Thermally coupled Constitutive Laws

[AlejandroCornejo]

Description
Dear @KratosMultiphysics/structural-mechanics and @KratosMultiphysics/constitutive-models

I'm implementing all the code infraestructure in order to be able to simulate thermo-mechanical calculations with a solver that couples the Structural and the ConvectionDiffusion Apps.

The solver was already there long time ago, I've just improved it a bit and corrected a detail.

The setting of the REFERENCE_TEMPERATURE, different options are implemented: using the GetValue() in the Geometry (has to be set outside by someone) and being a material property as:

void ThermalLinearPlaneStrain::InitializeMaterial(
    const Properties& rMaterialProperties,
    const GeometryType& rElementGeometry,
    const Vector& rShapeFunctionsValues
    )
{
    if (rElementGeometry.Has(REFERENCE_TEMPERATURE)) {
        mReferenceTemperature = rElementGeometry.GetValue(REFERENCE_TEMPERATURE);
    } else if (rMaterialProperties.Has(REFERENCE_TEMPERATURE)) {
        mReferenceTemperature = rMaterialProperties[REFERENCE_TEMPERATURE];
    }
}

Changelog
Please summarize the changes in one list to generate the changelog:
E.g.

• Erased the convective term in the Eulerian calculation
• Creation of the new thermal CL inside the ConstitutiveLawsApp, only the elastic basic ones
• etc...

Documentation

Now, the materials.json for the structural problem looks as this:

{
    "properties" : [{
        "model_part_name" : "Structure.Parts_Solid_Solid_Auto1",
        "properties_id"   : 1,
        "Material"        : {
            "constitutive_law" : {
                "name" : "ThermalLinearPlaneStrain"
                
            },
            "Variables"        : {
                "DENSITY"       : 7850.0,
                "YOUNG_MODULUS" : 2.1e11,
                "POISSON_RATIO" : 0.29,
                "THERMAL_EXPANSION_COEFFICIENT" : 7.2e-6,
                "REFERENCE_TEMPERATURE" : 22.5
            },
            "Tables"           : {
                "TEMPERATURE_vs_E" : {
                        "input_variable"  : "TEMPERATURE",
                        "output_variable" : "YOUNG_MODULUS",
                        "data"            : [[-1,   2.1e11],
                                             [3500, 2.1e8 ],
                                             [1e6,  2.1e8 ]]
                    }
            },
        "accessors"        : {
            "accessor_table_T_E" : {
                "accessor_type"  : "table_accessor", // here we indicate the table accessor
                "properties" : {
                    "table_input_variable"      : "TEMPERATURE",
                    "table_output_variable"     : "YOUNG_MODULUS",
                    "table_input_variable_type" : "node_historical" // we support several types

				}
            }
        }
        }
    }]
}

[loumalouomega]

Se my comment of VariableUtils

[loumalouomega]

1. In order to reduce code repetition in the `thermal_constitutive_laws `I'm going to slightly modify the original CL. This change consists in subtituting the `rValues.GetProperties()[YOUNG_MODULUS]` by a new virtual method called `GetMaterialProperty(YOUNG_MODULUS)`, which in the original cl will do the same operation as before. In the case of the thermal laws, the only method to be modified would be the `GetMaterialProperty(rVariable)`, where we modifiy the property according to a certain `TEMPERATURE` Table.

This has a small overhead..., I am OK because there is not easy manner. Except maybe templated static auxiliary methods...

[AlejandroCornejo]

1. In order to reduce code repetition in the `thermal_constitutive_laws `I'm going to slightly modify the original CL. This change consists in subtituting the `rValues.GetProperties()[YOUNG_MODULUS]` by a new virtual method called `GetMaterialProperty(YOUNG_MODULUS)`, which in the original cl will do the same operation as before. In the case of the thermal laws, the only method to be modified would be the `GetMaterialProperty(rVariable)`, where we modifiy the property according to a certain `TEMPERATURE` Table.

This has a small overhead..., I am OK because there is not easy manner. Except maybe templated static auxiliary methods...

Te other option is to repeat all the code and leave alone the original CL... IDK if more templates.. hehe

[AlejandroCornejo]

ok guys! I've done quite an effort in avoiding code repetition. In this case I avoided Templates (sorry Vicente, nothing personal ;) ). Now the elastic 3D thermal implements everything and the thermal plane strain and stress derive from it overriding only 3 methods.

I've moved a previously repeated method to the utilities to reuse the same implementation.

[loumalouomega]

ok guys! I've done quite an effort in avoiding code repetition. In this case I avoided Templates (sorry Vicente, nothing personal ;) ). Now the elastic 3D thermal implements everything and the thermal plane strain and stress derive from it overriding only 3 methods.

I've moved a previously repeated method to the utilities to reuse the same implementation.

It works for me

[loumalouomega]

Removing my block

[AlejandroCornejo]

Let's see if @rubenzorrilla @RiccardoRossi @philbucher have something to say

[AlejandroCornejo]

ping

[loumalouomega]

ping

Pong

[rubenzorrilla]

The tests subsubfolder are still in there. Besides, the subfolder is at the level of the files of another test, something that is way misleading.

Please, either wrap the already existent test files in a new folder or simply put the thermo_mechanical_accessor at the same level.

[AlejandroCornejo]

done

[AlejandroCornejo]

ping

Pong

ping

[loumalouomega]

ping

Pong

ping

@rubenzorrilla

[rubenzorrilla]

done

Thanks. Now it's much more tidy.

[rubenzorrilla]

All my concerns have been addressed. Good job.

[loumalouomega]

@AlejandroCornejo right now

[AlejandroCornejo]

@AlejandroCornejo right now

yeah!!!!!!!!!!!!!!!!!!!!! finally :)