Introduction

This document introduces the Burger_EDF_CIWAP_2021 which aims at modelling the following phenomena in concrete:

It is an isotropic constitutive law for all phenomena.

Two variants, described in this document, of the implementation of this behaviour are available in the MFrontGallery project:

The ACES project and the VERCORS benchmark

The Burger_EDF_CIWAP_2021 is shared on the MFront gallery in the framework of the European project ACES [1, 2], in order to give acces to the VERCORS benchmark 3rd participants [3].

Description of the behaviour

The Burger mechanical model was originally developed for EDF R&D by [46]. It was then modified in order to better represent long term basic creep according to [7], and implemented as a code_aster law and then in MFront.

The version of the law described in the present document and named Burger_EDF_CIWAP_2021 is different from the one implemented in code_aster, which is described in [6], on two points:

Strain decomposition

The total strain is the sum of 4 components:

\[ \underline{\varepsilon}^{\mathrm{to}}= \underline{\varepsilon}^{\mathrm{el}}+ \underline{\varepsilon}^{\text{shr}} + \underline{\varepsilon}^{\text{bc}} + \underline{\varepsilon}^{\text{dc}} \qquad{(1)}\]

where:

Drying shrinkage

The variation of drying shrinkage \({\dot{\underline{\varepsilon}}}^{\text{shr}}\) is considered proportional to the variation of relative humidity \(\dot{h}\):

\[ \dot{\underline{\varepsilon}}^{\text{shr}} = \dot{\underline{\varepsilon}}^{\text{shr}}\,\underline{I} = k^{\text{shr}}\dot{h}\,\underline{I} \qquad{(2)}\]

Elastic and basic creep strain

The combination of elastic and basic creep strains is known as a Burger rheological model. Elasticity and basic creep are isotropic and hence, the Burger model is duplicated in deviatoric and spherical chains (“d” relates to deviatoric while “s” relates to spherical).

For each chain the strain is composed of elastic strain (\(\varepsilon_{s}^{\text{el}}\) and \(\underline{\varepsilon}_{d}^{\text{el}}\)), reversible basic creep modeled by the Kelvin-Voigt element (\(\varepsilon_{s}^{\text{rbc}}\) and \(\underline{\varepsilon}_{d}^{\text{rbc}}\)) and irreversible basic creep modeled by the Maxwell element by using a viscosity which is not constant in time and dependent on the current state of strain (\(\varepsilon_{s}^{\text{ibc}}\) and \(\underline{\varepsilon}_{d}^{\text{ibc}}\)):

\[ \underline{\varepsilon}^{\text{bc}}=\underline{\varepsilon}^{\text{rbc}}+\underline{\varepsilon}^{\text{ibc}}= \varepsilon_{s}^{\text{rbc}}\,\underline{I}+\underline{\varepsilon}_{d}^{\text{rbc}}+ \varepsilon_{s}^{\text{ibc}}\,\underline{I}+\underline{\varepsilon}_{d}^{\text{ibc}} \]

The equations related to this Burger model are now recalled for the spherical chain (the deviatoric can be obtained by replacing “s” by “d” and using tensors instead of scalars).

The elastic strain is simply proportional to the stress (here written using the elastic bulk modulus \(k^{\text{el}}\) and shear coefficient \(\mu^{\text{el}}\) :

\[ \left\{ \begin{aligned} \varepsilon_{s}^{\text{el}} &= \frac{1}{3\,k^{\text{el}}}\,p\\ \underline{\varepsilon}_{d}^{\text{el}} &= \frac{1}{2\,\mu^{\text{el}}}\,\underline{\sigma}_{d} \end{aligned} \right. \qquad{(3)}\]

The basic creep strains are assumed proportional to the relative humidity. The reversible basic creep is governed by the equation:

\[ h\,p = k^{\text{rs}}\,\varepsilon_{s}^{\text{rbc}} + \eta^{\text{rs}}\,{\dot{\varepsilon}}_{s}^{\text{rbc}} \qquad{(4)}\]

while the irreversible basic creep follows the equation:

\[ h\,p = \eta^{\text{is}}{\dot{\varepsilon}}_{s}^{\text{ibc}} \qquad{(5)}\]

The viscosity of the irreversible contribution is assumed to be dependent on the current state of irreversible strains following the equation:

\[\eta^{\text{is}} = \eta_{0}^{\text{is}}\exp{\left( \frac{\left\|\underline{\varepsilon}_{m}^{\text{ibc}}\right\|}{\kappa} \right)}\]

where \(\left\|\underline{\varepsilon}_{m}^{\text{ibc}}\right\|= \max_{\tau\in[0,t]}\sqrt{\underline{\varepsilon}^{\text{ibc}}{\left(\tau\right)}\,\colon\,\underline{\varepsilon}^{\text{ibc}}{\left(\tau\right)}}\).

This feature induces non-linearity in the creep model. The treatment of this non-linearity is one of the most involved aspect of the proposed implementation and will be detailed in Section 2.3.

The model is also thermo-activated by adding a temperature dependence to the parameters of the previous equations. Parameters \(k^{\text{rs}}\), \(k^{\text{rd}}\), \(\eta^{\text{rs}}\), \(\eta^{\text{rd}}\), \(\eta^{\text{is}}\) and \(\eta^{\text{id}}\) have the following temperature dependence exemplified on \(k^{\text{rs}}\):

\[ k^{\text{rs}}{\left( T \right)} = k_{0}^{\text{rs}}\exp{\left( \frac{Q}{R}{\left( \frac{1}{T} - \frac{1}{T_{0}} \right)} \right)} \qquad{(6)}\]

Also,

\[ \kappa = {{\displaystyle \frac{\displaystyle \kappa_{0}}{\displaystyle \exp{\left( \frac{Q}{R}{\left( \frac{1}{T} - \frac{1}{T_{0}} \right)} \right)}}}} \qquad{(7)}\]

where \(Q\) is the activation energy for creep.

Drying creep strain

Finally, the desiccation creep is computed according to a modified version of a law called Bažant’s law, relating the excess of strain rate during drying to the rate of change of the water concentration [9]. It is supposed in this version of the model that desiccation creep occurs with a zero Poisson’s ratio and that its rate is proportional to the absolute value of the relative humidity rate. Also, following a proposition by Boucher [8, 10] taking inspiration in Benboudjema [4] who noted that the behavior is not symmetric between drying and imbibition. Here the imbibition desiccation strain is considered negligible. Also, the desiccation creep occurs only during first drying down to a given relative humidity.

\[ \dot{\underline{\varepsilon}}^{\text{dc}} = \frac{1}{\eta_{\text{fd}}}\underline{\sigma}\left\langle \dot{h} \right\rangle \qquad{(8)}\]

where:

\[ \langle\dot{h}\rangle = \left\{ \begin{aligned} \dot{h} &\quad\text{if}\quad \dot{h} < 0 \quad\text{and}\quad h{\left(t\right)} < \min_{\tau\in [0,t]} h{\left(\tau\right)}\\ 0 &\quad\text{otherwise} \end{aligned} \right. \]

Description of the implicit algorithm

To integrate this behaviour, we use an implicit scheme described in this page [11]. This scheme turns the constitutive equations, which forms a system of ordinary differential equations, into a system of non linear equations.

Such scheme is usually more efficient than explicit integration schemes based on (explicit) Runge-Kutta methods and allows deriving the consistent tangent operator which can lead to a quadratic convergence of the equilibrium at the global scale.

A standard Newton-Raphson algorithm will be used to find the solution of the implici system, which requires to compute the jacobian of the implicit system.

Choice of the state variables and auxilary state variables

The role of the implicit scheme is to determine the following increments:

\[ \Delta\,\underline{\varepsilon}^{\mathrm{el}},\,\Delta\,\underline{\varepsilon}^{\text{shr}},\, \Delta\,\underline{\varepsilon}^{\text{bc}},\,\Delta\,\underline{\varepsilon}^{\text{dc}},\,\Delta\,\left\|\underline{\varepsilon}_{m}^{\text{ibc}}\right\| \]

For convenience, the basic creep contribution \(\underline{\varepsilon}^{\text{bc}}\) is explicitly slitted into its reverible \(\underline{\varepsilon}^{\text{rbc}}\) and irreversible part \(\underline{\varepsilon}^{\text{ibc}}\). The set of increments to be determined is then:

\[ \Delta\,\underline{\varepsilon}^{\mathrm{el}},\,\Delta\,\underline{\varepsilon}^{\text{shr}}, \,\Delta\,\underline{\varepsilon}^{\text{rbc}}, \,\Delta\,\underline{\varepsilon}^{\text{ibc}},\,\Delta\,\underline{\varepsilon}^{\text{dc}},\,\Delta\,\left\|\underline{\varepsilon}_{m}^{\text{ibc}}\right\| \]

For one of those increments, one can either, as shown below:

  1. Find an implicit equation associated with this increment. In this case, the associated variable is said to be a state variable.
  2. Express this increment as a function of the other increments. In this case, the associated variable is said to be an auxiliary state variable. One obvious auxiliary state variable is the shrinkage strain \(\underline{\varepsilon}^{\text{shr}}\) which only depends on the variation of the relative humidity. Note that MFront does not automatically defines the increment associated with an auxiliary state variable.

From an numerical point of view, using an auxiliary state variable rather than a state variable is interesting as it reduces the size of the implicit system to be solved. However, it generally leads to more complex implementations.

For reasons that will become clearer later when we will examine each contributions, we choose the following set of state variables:

\[ \underline{\varepsilon}^{\mathrm{el}}, \underline{\varepsilon}^{\text{ibc}}, \left\|\underline{\varepsilon}_{m}^{\text{ibc}}\right\| \]

and the following set of auxiliary state variables:

\[ \underline{\varepsilon}^{\text{shr}}, \underline{\varepsilon}^{\text{rbc}}, \underline{\varepsilon}^{\text{dc}} \]

Split of the strain in spherical and deviatoric parts

In the proposed implementation, each inelastic strain will be split into its spherical and deviatoric part for the sake of clarity. With this choice, the final set of state variables is:

\[ \underline{\varepsilon}^{\mathrm{el}}, \varepsilon_{s}^{\text{ibc}}, \underline{\varepsilon}_{d}^{\text{ibc}}, \left\|\underline{\varepsilon}_{m}^{\text{ibc}}\right\| \]

and the following set of auxiliary state variables:

\[ \underline{\varepsilon}^{\text{shr}}, \varepsilon_{s}^{\text{rbc}}, \underline{\varepsilon}_{d}^{\text{rbc}}, \underline{\varepsilon}^{\text{dc}} \]

Drawbacks of this choice are an increase the size of the implicit system and a higher memory consumption (i.e. the number of values saved from one time step to the other).

This choice can easily be reverted, leading to a more compact and efficient implementation, while a bit less readable.

Update of state variables and auxiliary state variables

State variables are automatically updated when the solutions (the increment of the state variables) of the implicit equations are found. Auxiliary state variables are generally updated in a dedicated code block called @UpdateAuxiliaryStateVariables.

Reducing the memory consumption

For this behaviour, declaring the auxiliary state variables is convenient for post-processing, i.e. the behaviour integration only requires to determine their increments during the time step and does not require to know their values at the beginning of the time step. If such post-processing does not matter, a significant amount of memory can be saved by removing their declaration as auxiliary state variables and their update in the @UpdateAuxiliaryStateVariable code block.

Computation of the stress

The stress tensor is a function of the increment of the elastic strain \(\underline{\varepsilon}^{\mathrm{el}}\). Its value at the middle of the time step (at \(t+\theta\,\Delta\,t\)) is given by:

\[ \left\{ \begin{aligned} {\left.p\right|_{t+\theta\,\Delta\,t}} &= {\left(3\,{\left.\lambda^{\text{el}}\right|_{t+\theta\,\Delta\,t}}+2\,{\left.\mu^{\text{el}}\right|_{t+\theta\,\Delta\,t}}\right)}\,{\left.\varepsilon^{\text{el}}_{s}\right|_{t+\theta\,\Delta\,t}} &= 3\,{\left.k^{\text{el}}\right|_{t+\theta\,\Delta\,t}}\,{\left.\varepsilon^{\text{el}}_{s}\right|_{t+\theta\,\Delta\,t}} \\ {\left.\underline{\sigma}_d\right|_{t+\theta\,\Delta\,t}} &= 2\,{\left.\mu^{\text{el}}\right|_{t+\theta\,\Delta\,t}}\,{\left.\underline{\varepsilon}^{\mathrm{el}}_{d}\right|_{t+\theta\,\Delta\,t}} \end{aligned} \right. \qquad{(9)}\]

The derivatives of the pressure and the deviatoric stress with regard to the elastic strain increment are thus:

\[ {\displaystyle \frac{\displaystyle \partial {\left.p\right|_{t+\theta\,\Delta\,t}}}{\displaystyle \partial \Delta\,\underline{\varepsilon}^{\mathrm{el}}}} = 3\,\theta\,{\left.k^{\text{el}}\right|_{t+\theta\,\Delta\,t}}\,\underline{I} \quad\text{and}\quad {\displaystyle \frac{\displaystyle \partial {\left.\underline{\sigma}_d\right|_{t+\theta\,\Delta\,t}}}{\displaystyle \partial \Delta\,\underline{\varepsilon}^{\mathrm{el}}}} = 2\,\theta\,{\left.\mu^{\text{el}}\right|_{t+\theta\,\Delta\,t}}\,\underline{\underline{\mathbf{K}}} \qquad{(10)}\]

Implicit equation associated with basic creep

Reversible part: discretization of Equation (4)

Equation (4) can be discretized as follows:

\[ \begin{aligned} &&k^{\text{rs}}\,{\left.\varepsilon_{s}^{\text{rbc}}\right|_{t+\theta\,\Delta\,t}} + \eta^{\text{rs}}\,{{\displaystyle \frac{\displaystyle \Delta\,\varepsilon_{s}^{\text{rbc}}}{\displaystyle \Delta\,t}}}&={\left.h\right|_{t+\theta\,\Delta\,t}}\,{\left.p\right|_{t+\theta\,\Delta\,t}}\\ &\Rightarrow&{\left(\eta^{\text{rs}}+\theta\,\Delta\,t\,k^{\text{rs}}\right)}\,\Delta\,\varepsilon_{s}^{\text{rbc}}&=\Delta\,t\,{\left.h\right|_{t+\theta\,\Delta\,t}}\,{\left.p\right|_{t+\theta\,\Delta\,t}}-\Delta\,t\,k^{\text{rs}}\,{\left.\varepsilon_{s}^{\text{rbc}}\right|_{t}} \end{aligned} \]

Finally, the expression of the increment of the reversible part of the basic creep is:

\[ \Delta\,\varepsilon_{s}^{\text{rbc}} = \Delta\,t\,{{\displaystyle \frac{\displaystyle {\left.h\right|_{t+\theta\,\Delta\,t}}}{\displaystyle \eta^{\text{rs}}+\theta\,\Delta\,t\,k^{\text{rs}}}}}\,{\left.p\right|_{t+\theta\,\Delta\,t}}- \Delta\,t\,{{\displaystyle \frac{\displaystyle k^{\text{rs}}}{\displaystyle \eta^{\text{rs}}+\theta\,\Delta\,t\,k^{\text{rs}}}}}\,{\left.\varepsilon_{s}^{\text{rbc}}\right|_{t}} \qquad{(11)}\]

This equation shows that \(\Delta\,\varepsilon_{s}^{\text{rbc}}\) is a simple linear function of the pressure, i.e. a simple function of the elastic strain increment.

The derivative of \(\Delta\,\varepsilon_{s}^{\text{rbc}}\) with respect to the elastic strain increment is simply:

\[ {\displaystyle \frac{\displaystyle \partial \Delta\,\varepsilon_{s}^{\text{rbc}}}{\displaystyle \partial \Delta\,\underline{\varepsilon}^{\mathrm{el}}}}= \Delta\,t\,{{\displaystyle \frac{\displaystyle {\left.h\right|_{t+\theta\,\Delta\,t}}}{\displaystyle \eta^{\text{rs}}+\theta\,\Delta\,t\,k^{\text{rs}}}}}\,{\displaystyle \frac{\displaystyle \partial {\left.p\right|_{t+\theta\,\Delta\,t}}}{\displaystyle \partial \Delta\,\underline{\varepsilon}^{\mathrm{el}}}} \]

where the derivative \({\displaystyle \frac{\displaystyle \partial {\left.p\right|_{t+\theta\,\Delta\,t}}}{\displaystyle \partial \Delta\,\underline{\varepsilon}^{\mathrm{el}}}}\) is given by Equation (10).

Irreversible part: discretization of Equation (5)

Equation (5) can be discretized as follows:

\[ \Delta\,\varepsilon_{s}^{\text{ibc}} = \Delta\,t\,{{\displaystyle \frac{\displaystyle {\left.h\right|_{t+\theta\,\Delta\,t}}}{\displaystyle \eta^{\text{is}}{\left({\left.\left\|\underline{\varepsilon}_{m}^{\text{ibc}}\right\|\right|_{t+\theta\,\Delta\,t}}\right)}}}}\,{\left.p\right|_{t+\theta\,\Delta\,t}} \]

The implicit equation \(f_{\Delta\,\varepsilon_{s}^{\text{ibc}}}\) associated with \(\Delta\,\varepsilon_{s}^{\text{ibc}}\) is thus:

\[ f_{\Delta\,\varepsilon_{s}^{\text{ibc}}} = \Delta\,\varepsilon_{s}^{\text{ibc}}- \Delta\,t\,{{\displaystyle \frac{\displaystyle {\left.h\right|_{t+\theta\,\Delta\,t}}}{\displaystyle \eta^{\text{is}}{\left({\left.\left\|\underline{\varepsilon}_{m}^{\text{ibc}}\right\|\right|_{t+\theta\,\Delta\,t}}\right)}}}}\,{\left.p\right|_{t+\theta\,\Delta\,t}} \]

The following derivatives will be useful:

\[ \left\{ \begin{aligned} {\displaystyle \frac{\displaystyle \partial f_{\Delta\,\varepsilon_{s}^{\text{ibc}}}}{\displaystyle \partial \Delta\,\varepsilon_{s}^{\text{ibc}}}} &= 1 \\ {\displaystyle \frac{\displaystyle \partial f_{\Delta\,\varepsilon_{s}^{\text{ibc}}}}{\displaystyle \partial \Delta\,\underline{\varepsilon}^{\mathrm{el}}}} &= - \Delta\,t\,{{\displaystyle \frac{\displaystyle {\left.h\right|_{t+\theta\,\Delta\,t}}}{\displaystyle \eta^{\text{is}}{\left({\left.\left\|\underline{\varepsilon}_{m}^{\text{ibc}}\right\|\right|_{t+\theta\,\Delta\,t}}\right)}}}}\,{\displaystyle \frac{\displaystyle \partial {\left.p\right|_{t+\theta\,\Delta\,t}}}{\displaystyle \partial \underline{\varepsilon}^{\mathrm{el}}}} \\ {\displaystyle \frac{\displaystyle \partial f_{\Delta\,\varepsilon_{s}^{\text{ibc}}}}{\displaystyle \partial \Delta\,{\left.\left\|\underline{\varepsilon}_{m}^{\text{ibc}}\right\|\right|_{t+\theta\,\Delta\,t}}}} &= \Delta\,t\,{{\displaystyle \frac{\displaystyle {\left.h\right|_{t+\theta\,\Delta\,t}}}{\displaystyle {\left(\eta^{\text{is}}{\left({\left.\left\|\underline{\varepsilon}_{m}^{\text{ibc}}\right\|\right|_{t+\theta\,\Delta\,t}}\right)}\right)}^{2}}}}\,{\left.p\right|_{t+\theta\,\Delta\,t}}\, {\displaystyle \frac{\displaystyle \partial \eta^{\text{is}}}{\displaystyle \partial \Delta\,\left\|\underline{\varepsilon}_{m}^{\text{ibc}}\right\|}} \\ \end{aligned} \right. \]

Implicit equation associated with the maximum value of the norm of the irreversible part of the basic creep

In this document, we choose the following implicit equation to determine \(\Delta\,\left\|\underline{\varepsilon}_{m}^{\text{ibc}}\right\|\):

\[ f_{\left\|\underline{\varepsilon}_{m}^{\text{ibc}}\right\|}= \left\{ \begin{aligned} {\left.\left\|\underline{\varepsilon}_{m}^{\text{ibc}}\right\|\right|_{t}}+\Delta\,\left\|\underline{\varepsilon}_{m}^{\text{ibc}}\right\|- \sqrt{{\left.\underline{\varepsilon}^{\text{ibc}}\right|_{t+\Delta\,t}}\,\colon\,{\left.\underline{\varepsilon}^{\text{ibc}}\right|_{t+\Delta\,t}}} &\quad\text{if}\quad \sqrt{{\left.\underline{\varepsilon}^{\text{ibc}}\right|_{t+\Delta\,t}}\,\colon\,{\left.\underline{\varepsilon}^{\text{ibc}}\right|_{t+\Delta\,t}}}>{\left.\left\|\underline{\varepsilon}_{m}^{\text{ibc}}\right\|\right|_{t}}&\\ \Delta\,\left\|\underline{\varepsilon}_{m}^{\text{ibc}}\right\|&\quad\text{otherwise}& \end{aligned} \right. \qquad{(12)}\]

The derivatives of this equation are:

\[ \left\{ \begin{aligned} {\displaystyle \frac{\displaystyle \partial f_{\left\|\underline{\varepsilon}_{m}^{\text{ibc}}\right\|}}{\displaystyle \partial \Delta\,\left\|\underline{\varepsilon}_{m}^{\text{ibc}}\right\|}} &= 1\\ {\displaystyle \frac{\displaystyle \partial f_{\left\|\underline{\varepsilon}_{m}^{\text{ibc}}\right\|}}{\displaystyle \partial \Delta\,\underline{\varepsilon}^{\text{ibc}}}} &= -{{\displaystyle \frac{\displaystyle \Delta\,\underline{\varepsilon}^{\text{ibc}}}{\displaystyle \sqrt{{\left.\underline{\varepsilon}^{\text{ibc}}\right|_{t+\Delta\,t}}\,\colon\,{\left.\underline{\varepsilon}^{\text{ibc}}\right|_{t+\Delta\,t}}}}}}\quad\text{if}\quad \sqrt{{\left.\underline{\varepsilon}^{\text{ibc}}\right|_{t+\Delta\,t}}\,\colon\,{\left.\underline{\varepsilon}^{\text{ibc}}\right|_{t+\Delta\,t}}}>{\left.\left\|\underline{\varepsilon}_{m}^{\text{ibc}}\right\|\right|_{t}}\\ \end{aligned} \right. \]

An alternative implicit equation to determine the maximum value of the norm of the irreversible part of the basic creep

Using Implicit Equation (12) may lead to spurious oscillations of the Newton algorithm, although this is not observered in the test cases below.

This is due to the fact that the derivatives of this equation are not continuous.

A standard way to overcome this issue is to use a status method which can be implemented in MFront [12].

In this paragraph, we introduce an alternative treatment of this equation based on the regularised Fisher-Burmeister complementary function \(\Phi_{\varepsilon_{\text{FB}}}\), introduced in [13] and whose application to plasticity is described on a dedicated web page [14]:

\[ \Phi_{\varepsilon}{\left(x,y\right)}=x+y-\sqrt{x^{2}+y^{2}+\varepsilon_{\text{FB}}^{2}} \]

where \(\varepsilon_{\text{FB}}\) denotes a small numerical parameter.

\(\Phi_{\varepsilon}{\left(x,y\right)}\) has the following property:

\[ \Phi_{\varepsilon}{\left(x,y\right)}=0 \quad\Leftrightarrow\quad \left\{ \begin{aligned} x&\geq 0\\ y&\geq 0\\ 2\,x\,y&=\varepsilon^{2} \end{aligned} \right. \qquad{(13)}\]

Based on Property (13), the Implicit Equation (12) can be replaced by the following equation:

\[ f_{\left\|\underline{\varepsilon}_{m}^{\text{ibc}}\right\|}=\Phi_{\varepsilon_{\text{FB}}}{\left(\Delta\,\left\|\underline{\varepsilon}_{m}^{\text{ibc}}\right\|, {\left.\left\|\underline{\varepsilon}_{m}^{\text{ibc}}\right\|\right|_{t}}+\Delta\,\left\|\underline{\varepsilon}_{m}^{\text{ibc}}\right\|- \sqrt{{\left.\underline{\varepsilon}^{\text{ibc}}\right|_{t+\Delta\,t}}\,\colon\,{\left.\underline{\varepsilon}^{\text{ibc}}\right|_{t+\Delta\,t}}}\right)} \qquad{(14)}\]

Implicit Equation (14) has continuous derivatives.

\(\Phi_{\varepsilon}\) and its derivatives are implemented in the TFEL/Math library in Version 3.2 but their usage requires to include the TFEL/Math/FischerBurmeister.hxx header:

Implicit equation associated with the evolution of the drying creep strain

Equation (8) can be discretized as follows:

\[ \Delta\,\underline{\varepsilon}^{\text{dc}} = \frac{\Delta\,t}{\eta_{\text{fd}}}\left\langle \dot{h} \right\rangle\,{\left.\underline{\sigma}\right|_{t+\theta\,\Delta\,t}} \]

\(\Delta\,\underline{\varepsilon}^{\text{dc}}\) is thus a simple function of the elastic strain increment \(\Delta\,\underline{\varepsilon}^{\mathrm{el}}\).

Its derivative with respect to the elastic strain increment is:

\[ {\displaystyle \frac{\displaystyle \partial \Delta\,\underline{\varepsilon}^{\text{dc}}}{\displaystyle \partial \Delta\,\underline{\varepsilon}^{\mathrm{el}}}} =\frac{\Delta\,t}{\eta_{\text{fd}}}\left\langle \dot{h} \right\rangle\,{\displaystyle \frac{\displaystyle \partial {\left.\underline{\sigma}\right|_{t+\theta\,\Delta\,t}}}{\displaystyle \partial \Delta\,\underline{\varepsilon}^{\mathrm{el}}}} =\frac{\Delta\,t}{\eta_{\text{fd}}}\left\langle \dot{h} \right\rangle\, {\left(\underline{I}\,\otimes\,{\displaystyle \frac{\displaystyle \partial {\left.p\right|_{t+\theta\,\Delta\,t}}}{\displaystyle \partial \Delta\,\underline{\varepsilon}^{\mathrm{el}}}}+ {\displaystyle \frac{\displaystyle \partial {\left.\underline{\sigma}_{d}\right|_{t+\theta\,\Delta\,t}}}{\displaystyle \partial \Delta\,\underline{\varepsilon}^{\mathrm{el}}}}\right)} \]

Implicit equation associated with the elastic strain

The implicit equation \(f_{\underline{\varepsilon}^{\mathrm{el}}}\) associated with the elastic strain increment is given by the split of the strain (1) written in incremental form:

\[ f_{\underline{\varepsilon}^{\mathrm{el}}}=\Delta\,\underline{\varepsilon}^{\mathrm{el}}-\Delta\,\underline{\varepsilon}^{\mathrm{to}} +\Delta\,\underline{\varepsilon}^{\text{shr}} +\Delta\,\underline{\varepsilon}^{\text{bc}} +\Delta\,\underline{\varepsilon}^{\text{dc}} \qquad{(15)}\]

Description of the MFront implementations

Choice of the domain specific language and the StandardElasticity brick

The domain specific language (DSL) suitable for implementing an implicit scheme based on standard strain-based beahviour is called Implicit [11]. The @DSL keyword allows specifying the domain specific language to be used:

@DSL Implicit;

The standard elasticity brick [15] is appropriate for this behaviour which provides additional functionnalities such as:

The usage of the StandardElasticity is introduced as follows:

@Brick StandardElasticity;

Automic declaration of the elastic strain

The Implicit DSL automatically defines the elastic strain as the first state variable.

Description of the steps of the implicit scheme and the associated code blocks.
Figure 1: Description of the steps of the implicit scheme and the associated code blocks.

The main steps of the behaviour integration used by the Implicit DSL are reported on Figure 1:

Name of the behaviour

The @Behaviour keyword is required to name the behaviour, as follows:

@Behaviour Burger_EDF_CIWAP_2021;

Material

The @Material keyword allows to associate the behaviour to material:

@Material Concrete;

Metadata

The @Author keyword gives the name of the person who implemented the behaviour:

@Author François Hamon (EDF R&D ERMES - T65),
        Jean - Luc Adia (EDF R&D MMC - T25),
        Thomas Helfer (CEA IRESNE/DES/DEC/SESC);

The @Description keyword allows to introduce a small description of the implementation:

@Description {
`Burger_EDF_CIWAP_2021` model for concrete creep;
Compater to `code_aster`' `BETON_BURGER` constitutive equations, this behaviour
takes into account two points: 
-adding of shrinkage depending to relative humidity;
-drying creep depending to the positive part of derivative of relative humidity;
}

Choice of the algorithm and numerical parameters

For this implementation, we will use the standard Newton-Raphson:

@Algorithm NewtonRaphson;

The @Epsilon keyword allows to specify the default value of the stopping criteria:

@Epsilon 1e-14;

This default value can be changed at runtime using the epsilon parameter.

The @Theta keyword allows the specify the default value of the \(\theta\) parameter of the implicit scheme:

@Theta 1;

This default value can be changed at runtime using the theta parameter.

External headers

If we want to use the alternative implicit equation for \(\Delta\,\left\|\underline{\varepsilon}_{m}^{\text{ibc}}\right\|\), discussed in Section 2.3.4, we need to include the following header file using the @Includes keyword:

@Includes{
#include "TFEL/Math/FischerBurmeister.hxx"
}

Variables declarations

Material properties

Material properties are a classical way to create behaviours that can be used to describe a wide range of materials. Material properties are provided by the calling solver and are generally read in the input file.

Table 1: Name of the material properties and descriptions
MFront variable Description
young Young modulus \(E^{\text{el}}\)
nu Poisson ratio \(\nu^{\text{el}}\)
K_SH Shrinkage coefficient \(k^{\text{shr}}\)
K_RS Spherical reversible basic creep stiffness \(k_{0}^{\text{rs}}\)
ETA_RS Spherical reversible basic creep viscosity \(\eta_{0}^{\text{rs}}\)
KAPPA Reference strain for evolution of irreversible creep viscosity \(\kappa_{0}\)
ETA_IS Spherical irreversible creep viscosity \(\eta_{0}^{\text{is}}\)
K_RD Deviatoric reversible basic creep stiffness \(k_{0}^{\text{rd}}\)
ETA_RD Deviatoric reversible basic creep viscosity \(\eta_{0}^{\text{rd}}\)
ETA_ID Deviatoric irreversible creep viscosity \(\eta_{0}^{\text{id}}\)
QSR_K Activation energy \(\frac{Q}{R}\)
TEMP_0_C Reference temperature \(T_{0}\)
ETA_FD Drying creep coefficient \(\eta_{\text{fd}}\)

The material properties required by the Burger_EDF_CIWAP_2021 behaviour are listed in Table 1.

Material properties are declared using the @MaterialProperty keyword, as follows:

//! Elastic Young Modulus
@MaterialProperty stress young;
young.setGlossaryName("YoungModulus");
//! Elastic Poisson Ratio
@MaterialProperty real nu;
nu.setGlossaryName("PoissonRatio");
//! Shrinkage coefficient
@MaterialProperty real K_SH;
K_SH.setEntryName("ShrinkageFactor");
//! Spherical reversible basic creep stiffness
@MaterialProperty real K_RS;
K_RS.setEntryName("SphericReversibleStiffness");
//! Spherical reversible basic creep viscosity
@MaterialProperty real ETA_RS;
ETA_RS.setEntryName("SphericReversibleViscosity");
//! Reference strain for evolution of irreversible creep viscosity
@MaterialProperty real KAPPA;
KAPPA.setEntryName("IrreversibleCreepViscosityReferenceStrain");
//! Spherical irreversible creep viscosity
@MaterialProperty real ETA_IS;
ETA_IS.setEntryName("SphericIrreversibleCreepViscosity");
//! Deviatoric reversible basic creep stiffness
@MaterialProperty real K_RD;
K_RD.setEntryName("DeviatoricReversibleStiffness");
//! Deviatoric reversible basic creep viscosity
@MaterialProperty real ETA_RD;
ETA_RD.setEntryName("DeviatoricReversibleViscosity");
//! Deviatoric irreversible creep viscosity
@MaterialProperty real ETA_ID;
ETA_ID.setEntryName("DeviatoricIrreversibleViscosity");
//! Activation energy of basic creep parameter
@MaterialProperty real QSR_K;
QSR_K.setEntryName("ActivationEnergy");
//! Reference temperature for temperature activation of basic creep
@MaterialProperty real TEMP_0_C;
TEMP_0_C.setEntryName("ReferenceTemperature");
//! Drying creep viscosity
@MaterialProperty real ETA_FD;
ETA_FD.setEntryName("DryingCreepVicosity");

Note that we associated to each material property an external name (either a glossary name or a so-called entry name) using the setGlossaryName or setEntryName methods. The external name is the name of this variable as seen by the calling solver and its user when this solver supports deep integration of MFront behaviours.

Glossary names are a restricted list of names described on a dedicated page [16]. The setGlossaryName fails if the given name is not in this list, avoiding typing errors.

Material properties vs parameters

In most MFront tutorials, we usually prefer to use parameters, which are declared using the @Parameter keyword.

Contrary to material properties, parameters have default values. Parameters are thus useful to build behaviours specific to one particular material. Such behaviours are self-contained, which considerably eases building a strict material knowledge management policy.

Parameters are also stored in a global structure. This means that parameters are uniform. In some codes, material properties may be defined on a per integration point basis.

State variables

Table 2: Name of the state variables and descriptions
MFront variable Description
ELIM Limit strain for the irreversible creep model
ESPHI Spherical irreversible basic reep strain \(\varepsilon_{s}^{\text{ibc}}\)
EDEVI Deviatoric irreversible basic creep strain \(\varepsilon_{d}^{\text{ibc}}\)

Following Section 2.1, the state variables required by the Burger_EDF_CIWAP_2021 behaviour are listed in Table 2.

The state variables are introduced by the @StateVariable as follows:

@StateVariable strain ELIM;
ELIM.setEntryName("MaximumValueOfTheIrreversibleStrain");
//! Spherical irreversible basic creep strain
@StateVariable strain ESPHI;
ESPHI.setEntryName("SphericIrreversibleStrain");
//! Deviatoric irreversible basic creep strain
@StateVariable StrainStensor EDEVI;
EDEVI.setEntryName("DeviatoricIrreversibleStrain");

Auxiliary state variables

Table 3: Name of the auxiliary state variables and descriptions
MFront variable Description
ESH Shrinkage strain \(\varepsilon^{\text{shr}}\)
ESPHR Spherical reversible basic creep strain \(\varepsilon_{s}^{\text{rbc}}\)
EDEVR Deviatoric reversible basic creep strain \(\varepsilon_{d}^{\text{rbc}}\)
Edess Drying creep strain \(\varepsilon^{\text{dc}}\)
rHmin Historical min value of relative humidity
EF Basic creep strain tensor

Following Section 2.1, the auxiliary state variables required by the Burger_EDF_CIWAP_2021 behaviour are listed in Table 3.

The auxiliary state variables are introduced by the @AuxiliaryStateVariable as follows:

//! Spherical Reversible basic creep strain
@AuxiliaryStateVariable strain ESPHR;
ESPHR.setEntryName("SphericReversibleStrain");
//! Deviatoric Reversible basic creep strain
@AuxiliaryStateVariable StrainStensor EDEVR;
EDEVR.setEntryName("DeviatoricReversibleStrain");
//! Shrinkage strain
@AuxiliaryStateVariable strain ESH;
ESH.setEntryName("ShrinkageStrain");
//! Drying creep strain
@AuxiliaryStateVariable StrainStensor Edess;
Edess.setEntryName("DryingCreepStrain");
/*!
 * Shifted value of the historical minimale of relative humidity 
 *
 * This value has been shifted (value + 1) so the implicit initialization
 * to zero of this variable in most finite element solvers is meaningfull.
 * So for initialization at value different to 1, please don't forget to
 * take into account this shift.
 */
@AuxiliaryStateVariable strain rHmin;
rHmin.setEntryName("ShiftedHistoricalMinimumRelativeHumidity");
//! Basic creep strain tensor
@AuxiliaryStateVariable StrainStensor EF;
EF.setEntryName("BasicCreepStrain");

External state variables

Table 4: Name of the external state variables and descriptions
MFront variable Description
T Temperature
rH Relative humidity

The external state variables required by the behaviour are listed on Table 4.

The temperature is automatically defined by MFront, hence only the relative humidity must be declared:

@ExternalStateVariable real rH;
rH.setEntryName("RelativeHumidity");

Note for code_aster users

Due to the specific way used by code_aster to handle external state variables, the external name of the relative humidity must be changed to SECH as follows:

rH.setEntryName("SECH");

Local variables

Local variables are a powerfull feature of MFront which allows to define variables which will be accessible in each code blocks.

In this implementation, we will need local variables for two distinct purposes:

  1. Precompute some quantities before the resolution of the implicit system.
  2. Save the increment of auxiliary state variables for their update after the resolution of the implicit system.

Local variables to optimize computation times

A few variables can be computed before the resolution of the implicit system in the @InitLocalVariables code block. The most important ones are coefficients of creep mechanisms whose computations involves exponentials of the temperature following the Arrhenius law (see Equations (6) and (7)).

Those local variables are declared as follows:

//! first Lamé coeffient
@LocalVariable real lambda;
//! second Lamé coeffient
@LocalVariable real mu;
//! variable to compute effect of relative on drying creep
@LocalVariable real VrH;
//! inverse of drying strain viscosity
@LocalVariable real inv_ETA_FD;
/* variables for impact of temperature on basic creep */
@LocalVariable real KRS_T;
@LocalVariable real KRD_T;
@LocalVariable real NRS_T;
@LocalVariable real NRD_T;
@LocalVariable real NIS_T;
@LocalVariable real NID_T;
@LocalVariable real KAPPA_T;

Local variables saving the increments of the auxiliary state variables

As discussed in Section 2, the increment of auxiliary state variables can be expressed as a function of the increments of the state variables. Those increments must be evaluated during the evaluation of the implicit system in the @Integrator code block for each new estimates of the increments of the state variables.

Using local variables allows to save the last estimates of the increment of the auxiliary state variables to update those variables in the @UpdateAuxiliaryStateVariables code blocks.

Those local variables are declared as follows:

//! increment of spherical reversible strain
@LocalVariable strain dESPHR;
//! increment of deviatoric reversible strain
@LocalVariable StrainStensor dEDEVR;
//! increment of shrinkage strain
@LocalVariable strain dESH;
//! increment of drying creep strain
@LocalVariable StrainStensor dEdess;

Initialisation of the local variables

The @InitLocalVariables code block is now described.

@InitLocalVariables {

First we explicitly compute the Lamé coefficients using the built-in computeLambda and computeMu functions:

  lambda = computeLambda(young, nu);
  mu = computeMu(young, nu);

Then, we compute the effective creep properties for the temperature at the middle of the time step:

  const auto Tm = T + dT / 2;
  const auto iT = 1 / (273 + Tm) - 1 / (273 + TEMP_0_C);
  const auto e = exp(QSR_K * iT);
  KRS_T = K_RS * e;
  KRD_T = K_RD * e;
  NRS_T = ETA_RS * e;
  NRD_T = ETA_RD * e;
  NIS_T = ETA_IS * e;
  NID_T = ETA_ID * e;
  KAPPA_T = KAPPA / e;

Then we compute the value of the shrinkage strain increment \(\Delta\,\epsilon^{\text{shr}}\):

  dESH = K_SH * drH;

Finally, we compute the quantities relative the the drying strain:

  if ((drH <= 0) && (rH <= rHmin + 1)) {
    VrH = fabs(drH);
    rHmin = rH - 1;
  } else {
    VrH = 0;
  }
  if (ETA_FD > real(0)) {
    inv_ETA_FD = 1 / ETA_FD;
  } else {
    inv_ETA_FD = real(0);
  }
}

Note that this implementation uses a little trick to activate the drying strain base on the value of \(\eta_{\text{fd}}\). In the latter is negative or null, drying creep is desactivated.

Implicit system

The @Integrator code block defines the values of residuals and their derivatives with respect to the state variables.

@Integrator {

We start this implementation by defining the current estimates of the pressure and deviatoric stress at the middle of the time steps and their derivatives with respect to the elastic strain increment:

  constexpr auto eeps = strain(1e-14);
  const auto id = Stensor::Id();
  const auto pr = trace(sig) / 3;
  const auto s = deviator(sig);
  const auto dpr_ddeel = (lambda + 2 * mu / 3) * theta * id;
  const auto ds_ddeel = 2 * mu * theta * Stensor4::K();
  const auto dsig_ddeel = ds_ddeel + (dpr_ddeel ^ id);

The following line computes the relative humidity at the middle of the time step:

  const auto rH_mts = rH + theta * drH;

The computation of the increment of the spherical and deviatoric parts of the basic creep given by Equation (11) is implemented as follows:

  const auto a_rs = theta * KRS_T * dt / NRS_T;
  dESPHR = (rH_mts * pr - KRS_T * ESPHR) * dt / (NRS_T * (1 + a_rs));
  const auto ddESPHR_dpr = rH_mts * dt / (NRS_T * (1 + a_rs));
  const auto a_rd = theta * KRD_T * dt / NRD_T;
  dEDEVR = (rH_mts * s - KRD_T * EDEVR) * dt / (NRD_T * (1 + a_rd));
  const auto ddEDEVR_dds = rH_mts * dt / (NRD_T * (1 + a_rd));

The implementation of Implicit Equation (12) is straightforward:

  const auto e = (ESPHI + dESPHI) * id + EDEVI + dEDEVI;
  const auto ne = sqrt(e | e);
  if (ne > ELIM) {
    fELIM += ELIM - ne;
    const auto ine = 1 / max(ne, strain(1.e-14));
    dfELIM_ddESPHI = -trace(e) * ine;
    dfELIM_ddEDEVI = -e * ine;
  }

Implementation of Implicit Equation (14)

As discussed in Section 2.3.4, an alternative implementation based on the Fisher-Burmeister complementary function may is possible. This implementation is as follows:

  const auto e = (ESPHI + dESPHI) * id + EDEVI + dEDEVI;
  const auto ne = sqrt(e | e);
  const auto ine = 1 / max(ne, eeps);
  const auto dne_ddESPHI = trace(e) * ine;
  const auto dne_ddEDEVI = e * ine;
  const auto myield_ELIM = ELIM + dELIM - ne;
  fELIM = regularisedFischerBurmeisterFunction(dELIM, myield_ELIM, eeps);
  const auto [dfELIM_dx, dfELIM_dy] =
      regularisedFischerBurmeisterFunctionFirstDerivatives(dELIM, myield_ELIM,
                                                           eeps);
  dfELIM_ddELIM = dfELIM_dx + dfELIM_dy;
  dfELIM_ddESPHI = -dfELIM_dy * dne_ddESPHI;
  dfELIM_ddEDEVI = -dfELIM_dy * dne_ddEDEVI;

We know implement the implicit equations associated with the spherical and deviatoric parts of the irreversible part of the basic creep:

  elim = (ELIM + theta * dELIM) / KAPPA_T;
  auto delim_ddELIM = theta / KAPPA_T;
  if (abs(elim) > 200) {
    elim = (elim / abs(elim)) * 200;
    delim_ddELIM = 0;
  }
  const auto eexp = exp(-elim);
  const auto deexp_ddELIM = -eexp * delim_ddELIM;
  fESPHI -= (rH_mts * eexp * dt / NIS_T) * pr;
  dfESPHI_ddeel = -(rH_mts * eexp * dt / NIS_T) * dpr_ddeel;
  dfESPHI_ddELIM = -(rH_mts * deexp_ddELIM * dt / NIS_T) * pr;
  fEDEVI -= (rH_mts * eexp * dt / NID_T) * s;
  dfEDEVI_ddeel = -(rH_mts * eexp * dt / NID_T) * ds_ddeel;
  dfEDEVI_ddELIM = -(rH_mts * deexp_ddELIM * dt / NID_T) * s;

The computation of the increment of the drying creep strain and its derivative with respect to the elastic strain increment is straightforward:

  dEdess = inv_ETA_FD * VrH * sig;
  const auto dEdess_ddeel = inv_ETA_FD * VrH * dsig_ddeel;

Finally, the residual \(f_{\underline{\varepsilon}^{\mathrm{el}}}\) associated with the elastic strain is given by Equation (15) which can be implemented in a straightforward manner:

  feel += dEDEVR + dEDEVI + dEdess + (dESPHR + dESPHI + dESH) * id;
  dfeel_ddeel += ddEDEVR_dds * ds_ddeel + dEdess_ddeel +  //
                 (id ^ (ddESPHR_dpr * dpr_ddeel));
  dfeel_ddEDEVI = Stensor4::Id();
  dfeel_ddESPHI = id;
}

Update of the auxiliary state variables

Following Section 2.1, the @UpdateAuxiliaryStateVariables code block is used to update the auxiliary state variables:

@UpdateAuxiliaryStateVariables {
  ESH += dESH;
  Edess += dEdess;
  ESPHR += dESPHR;
  EDEVR += dEDEVR;
  //
  const auto id = Stensor::Id();
  EF += dEDEVR + dEDEVI + (dESPHR + dESPHI) * id;
}

Test cases

MTest non regression test cases

5 MTest’ unit tests associated with the Burger_EDF_CIWAP_2021 behaviour have been introduced in the MFrontGallery project. Thoses tests are automatically executed when MFront evolves as part of the its continuous integration strategy.

This ensures that the implementations of the Burger_EDF_CIWAP_2021 behaviour compiles despite MFront’ evolutions and that the MTest’ tests gives the same results.

It is also worth mentionning that each test is executed \(5\) times which differents roundings modes (including a so-called random rounding mode where the rounding mode is changed a various predefined steps by MTest), providing a basic and naive check of the numerical stability of the results.

In the following test cases, the material properties given in the table below are used.

Values of the material properties used for the unit tests.
Material Properties Values Units
Young Modulus \(E^{\text{el}}\) 24.2e9 Pa
Poisson ratio \(\nu^{\text{el}}\) 0.2 -
Shrinkage coefficient \(k^{\text{shr}}\) 0.00951974 (and 0 for cases 4 and 5) -
Spherical reversible basic creep stiffness \(k_{0}^{\text{rs}}\) 3.9e10 Pa
Spherical reversible basic creep viscosity \(\eta_{0}^{\text{rs}}\) 4.6e17 Pa.s
Reference strain for evolution of irreversible creep viscosity \(\kappa_{0}\) 1.20e-4 -
Spherical irreversible creep viscosity \(\eta_{0}^{\text{is}}\) 2.6e18 Pa.s
Deviatoric reversible basic creep stiffness \(k_{0}^{\text{rd}}\) 1.95e10 Pa
Deviatoric reversible basic creep viscosity \(\eta_{0}^{\text{rd}}\) 2.30e17 Pa.s
Deviatoric irreversible creep viscosity \(\eta_{0}^{\text{id}}\) 1.30e18 Pa.s
Activation energy \(\frac{Q}{R}\) 7677.42 K
Reference temperature \(T_{0}\) 20 °C
Drying creep coefficient \(\eta_{\text{fd}}\) 6.2e9 Pa.s

Case 1: basic creep at 20°C

This test is available here.

The T-H-M loading is as follow:

@ExternalStateVariable "Temperature" 20.;
@ImposedStress "SXX" {0.0: 12e6, 8.64000000e+08: 12e6};
@ExternalStateVariable "SECH" {0.0: 1.0, 8.64000000e+08: 1.};
Results of the second test

Case 2: basic creep at 40°C

This test is available here.

The T-H-M loading is as follow:

@ExternalStateVariable 'Temperature' {0.0: 40.0, 8.64000000e+08: 40.0};
@ImposedStress "SXX" {0.0: 12e6, 8.64000000e+08: 12e6};
@ExternalStateVariable "SECH" {0.0: 1.0, 8.64000000e+08: 1.};
Results of the third test

Case 3: drying creep at 20°C and 75%RH

This test is available here.

The T-H-M loading is as follow:

@ExternalStateVariable "Temperature" 20.;
@ImposedStress "SXX" {0.0: 12e6, 8.64000000e+08: 12e6};
@ExternalStateVariable "SECH" {0.0: 1.0, 8.64000000e+08:0.5};
Results of the fourth test

Case 4: drying-humidification creep at 20°C and 50%RH

This test is available here.

The T-H-M loading is as follow:

@ExternalStateVariable "Temperature" 20.;
@ImposedStress "SXX" {0.0: 12e6, 8.64000000e+08: 12e6};
@ExternalStateVariable "SECH" {0.0: 0.5, 8.64000000e+04: 0.5,
9.60000000e+07: 0.5,
9.60864000e+07:0.8, 1.92000000e+08:0.8, 1.92086400e+08:0.5,
2.88000000e+08:0.5, 2.88086400e+08:0.8, 3.84000000e+08:0.8,
3.84086400e+08:0.5, 4.80000000e+08:0.5, 4.80086400e+08:0.8,
5.76000000e+08:0.8, 5.76086400e+08:0.5, 6.72000000e+08:0.5,
6.72086400e+08:0.8, 7.68000000e+08:0.8, 7.68086400e+08:0.5,
Results of the fourth test

Case 5: Shrinkage at 20°C

This test is available here.

The T-H-M loading is as follow:

@ExternalStateVariable "Temperature" 20.;
@ImposedStress "SXX" {0.0: 0, 8.64000000e+08: 0};
@ExternalStateVariable "SECH" {0.0: 1.0, 8.64000000e+08: 0.5};
Results of the first test

code_aster test cases:

The same material properties as previously are used. For test cases with drying, the GRANGER model with the parameters below is use to simulate drying before mechanical modelling with the Burger_EDF_CIWAP_2021 constitutive law.

GRANGER material Properties Values Units
\(A\) 4.3e-13 m2/s
\(B\) 0.07 -
\(\frac{Q}{R}\) 4679 K

A desorption isotherm model is use to pass from humidity to water content and inversely. This desorption isotherm is not described here but is in the computation files. Firstly, the drying problem is solved in term of water content. Then, this water content evolution is converted into a relative humidity field by using desorption isotherm model.

Profile of water concentration (left). Profile of relative humidity (right).

The water content of the three test cases below is presented here.

Evolution of the water content

Case 1: Basic creep at 20°C

This test case concerns the modelling of uniaxial basic creep test on 16X100 cm specimen with 12 MPa of compression as proposed in [8]. A comparison between BETON_BURGER and Burger_EDF_CIWAP_2021 is shown below.

Result of the first code_aster test

Case 1: Drying creep at 20°C and 60%RH

This test case concerns the modelling of uniaxial drying creep test on 16X100 cm specimen with 12 MPa of compression as proposed in [8]. A comparison between BETON_BURGER and Burger_EDF_CIWAP_2021 is shown below.

Result of the second code_aster test

Case 1: Drying creep at 20°C in sinusoidal relative humidity condition

This test case concerns the modelling of uniaxial drying creep test on 16X100 cm specimen with 12 MPa of compression as proposed in [8] in case of drying-humidification cycle condition. A comparison between BETON_BURGER and Burger_EDF_CIWAP_2021 is shown below.

Result of the third code_aster test

References

1.
2.
ACES H2020, LinkedIn showcase. Available from: https://www.linkedin.com/showcase/aces-h2020/
3.
Charpin, Laurent, Niepceron, Julien, Corbin, Manuel, Masson, Benoît, Mathieu, Jean-Philippe, Haelewyn, Jessica, Hamon, François, Åhs, Magnus, Aparicio, Sofía, Asali, Mehdi, Capra, Bruno, Azenha, Miguel, Bouhjiti, David E. -M., Calonius, Kim, Chu, Meng, Herrman, Nico, Huang, Xu, Jiménez, Sergio, Mazars, Jacky, Mosayan, Mahsa, Nahas, Georges, Stepan, Jan, Thenint, Thibaud and Torrenti, Jean-Michel. Ageing and air leakage assessment of a nuclear reactor containment mock-up: VERCORS 2nd benchmark. Nuclear Engineering and Design. 1 June 2021. Vol. 377, p. 111136. DOI 10.1016/j.nucengdes.2021.111136. Available from: https://www.sciencedirect.com/science/article/pii/S0029549321000881
4.
Benboudjema, Farid. Modélisation des déformations différées du béton sous sollicitations biaxiales. Application aux enceintes de confinement de bâtiments réacteurs des centrales nucléaires. Theses. Université de Marne la Vallée, 2002. Available from: https://tel.archives-ouvertes.fr/tel-00006945
5.
Benboudjema, F., Meftah, F. and Torrenti, J. M. Interaction between drying, shrinkage, creep and cracking phenomena in concrete. Engineering Structures. 1 January 2005. Vol. 27, no. 2, p. 239–250. DOI 10.1016/j.engstruct.2004.09.012. Available from: https://www.sciencedirect.com/science/article/pii/S0141029604003359
6.
EDF. Relation de comportement BETON_BURGER_FP pour le fluage du béton. code_aster website. March 2020. Available from: https://www.code-aster.org/V2/doc/v12/fr/man_r/r7/r7.01.35.pdf
7.
Sellier, Alain and Buffo-Lacarrière, Laurie. Vers une modélisation simple et unifiée du fluage propre, du retrait et du fluage en dessiccation du béton. European Journal of Environmental and Civil Engineering. 2009. Vol. 13, no. 10, p. 1161–1182. DOI 10.1080/19648189.2009.9693184. Available from: https://doi.org/10.1080/19648189.2009.9693184
8.
Boucher, Maxime. Analyse de la corrélation spatio-temporelle des déformations entre le coeur d’un ouvrage épais et son parement : Application aux enceintes de confinement. PhD thesis. 2016. Available from: http://www.theses.fr/2016GREAI067
9.
Bazant, Z. P. and Chern, J. C. Concrete creep at variable humidity: Constitutive law and mechanism. Materials and Structures. 1 January 1985. Vol. 18, no. 1, p. 1. DOI 10.1007/BF02473360. Available from: https://doi.org/10.1007/BF02473360
10.
Bouhjiti, D. E. -M., Boucher, M., Briffaut, M., Dufour, F., Baroth, J. and Masson, B. Accounting for realistic thermo-hydro-mechanical boundary conditions whilst modeling the ageing of concrete in nuclear containment buildings: Model validation and sensitivity analysis. Engineering Structures. 1 July 2018. Vol. 166, p. 314–338. DOI 10.1016/j.engstruct.2018.03.015. Available from: https://www.sciencedirect.com/science/article/pii/S0141029617316954
11.
CEA/EDF. The implicit domain specific language. 17 March 2021. Available from: http://tfel.sourceforge.net/implicit-dsl.html
12.
CEA/EDF. Implementation of a multi-surface, compressible and perfect plastic behaviour using the drucker-prager yield criterion and a cap. March 2021. Available from: http://tfel.sourceforge.net/drucker-prager-cap.html
13.
Liao-McPherson, Dominic, Huang, Mike and Kolmanovsky, Ilya. A regularized and smoothed fischer–burmeister method for quadratic programming with applications to model predictive control. IEEE Transactions on Automatic Control. July 2019. Vol. 64, no. 7, p. 2937–2944. DOI 10.1109/TAC.2018.2872201.
14.
CEA/EDF. On the fisher-burmeister complementary function and its applications to multi-surface plasticity. March 2021. Available from: http://tfel.sourceforge.net/implicit-dsl.html
15.
CEA/EDF. Behaviour bricks. March 2020. Available from: http://tfel.sourceforge.net/BehaviourBricks.html
16.
CEA/EDF. TFEL glossary description. March 2020. Available from: http://tfel.sourceforge.net/glossary.html