/17/

C++

The physical constants are available as constexpr static members of tfel::PhysicalConstants class defined in the TFEL/PhysicalConstants.hxx header.

using Cste   = tfel::PhysicalConstants<real>;
const auto R = Cste::R;

Python

The physical constants are available as members of the tfel.PhysicalConstants class, as follows:

from tfel import PhysicalConstants as Cste
print(Cste.Na)