The page describes the new functionalities of Version 5.0.0 of the TFEL project.

Version 5.0.0 has been released on December, 17th 2024. This version inherits from the bug fixes of:

• Version 3.0.14,
• Version 3.1.14,
• Version 3.2.11,
• Version 3.3.6.
• Version 3.4.7.
• Version 4.0.4.
• Version 4.1.3.
• Version 4.2.2.

1 Tested configurations

There are currently \(20\,576\) unit tests.

Versions of clang prior to Version \(17\) are known not to work.

2 New TFEL/Utilities features

2.1 Support for C++ digit separator in CxxTokenizer

The CxxTokenizer class now supports C++ digit separators in numbers, (integer of floatting point numbers), as illustrated in the following example:

@DSL IsotropicMisesCreep;
@Behaviour DigitSeparatorTest;

@UseQt true;
@Epsilon 0.000'000'000'1;
@IterMax 1'000;

@ElasticMaterialProperties{150'000'000'000, 0.3};
@Parameter stress s0 = 10'000'000;
@Parameter strainrate de0 = 8.230'512e-67;
@Parameter strainrate E = 8.2;

@FlowRule{
  f = de0 * pow(seq / s0, E);
}

3 New TFEL/Math features

3.1 Tiny matrices product

The product of two tiny matrices has been implemented:

const auto m1 = tmatrix<2u, 2u, int>{0, 1,  //
                                     2, 3};
const auto m2 = tmatrix<2u, 2u, int>{4, -1,  //
                                     5, 2};
const auto m3 = m1 * m2;

3.2 New solver eigen solver

A new eigen solver based on Harari’s analytical solution have been introduced for symmetric tensors. The computation of eigen values is done with Harari’s algorithm [1] and the computation of eigen vectors is done with the default eigen solver for symmetric tensors of TFEL. Such computations are more efficient and more accurate than the default TFEL algorithm.

Those algorithms are available in 3D. For 2D symmetric tensors, we fall back to some default algorithm as described below.

The various eigen solvers available are enumerated in Table 1.

The eigen solver is passed as a template argument of the computeEigenValues or the computeEigenVectors methods as illustrated in the code below:

tmatrix<3u,3u,real> m2;
tvector<3u,real>    vp2;
std::tie(vp,m)=s.computeEigenVectors<stensor::HARARIEIGENSOLVER>();

3.2.0.1 Some benchmarks

We have compared the available algorithm on \(10^{6}\) random symmetric tensors whose components are in \([-1:1]\).

For a given symmetric tensor, we consider that the computation of the eigenvalues and eigenvectors failed if: \[ \Delta_{\infty}=\max_{i\in[1,2,3]}\left\|\underline{s}\,\cdot\,\vec{v}_{i}-\lambda_{i}\,\vec{v}_{i}\right\|>10\,\varepsilon \] where \(\varepsilon\) is the accuracy of the floatting point considered.

The results of those tests are reported on Tables 2, 3 and 4. The Harari eigen solver offers a better compromise between accuracy and numerical efficiency than the default TFEL solver.

4 New TFEL/Material features

4.1 Homogenization

The homogenization functions are part of the namespace tfel::material::homogenization. A specialization for elasticity is defined: tfel::material::homogenization::elasticity, in prevision of further homogenization developments in other physics.

4.1.1 Ellipsoidal inclusion embedded in isotropic matrix

The function computeEshelbyTensor computes the Eshelby tensor of an ellipsoid whose semi-axis lengths are a, b, c, embedded in an isotropic matrix. There is also computeSphereEshelbyTensor, computeAxisymmetricalEshelbyTensor, and also computeCircularCylinderEshelbyTensor and computeEllipticCylinderEshelbyTensor for plane strain elasticity.

Three functions also compute the strain localisation tensor of an ellipsoid embedded in an isotropic matrix and submitted to an external uniform strain field : computeEllipsoidLocalisationTensor, computeAxisymmetricalEllipsoidLocalisationTensor and computeSphereLocalisationTensor.

4.1.2 Homogenization schemes

Different schemes are implemented and return the homogenized stiffness of the material. The available functions are computeMoriTanakaScheme, computeDiluteScheme, computeSphereDiluteScheme, computeSphereMoriTanakaScheme. If a distribution of ellipsoids is considered, three types of distributions are considered. The corresponding functions are computeIsotropicDiluteScheme, computeTransverseIsotropicDiluteScheme, computeOrientedDiluteScheme, computeIsotropicMoriTanakaScheme, computeTransverseIsotropicMoriTanakaScheme and computeOrientedMoriTanakaScheme.

5 MFront

5.1 Improvements to the MaterialProperty DSL

5.1.1 The @Data keyword

The @Data keyword allow the definition of a material properties using the interpolation of a set of values.

The @Data keyword is followed by a set of options defining:

• the interpolation option,
• the set of values,
• the extrapolation option.

The precise syntax depends on the number of inputs of the material property.

5.1.1.1 Material properties without input

The only option accepted is value.

5.1.1.2 Material properties with only one input

The values option is required. It must be a map associating values of the input and values of the output.

The interpolation option is optional. It must be a string. The values linear and cubic_spline are accepted.

The extrapolation option is optional. It must be a boolean or a string:

• The boolean value true means that a linear extrapolation will be performed.
• The boolean value false or the string value bound_to_last_value or the string value constant means that a constant value will be used for extrapolation.

5.1.1.3 Example of usage

@DSL MaterialProperty;
@Law LinearDataInterpolation;

@UseQt true;
@UnitSystem SI;

@Output stress E;
E.setGlossaryName("YoungModulus");

@StateVariable temperature T;
T.setGlossaryName("Temperature");

@Data {
  values: { 293.15 : 240e9, 693.15 : 180e9, 893.15 : 170e9 },
  interpolation : "linear"
};

5.2 Improvements to the StandardElastoViscoPlasticity brick

5.2.1 Strain rate sensitive isotropic hardening rule

The StrainRateSensitive isotropic hardening rule is defined by: \[ R{\left(p, \dot{p}\right)} = R_{0}{\left(p\right)}\,R_{rs}{\left(\dot{p}\right)}, \]

where:

• \(R_{0}{\left(p\right)}\) is the yield radius corresponding to an infinitly slow loading. It can be built by summing any isotropic hardening rule already implemented in the `StandardElastoViscoPlasticity brick.
• \(R_{rs}{\left(\dot{p}\right)}\) is a correction describing the strain rate sensitivity. \(R_{rs}{\left(0\right)}\) must be equal to 1.

This isotropic hardening rule can be parametrised using the following options:

• rate_independent_isotropic_hardening: this option introduces a contribution to \(R_{0}{\left(p\right)}\). This option can be repeated multiple times.
• rate_sensitivity_factor: this option introduces the rate sensivity factor \(R_{rs}{\left(\dot{p}\right)}\). The list of available rate sensivity factors is given in section 5.2.1.2.

5.2.1.1 Example of usage

@Brick StandardElastoViscoPlasticity{
  stress_potential : "Hooke" {young_modulus : 210e9, poisson_ratio : 0.3},
  inelastic_flow : "Plastic" {
    criterion : "Mises",
    isotropic_hardening : "StrainRateSensitive" {
      rate_independent_isotropic_hardening :
          "Swift" {R0 : "alpha * Ks * (p0s ** ns)", p0 : "p0s", n : "ns"},
      rate_independent_isotropic_hardening : "Voce" {
        R0 : "(1 - alpha) * Q1 * (1 - Q2)",
        Rinf : "(1 - alpha) * Q1",
        b : "Q3"
      },
      rate_sensitivity_factor :
          "CowperSymonds" {dp0 : "dp0cs", n : "ncs", Rs_eps : 1e-8}
    }
  }
};

5.2.1.2 List of rate sensitivity factors

• Cowper-Symonds’s rate sensitivity factor: \[ R_{rs}{\left(\dot{p}\right)}=1+A\,\left(\frac{\dot{p}}{\dot{p}_{0}}\right)^{n} \] When the exponent \(n\) is lower than one, the following regularised version, based on the user defined value \(R_{\epsilon}\) is available: \[ R_{rs}{\left(\dot{p}\right)}=1+ \left\{ \begin{aligned} R_{\epsilon}\,\left(\frac{\dot{p}}{\dot{p}_{\epsilon}}\right) &&\text{if}&&\dot{p}<\dot{p}_{\epsilon}\\ A\,\left(\frac{\dot{p}}{\dot{p}_{0}}\right)^{n} &&\text{if}&&\dot{p}\geq\dot{p}_{\epsilon} \end{aligned} \right. \] where \(\dot{p}_{\epsilon}\) is such that \(R_{\epsilon} = A\,\left(\frac{\dot{p}_{\epsilon}}{\dot{p}_{0}}\right)^{n}\)

• Johnson-Cook’s rate sensitivity factor: \[ R_{rs}{\left(\dot{p}\right)}=1+ \left\{ \begin{aligned} 0 &&\text{if}&&\dot{p}<\dot{p}_{0}\\ A\,\log\left(\frac{\dot{p}}{\dot{p}_{0}}\right)&&\text{if}&&\dot{p}\geq\dot{p}_{0} \end{aligned} \right. \]

5.3 New DSL options

5.3.1 The disable_runtime_checks option

If this option is set to true, interfaces may disable as much runtime checks as possible. Those runtime checks include checking standard bounds and physical bounds for instance.

5.4 generic interface improvements

5.4.1 The @SelectedModellingHypothesis and @SelectedModellingHypotheses keywords

The @SelectedModellingHypothesis and @SelectedModellingHypotheses keywords allows to select which modelling hypotheses will be generated.

Their syntaxes are similar to the keywords @ModellingHypothesis and @ModellingHypotheses, respectively.

The selected modelling hypotheses must be a sub-set of the modelling hypotheses supported by the behaviour.

5.4.1.1 Example of usage

$ mfront --obuild --interface=generic --@SelectedModellingHypothesis=PlaneStrain Plasticity.mfront
Treating target : all
The following library has been built :
- libBehaviour.so :  Plasticity_PlaneStrain

6 Documentation

The page Libaries usage in C++ describe how to use the TFEL libraries in C++ projects, using either the tfel-config utility or cmake packages.

7 Issues fixed

7.1 Issue 535: [mfront-doc] add support for material properties

Allows mfront-doc to be used with the MaterialLaw DSL. Example of use:

$ mfront-doc T91MartensiticSteel_gamma1_ROUX2007.mfront

For more details, see https://github.com/thelfer/tfel/issues/535

7.2 Issue 582: [cast3m interface] add explicit names and 4-letter mapping of internal variables to the castem file

Adds the correspondence between variable names and the 4-letter names used in castem to the example .dgibi file, generated by MFront with castem interfaces. The mapping is written in the file as a comment, for example in the form :

** List of material properties:
**
** - YOUN: YoungModulus
** - NU: PoissonRatio
** - RHO: MassDensity
** - H: HardeningSlope
** - SO: Yield strength

For more details, see https://github.com/thelfer/tfel/issues/582

7.3 Issue 585: mfront Allow to override inputs in material properties

$ mfront-query --state-variables Inconel600_YoungModulus.mfront 
- Temperature (TK): the temperature
$ mfront-query --parameters Inconel600_YoungModulus.mfront

$ mfront-query --state-variables --dsl-option='overriding_parameters:{"TK": 400}}' Inconel600_YoungModulus.mfront
$ mfront-query --parameters --dsl-option='overriding_parameters:{"TK": 400}}' Inconel600_YoungModulus.mfront
- Temperature (TK): the temperature

For more details, see https://github.com/thelfer/tfel/issues/585

7.4 Issue 567: MTestCurrentState.copy() produces a shallow copy in python bindings

This behaviour is consistent with the copy constructor of the StudyCurrentState class.

To make a deep copy of this object, the makeDeepCopy method has been introduced.

For more details, see https://github.com/thelfer/tfel/issues/567

7.5 Issue 557: [mfront] allow to specify dsl options in MFrontBase::getDSL

For more details, see https://github.com/thelfer/tfel/issues/557

7.6 Issue 556: [cmake] export compile flags in dedicated variables

For more details, see https://github.com/thelfer/tfel/issues/556

7.7 Issue 555: [cmake] better handling of dependencies in exported cmake files

For more details, see https://github.com/thelfer/tfel/issues/556

7.8 Issue 526: The @UseQt keyword is not mentioned in the MaterialLaw’s keywords help page

For more details, see https://github.com/thelfer/tfel/issues/526

7.9 Issue 476: [generic interface] Add support for arrays of thermodynamic forces

For more details, see https://github.com/thelfer/tfel/issues/476

7.10 Issue 370: [tfel-utilities] Support for C++ digit separator

For more details, see https://github.com/thelfer/tfel/issues/370