Architecture overview

TC-Python contains classes of these types:

  • TCPython – this is where you start with general settings.

  • SystemBuilder and System – where you choose database and elements etc.

  • Calculation – where you choose and configure the calculation.

  • Result – where you get the results from a calculation you have run.

TCPython

This is the starting point for all TC-Python usage.

You can think of this as the start of a “wizard”.

You use it to select databases and elements. That will take you to the next step in the wizard, where you configure the system.

Example:

from tc_python import *

with TCPython() as start:
    start.select_database_and_elements(...
    # e.t.c
# after with clause

# or like this
with TCPython():
    SetUp().select_database_and_elements(...
    # e.t.c
# after with clause

Tip

If you use TC-Python from Jupyter Lab / Notebook, you should use TC-Python slightly different to be able to use multiple cells. See Using TC-Python within a Jupyter Notebook or the Python console for details.

Note

When your python script runs a row like this:

with TCPython() as start:

a process running a calculation server starts. Your code, via TC-Python, uses socket communication to send and receive messages to and from that server.

When your Python script has run as far as this row:

# after with clause

the calculation server automatically shuts down, and all temporary files are deleted. It is important to ensure that this happens by structuring your Python code using a with() clause as in the above example.

Note

To re-use results from previous calculations, set a folder where TC-Python saves results, and looks for previous results.

This is done with the function set_cache_folder().

from tc_python import *

with TCPython() as start:
   start.set_cache_folder("cache")

This folder can be a network folder and shared by many users. If a previous TC-Python calculation has run with the same cache_folder and EXACTLY the same system and calculation settings, the calculation is not re-run. Instead the result is automatically loaded from disk.

It is also possible to explicitly save and load results.

from tc_python import *

with TCPython() as start:
   #... diffusion calculation (could be any calculation type)
   calculation_result.save_to_disk('path to folder')
   #...
   loaded_result = start.load_result_from_disk().diffusion('path to folder')

SystemBuilder and System

A SystemBuilder is returned when you have selected your database and elements in TCPython.

The SystemBuilder lets you further specify your system, for example the phases that should be part of your system.

Example:

from tc_python import *

with TCPython() as start:
    start.select_database_and_elements("ALDEMO", ["Al", "Sc"])
         # e.t.c

When all configuration is done, you call get_system() which returns an instance of a System class. The System class is fixed and cannot be changed. If you later want to change the database, elements or something else, change the SystemBuilder and call get_system() again, or create a new SystemBuilder and call get_system().

From the System you can create one or more calculations, which is the next step in the “wizard”.

Note

You can use the same System object to create several calculations.

Calculation

The best way to see how a calculation can be used is in the TC-Python examples included with the Thermo-Calc installation.

Some calculations have many settings. Default values are used where it is applicable, and are overridden if you specify something different.

When you have configured your calculation you call calculate() to start the actual calculation. That returns a Result, which is the next step.

Single equilibrium calculations

In single equilibrium calculations you need to specify the correct number of conditions, depending on how many elements your System contains.

You do that by calling set_condition().

An important difference from other calculations is that single equilibrium calculations have two functions to get result values.

The calculate() method, which gives a SingleEquilibriumTempResult, is used to get actual values. This result is “temporary”, meaning that if you run other calculations or rerun the current one, the resulting object no longer gives values corresponding to the first calculation.

This is different from how other calculations work. If you want a Result that you can use after running other calculations, you need to call calculate_with_state(), which returns a SingleEquilibriumResult.

Note

calculate() is the recommended function and works in almost all situations. Also it has much better performance than calculate_with_state().

Example:

from tc_python import *

with TCPython() as start:
    gibbs_energy = (
        start.
            select_database_and_elements("FEDEMO", ["Fe", "Cr", "C"]).
            get_system().
            with_single_equilibrium_calculation().
                set_condition(ThermodynamicQuantity.temperature(), 2000.0).
                set_condition(ThermodynamicQuantity.mole_fraction_of_a_component("Cr"), 0.1).
                set_condition(ThermodynamicQuantity.mole_fraction_of_a_component("C"), 0.01).
                calculate().
                get_value_of("G")
        )

Batch equilibrium calculations

Batch equilibrium calculations are used when you want to do many single equilibrium calculations and it is known from the beginning which result values are required from the equilibrium. This is a vectorized type of calculation that can reduce the overhead from Python and TC-Python similar to the approach used in numpy-functions for example.

Tip

The performance of batch equilibrium calculations can be significantly better than looping and using single equilibrium calculations if the actual Thermo-Calc calculation is fast. There is little advantage if the Thermo-Calc equilibrium calculations take a long time (typically for large systems and databases).

Example:

from tc_python import *

with TCPython() as start:
    calculation = (
        start
            .set_cache_folder(os.path.basename(__file__) + "_cache")
            .select_database_and_elements("NIDEMO", ["Ni", "Al", "Cr"])
            .get_system()
            .with_batch_equilibrium_calculation()
            .set_condition("T", 800.1)
            .set_condition("X(Al)", 1E-2)
            .set_condition("X(Cr)", 1E-2)
            .disable_global_minimization()
    )

    list_of_x_Al = [1, 2, 3, 4, 5, 6, 7, 8, 9, 10]
    list_of_x_Cr = [3, 5, 7, 9, 11, 13, 15]
    lists_of_conditions = []
    for x_Al in list_of_x_Al:
        for x_Cr in list_of_x_Cr:
            lists_of_conditions.append([
                ("X(Al)", x_Al / 100),
                ("X(Cr)", x_Cr / 100)])
    calculation.set_conditions_for_equilibria(lists_of_conditions)

    results = calculation.calculate(["BM", "VM"])

    masses = results.get_values_of("BM")
    volumes = results.get_values_of('VM')

print(masses)
print(volumes)

Precipitation calculations

All that can be configured in the Precipitation Calculator in Graphical Mode can also be done here in this calculation. However, you must at least enter a matrix phase, a precipitate phase, temperature, simulation time and compositions.

Example:

from tc_python import *

with TCPython() as start:
    precipitation_curve = (
        start.
            select_thermodynamic_and_kinetic_databases_with_elements("ALDEMO", "MALDEMO", ["Al", "Sc"]).
            get_system().
            with_isothermal_precipitation_calculation().
                set_composition("Sc", 0.18).
                set_temperature(623.15).
                set_simulation_time(1e5).
                with_matrix_phase(MatrixPhase("FCC_A1").
                                  add_precipitate_phase(PrecipitatePhase("AL3SC"))).
                calculate()
    )

Scheil calculations

All Scheil calculations available in Graphical Mode or Console Mode can also be done here in this calculation. The minimum you need to specify are the elements and compositions. Everything else is set to a default value.

Example:

from tc_python import *

with TCPython() as start:
    temperature_vs_mole_fraction_of_solid = (
        start.
            select_database_and_elements("FEDEMO", ["Fe", "C"]).
            get_system().
            with_scheil_calculation().
                set_composition("C", 0.3).
                calculate().
                get_values_of(ScheilQuantity.temperature(),
                              ScheilQuantity.mole_fraction_of_all_solid_phases())
        )

Property diagram calculations

For the property diagram (step) calculation, everything that you can configure in the Equilibrium Calculator when choosing One axis in Graphical Mode can also be configured in this calculation. In Console Mode the property diagram is created using the Step command. The minimum you need to specify are elements, conditions and the calculation axis. Everything else is set to default values, if you do not specify otherwise.

Example:

from tc_python import *

with TCPython() as start:
    property_diagram = (
        start.
            select_database_and_elements("FEDEMO", ["Fe", "C"]).
            get_system().
            with_property_diagram_calculation().
                with_axis(CalculationAxis(ThermodynamicQuantity.temperature()).
                    set_min(500).
                    set_max(3000)).
                set_condition(ThermodynamicQuantity.mole_fraction_of_a_component("C"), 0.01).
                calculate().
                get_values_grouped_by_stable_phases_of(ThermodynamicQuantity.temperature(),
                                                       ThermodynamicQuantity.volume_fraction_of_a_phase("ALL"))
        )

Phase diagram calculations

For the phase diagram (map) calculation, everything that you can configure in the Equilibrium Calculator when choosing Phase diagram in Graphical Mode can also be configured in this calculation. In Console Mode the phase diagram is created using the Map command. The minimum you need to specify are elements, conditions and two calculation axes. Everything else is set to default values, if you do not specify otherwise.

Example:

from tc_python import *

with TCPython() as start:
    phase_diagram = (
        start.
            select_database_and_elements("FEDEMO", ["Fe", "C"]).
            get_system().
            with_phase_diagram_calculation().
                with_first_axis(CalculationAxis(ThermodynamicQuantity.temperature()).
                    set_min(500).
                    set_max(3000)).
                with_second_axis(CalculationAxis(ThermodynamicQuantity.mole_fraction_of_a_component("C")).
                    set_min(0).
                    set_max(1)).
                set_condition(ThermodynamicQuantity.mole_fraction_of_a_component("C"), 0.01).
                calculate().
                get_values_grouped_by_stable_phases_of(ThermodynamicQuantity.mass_fraction_of_a_component("C"),
                                                       ThermodynamicQuantity.temperature())
        )

Diffusion calculations

For diffusion calculations, everything that you can configure in the Diffusion Calculator can also be configured in this calculation. The minimum you need to specify are elements, temperature, simulation time, a region with a grid and width, a phase and an initial composition.

Example:

from tc_python import *

with TCPython() as start:
    diffusion_result = (
        start.
            select_thermodynamic_and_kinetic_databases_with_elements("FEDEMO", "MFEDEMO", ["Fe", "Ni"]).
            get_system().
            with_isothermal_diffusion_calculation().
                set_temperature(1400.0).
                set_simulation_time(108000.0).
                add_region(Region("Austenite").
                    set_width(1E-4).
                    with_grid(CalculatedGrid.linear().set_no_of_points(50)).
                    with_composition_profile(CompositionProfile().
                        add("Ni", ElementProfile.linear(10.0, 50.0))
                    ).
                add_phase("FCC_A1")).
        calculate())

    distance, ni_fraction = diffusion_result.get_mass_fraction_of_component_at_time("Ni", 108000.0)

Property Model calculations

For Property Model calculations, everything that you can configure in the Property Model Calculator in Graphical Mode can also be configured in this calculation. The minimum you need to specify are elements, composition and which Property Model you want to use.

Example:

from tc_python import *

with TCPython() as start:
    print("Available Property Models: {}".format(start.get_property_models()))
    property_model = (
        start.
            select_database_and_elements("FEDEMO", ["Fe", "C"]).
            get_system().
            with_property_model_calculation("Driving force").
            set_composition("C", 1.0).
            set_argument("precipitate", "GRAPHITE"))

    print("Available arguments: {}".format(property_model.get_arguments()))
    result = property_model.calculate()

    print("Available result quantities: {}".format(result.get_result_quantities()))
    driving_force = result.get_value_of("normalizedDrivingForce")

Material to Material calculations

Material to Material calculations are generally regular single equilibrium, property diagram or phase diagram calculations but they are specialised to handle the mixture of two materials A and B. Everything that you can configure in the Material to Material Calculator in Graphical Mode can also be configured in this calculation. The minimum required configuration is shown below for a Property diagram calculation for varying amount of material B. The other calculators (single fraction of material B and phase diagram calculations) are configured in a similar way.

Example:

from tc_python import *

with TCPython() as start:
    material_to_material_property_diagram = (
        start.
            select_database_and_elements("FEDEMO", ["Fe", "Cr", "Ni", "C"]).
            get_system().
            with_material_to_material().
            with_property_diagram_calculation().
            set_material_a({"Cr": 10.0, "Ni": 15.0}, "Fe").
            set_material_b({"Cr": 15.0, "Ni": 10.0}, "Fe").
            set_activities({"C": 0.1}).
            with_constant_condition(ConstantCondition.temperature(800 + 273.15)).
            with_axis(MaterialToMaterialCalculationAxis.fraction_of_material_b(from_fraction=0.0,
                                                                               to_fraction=1.0,
                                                                               start_fraction=0.5))
    )

    result = material_to_material_property_diagram.calculate()
    data = result.get_values_grouped_by_quantity_of(MATERIAL_B_FRACTION,
                                                    ThermodynamicQuantity.volume_fraction_of_a_phase(ALL_PHASES))

    for group in data.values():
        fractions_of_b = group.x
        volume_fractions_of_phase = group.y
        phase_name = group.label

Process Metallurgy calculations

Process Metallurgy calculations are specialized to support the convenient handling of component-based additions (i.e., slag compositions such as 50% Al2O3 - 30% CaO - 20% SiO2), provide tailor-made result quantities, a framework for developing kinetic process simulations, and more useful features.

There are two distinct types of calculations:

Equilibrium calculation example:

Equilibrium calculations are useful in a large range of situations when considering the kinetics of a process is unnecessary.

from tc_python import *

with TCPython() as session:
    metal = EquilibriumAddition({"Fe": None, "C": 4.5, "Si": 1.0}, 100e3, temperature=1650 + 273.15)
    slag = EquilibriumAddition({"CaO": 75, "Al2O3": 25}, 3e3, temperature=1600 + 273.15)
    gas = EquilibriumGasAddition({"O2": 100}, 1000, amount_unit=GasAmountUnit.NORM_CUBIC_METER)
    calc = session.with_metallurgy().with_adiabatic_equilibrium_calculation(ProcessDatabase.OXDEMO)

    (calc
     .add_addition(metal)
     .add_addition(slag)
     .add_addition(gas))

    result = calc.calculate()

    print(f"Stable phases: {result.get_stable_phases()}, temperature: {result.get_temperature()} K")

Process simulation example:

TC-Python is providing a framework for modelling in principle any process in metallurgy, especially steel-making. It is up to the user to actually develop a concrete model for the process in question. The framework is in the current release limited to one reaction zone connecting two bulk zones. These bulk zones are typically the steel melt and the top slag, but not limited to that. The framework in its current version has proven to be useful to model industrial ladle furnaces, AOD- and VOD-converters and more. Process features such as heating and cooling, heat transfer between the bulk zones, inclusion formation and their flotation, etc., can be modelled.

This is a very simplified minimal but complete model mimicking a BOF process:

from tc_python import *

with TCPython() as session:
    calc = (session.with_metallurgy()
            .with_adiabatic_process_calculation(ProcessDatabase.OXDEMO)
            .set_end_time(15 * 60))

    steel_zone = MetalBulkZone(density=7800)
    slag_zone = SlagBulkZone(density=4500)

    steel_zone.add_addition(SingleTimeAddition({"Fe": None, "C": 4.5, "Si": 1.0}, 120e3,
                                               temperature=1600 + 273.15), time=0)
    slag_zone.add_addition(SingleTimeAddition({"CaO": 75, "SiO2": 25}, 1.2e3,
                                              temperature=1500 + 273.15,
                                              composition_unit=CompositionUnit.MOLE_PERCENT), time=0)

    steel_zone.add_continuous_addition(ContinuousGasAddition({"O2": 100}, 1,
                                                             rate_unit=GasRateUnit.NORM_CUBIC_METER_PER_SEC))

    calc.with_reaction_zone(ReactionZone(area=10.0,
                                         left_zone=steel_zone, mass_transfer_coefficient_left=1.0e-5,
                                         right_zone=slag_zone, mass_transfer_coefficient_right=1.0e-6))

    result = calc.calculate()

    print(f"Stable phases in the steel melt: {result.get_stable_phases(steel_zone)}")
    print(f"C-content in steel vs. time: {result.get_composition_of_phase_group(steel_zone,
                                                                                PhaseGroup.ALL_METAL)['C']}")

Additive Manufacturing calculations

For Additive Manufacturing calculations, everything that you can configure in the AM Calculator in Graphical Mode can also be configured in this calculation. In the minimal case, you do not need to specify any options (the material used by default is IN625).

from tc_python import *

with TCPython() as start:
    am_result = (start.with_additive_manufacturing()
                 .with_steady_state_calculation()
                 .calculate())

    print(f"Melt pool depth: {am_result.get_meltpool_depth()}")
    plotter, mesh = am_result.get_pyvista_plotter()
    plotter.add_mesh(mesh)
    plotter.show()

With several custom settings a basic calculation looks like this:

from tc_python import *

with TCPython() as start:
    am_result = (start.with_additive_manufacturing()
                 .with_steady_state_calculation()
                 .disable_fluid_flow_marangoni()
                 .disable_separate_materials()
                 .with_material_properties(MaterialProperties.from_library("IN718"))
                 .set_powder_density(90)
                 .set_ambient_temperature(15 + 273.15)
                 .set_base_plate_temperature(25 + 273.15)
                 .with_top_boundary_conditions(TopBoundaryConditions().set_radiation_emissivity(0.2))
                 .with_heat_source(HeatSource.conical()
                                   .set_power(210)
                                   .set_absorptivity(80.0))
                 .calculate())

    print(f"Melt pool depth: {am_result.get_meltpool_depth()}")
    plotter, mesh = am_result.get_pyvista_plotter()
    plotter.add_mesh(mesh)
    plotter.show()

Tip

The graphical visualization and extraction of data from the 3D result dataset requires the package pyvista. More details are provided in the section Optional packages.

Result

All calculations have a method called calculate() that starts the calculations and when finished, returns a Result.

The Result classes have very different methods, depending on the type of calculation.

The Result is used to get numerical values from a calculation that has run.

The Result can be saved to disk by the method save_to_disk().

Previously saved results can be loaded by the method load_result_from_disk() on the SetUp class.

Example:

# code above sets up the calculation
r = calculation.calculate()
time, meanRadius = r.get_mean_radius_of("AL3SC")

The Result objects are completely independent from calculations done before or after they are created. The objects return valid values corresponding to the calculation they were created from, for their lifetime. The only exception is if you call calculate() and not calculate_with_state() on a single equilibrium calculation.

As in the following example you can mix different calculations and results, and use old results after another calculation has run.

Example:

# ...
# some code to set up a single equilibrium calculation
# ...

single_eq_result = single_eq_calculation.calculate_with_state()

# ...
# some code to set up a precipitation calculation
# ...

prec_result = precipitation_calculation.calculate()

# ...
# some code to set up a Scheil calculation
# ...

scheil_result = scheil_calculations.calculate()

# now it is possible to get results from the single equilibrium calculation,
# without having to re-run it (because it has been calculated with saving of the state)

gibbs = single_eq_result.get_value_of("G")

DiffusionResult

The DiffusionResult class, that is returned when calling calculate() on any DiffusionCalculation, has the possibility to create a ContinuedDiffusionCalculation, in addition to the “normal” functionality for results. This makes it possible to run a diffusion calculation and then, depending on the result, change some settings and continue.

Example:

# ...
# some code to set up a Diffusion calculation
# ...
first_diffusion_result = diffusion_calculation.calculate()

continued_calculation = first_diffusion_result.with_continued_calculation()

continued_calculation.set_simulation_time(110000.0)
continued_calculation.with_left_boundary_condition(BoundaryCondition.mixed_zero_flux_and_activity().set_activity_for_element('C', 1.0))
second_result = continued_calculation.calculate()
# ...
# Now you can use get second_result to get calculated values, just as normal.
# You can also use first_diffusion_result even after second_result is created.
# You can also use second_result (and even first_diffusion_result) to create a new ContinuedDiffusionCalculation by calling with_continued_calculation.

Property Model Framework

The Python Property Model SDK extends the Thermo-Calc software to enable you to create your own Property Models. A Property Model is a Python-based calculation that can use any TC-Python functionality (including diffusion and precipitation calculations) but is usable through the Graphical User Interface (UI) of Thermo-Calc in a more simple way. It is typically used to model material properties but by no means limited to that. Examples of Property Models provided by Thermo-Calc include Martensite and Pearlite formation in steel.

The Property Model Framework uses standard Python 3 beginning with Thermo-Calc 2021a and can access all TC-Python functionality and any Python package including numpy, scipy, tensorflow, etc. The actual calculation code is nearly identical, regardless if called from within a Property Model or from standard Python.

This is a complete rewrite of the original version of the framework that was based on Jython 2.7 and therefore had a number of limitations. Property models written with the old Property Model Framework before Thermo-Calc 2021a are not compatible with the new framework. However, the migration should be relatively easy because the syntax was changed as little as possible.

Property models vs. TC-Python

The main difference between a Property Model and regular TC-Python code is that a Property Model is directly integrated into the UI of Thermo-Calc via a plugin architecture while TC-Python code can only be accessed by programs and scripts written in Python.

The user should develop a Property Model if the functionality needs to be available from the Thermo-Calc UI, especially if it should be applied by other users not familiar to programming languages. Otherwise it is preferable to implement the functionality directly in a TC-Python program. If required, Property Models can as well be accessed from within TC-Python.

Architecture

Every Property Model needs to contain a class that implements the interface tc_python.propertymodel_sdk.PropertyModel. There are naming conventions that must to be fulfilled: the file name is required to follow the the pattern XYPythonModel.py and the name of the class needs to match this. Additionally the file must be placed in a directory named XYPython within the Property Model directory. The content of the placeholder XY can be freely chosen.

A simple complete Property Model, saved in a file called SimplePythonModel.py in the directory SimplePython, looks like this:

from tc_python import *


class SimplePythonModel(PropertyModel):
    def provide_model_category(self) -> List[str]:
        return ["Demo"]

    def provide_model_name(self) -> str:
        return "My Demo Model"

    def provide_model_description(self) -> str:
        return "This is a demo model."

    def provide_ui_panel_components(self) -> List[UIComponent]:
        return [UIBooleanComponent("CHECKBOX", "Should this be checked?", "Simple checkbox", setting=False)]

    def provide_calculation_result_quantities(self) -> List[ResultQuantity]:
        return [create_general_quantity("RESULT", "A result")]

    def evaluate_model(self, context: CalculationContext):
        if context.get_ui_boolean_value("CHECKBOX"):
            self.logger.info("The checkbox is checked")

        # obtain the entered values from the GUI
        composition_as_mass_fraction = context.get_mass_fractions()
        temp_in_k = context.get_temperature()
        calc = context.system.with_single_equilibrium_calculation()
        # continue with a TC-Python calculation now ...

        context.set_result_quantity_value("RESULT", 5.0)  # the value would normally have been calculated

The basic building blocks of the Property Model API are:

  • tc_python.propertymodel_sdk.ResultQuantity: Defines a calculation result of a Property Model that will be provided to the UI after each model evaluation

  • tc_python.propertymodel_sdk.CalculationContext: Provides access to the data from the UI (such as the entered composition and temperature) and to the current TC-Python system object which is the entrypoint for using TC-Python from within the Property Model

  • tc_python.propertymodel_sdk.UIComponent: These are the UI-components that create the user interface of the Property Model within the model panel of the Thermo-Calc application UI. Different components are available (for example checkboxes, text fields and lists).

Property Model directory

The Property Model py-files need to be located within subdirectories of the Property Model directory, e.g. PropertyModels/XYPython/XYPythonModel.py. The default Property Model directory can be changed in the menu Tools -> Options in the graphical user interface.

Operating system

Default Property Model directory

Windows

C:\Program Files\Thermo-Calc\2024b\PropertyModels

Linux

/home/UserName/Thermo-Calc/2024b/PropertyModels or: /opt/Thermo-Calc/2024b/PropertyModels

MacOS

/Applications/Thermo-Calc-2024b.app/Contents/Resources/PropertyModels

Note

The Property Model directory location has been changed in the release 2023b.